Skip to main content
SearchLoginLogin or Signup

Extending Intention to Use Toward Postadoption Behavior—Conceptualizing Actual Usage for Information Technology-Enabled Banking Services

Volume 5, Issue 2. DOI: 10.1037/tmb0000132

Published onJun 10, 2024
Extending Intention to Use Toward Postadoption Behavior—Conceptualizing Actual Usage for Information Technology-Enabled Banking Services
·

Abstract

This study aims to address the gap in technology adoption research by focusing on actual postadoption behavior in consumer marketing and operationalizing the construct of actual usage for IT-enabled banking services. The research uses a rigorous two-stage scale construction and validation process using partial least square-structural equation modelling. Results of the present research article conceptualize actual usage measurement model as a second-order formative-reflective construct with four first-order reflective constructs. This study also provides a foundation for future research studies incorporating Technology Readiness Index 2.0 and self-reported actual usage as a multidimensional construct. This operationalization confirms the uniqueness of different dimensions of actual usage, which should be configured as strategy devising tools for sensing and seizing opportunities to tap the market for any technology product or service.

Keywords: postadoption behavior, multidimensional construct, formative construct, partial least square-structural equation modelling

Funding: No funding was received for this research.

Disclosures: The authors declare no conflicts of interest.

Data Availability: The authors will make the data, analytic methods, and study materials available to other researchers. There is no prior use of the data used in this study.

Open Access License: This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License (CC-BY- NC-ND). This license permits copying and redistributing the work in any medium or format for noncommercial use provided the original authors and source are credited and a link to the license is included in attribution. No derivative works are permitted under this license.

Correspondence concerning this article should be addressed to Kritika Nagdev, School of Business Studies, Vivekananda Institute of Professional Studies, Guru Gobind Singh Indraprastha University, AU Block, Pitampura, New Delhi 110034, India. Email: [email protected]


Video Abstract


Digital transformation of banking and financial services brings multifaceted benefits to customers. The pandemic upended the consumer market across industries globally and accelerated the shift toward digital banking (Gopakumar, 2020). However, regardless of this growing customer demand and potential assistance of the services, consumers still choose to avoid, refuse, or postpone their adoption of technology (Blut & Wang, 2020; Deloitte Switzerland, 2021; Laukkanen et al., 2008; Seth et al., 2020; Talwar et al., 2020). For a long time in the past, marketers have only concentrated on adoption strategies of their technology, but studying postadoptive behaviors has become a necessity in the current times (Oertzen & Odekerken-Schröder, 2019; Shih & Venkatesh, 2004). The strongest indicator of continued consumer acceptance is postadoption actual usage behavior and the degree of usage (Alzoubi et al., 2022; Liu & Forsythe, 2010). Instead of initial adoption, ongoing use of technology is critical for technical enterprises’ significant success and long-term survival (Son & Han, 2011). Therefore, focusing solely on improving customers’ favorable perceptions or positive attitudes toward using technology might not be an efficient approach for the prediction and comprehension of individual consumer behavior. It is, therefore, important to go beyond the adoption diffusion paradigm and delve into the actual usage (AU) as the ultimate goal of any marketer is to get consumers to use the technology postadoption across industries (Dimitrova et al., 2022; Oertzen & Odekerken-Schröder, 2019). To address this, the present study measures actual usage behavior for IT-enabled banking services (ITeBS).

In the extant literature, the actual usage construct has not been measured efficiently as they are neither appropriately operationalized nor observed in terms of depth and duration of usage. Since modern technologies can perform diverse functions, Huh and Kim (2008) categorized user behavior in terms of basic and innovative functions of utilization. Shih and Venkatesh (2004) segmented the respondents into four categories of users based on two dimensions, the rate and variety of usage. It defines variety as the diverse ways a product may be utilized, and usage rate as the frequency of use of a product, regardless of the various applications for which it may be used. In a latest research, technology acceptance model (TAM) by Davis (1989) has been used to measure actual use as a dependent variable, in terms of frequency and duration of use (Mariano et al., 2022). However, there still exists a huge research gap in examining the depth of use of technology that is, features used, and so on (Burton-Jones & Straub, 2006).

It is essential to note that any technology comes with its inherent adoption and usage challenges. User characteristics, easy access, and earlier experience in technology also have an impact on consumer behavior (Kim et al., 2020; Trinh et al., 2020). Consumer’s inability to use may lead to negative perceptions or feelings about the usage of technology. Despite several debates, discussions, and recommendations to incorporate usage, majority of technology adoption research ends up evaluating only the internal perceptual state of consumers like perception toward technology, their likelihood and intention to use technology. A bibliometric analysis of literature published in a leading journal for bank marketing also states the focus of the research in online banking is “adoption,” with a particular emphasis on the “predictor” of “behavioral intention” and “willingness” to adopt and use “online banking service” (Kumar et al., 2022). The ever-evolving technology spheres and increased cybercrime rate, demand for measuring the technology readiness of the consumer for actual usage. Hence, it is essential to conduct relevant research to corroborate the theory of the Technology Readiness Index (TRI) by Parasuraman and Colby (2015) which evaluates the overall perception by evaluating contributing and inhibiting factors in the current context.

From the managerial perspective, banks would also be more interested in knowing the customers’ actual usage of ITeBS. It would be worthwhile to examine the extent of technology usage, rather than only measuring its perception or likeliness to use technology. Hence, it is an inescapable research gap. Therefore, the study accepts the proposition of Blut and Wang (2020), which recommended predicting actual behavior rather than an internal perceptual state of consumers. It establishes a process-driven approach to measure consumer acceptance, where technology readiness (TR) determines behavioral intention and resultantly intention affects actual usage of ITeBS. Thus, this research work remains one of the earliest for investigating both the behavioral aspects—relative (intention to use) and actual behavior along with, the recommendations of Polites et al. (2012) to conceptualize a self-reported form of actual usage as a multidimensional construct (by way of measuring adoption behavior through an established formative scale).

Literature Review

Technology has become a ubiquitous part of almost all processes. Usage of a technological product or service as defined in literature refers to “an individual’s actual usage of the product or service for personal purposes” (Son & Han, 2011). Technology has also modulated behaviors and expectations of any service or process consumed (Ivanov et al., 2017; Sebastian et al., 2023). This may also be impacted by TR—which may be defined as the user’s propensity to actually embrace and eventually use various technologies (Hong & Park, 2024; Parasuraman, 2000). Thus, the present study adapts the TRI which is devised to indicate the consumers’ perception of technology.

Theoretical Background

The true objective of consumer research is to represent the consumer’s voice by investigating attitudes, needs, motivations, and behavior related to a product or service. In technology-based service frameworks, services are cocreated and cannot be delivered without customers’ active involvement (Lovelock & Wirtz, 2004). This high level of contribution of consumer behavior cannot be accurately explained by the theory of TAM and its extended versions, namely the Unified Theory of Acceptance and Use of Technology by Venkatesh et al. (2012), which are invariably adopted in consumer marketing settings (Negm, 2023; Omar et al., 2023).

In addition, the theory proposed by Parasuraman, the TRI in the year 2000 is an “individual-specific” attitudinal scale (C. H. Lin et al., 2007). TR construct is a structural arrangement of perceptual enablers and inhibitors that normally result in an overall mindset that determines a person’s inclination toward technologies (Parasuraman, 2000). Simply put, enablers are positive characteristics of technology as perceived by users. In contrast, inhibitors are the negative perceptions associated with technology. For example, security concerns while using alternate banking channels, lack of trust to switch to modern technology, avoidance of dependence on using technology for every task, and so on. The index is a multiple-dimension construct, comprising four distinct factor dimensions, namely, innovativeness, optimism, insecurity, and discomfort.

Technology readiness dimensions have also been tested as moderating variables to influence customer experience and continued intentions to adopt smart banking services (Khashan et al., 2023). Hence, readiness to use could come out as the major differentiating variable that affects individuals’ acceptance of IT-enabled banking services.

To understand the dynamics behind the acceptance of ITeBS, the four TR dimensions of TRI 2.0 (Parasuraman & Colby, 2015) are used in the study to measure the readiness of customers. (TRI 1.0) by Parasuraman (2000), the first version of the scale was a lengthy scale comprising 36 items. The updated version TRI 2.0 was developed in the year 2015 (Parasuraman & Colby, 2015), which is widely applied as it is less troublesome with only 16 dimensions from the original 36-item scale; it is also less burdensome in terms of assessment of multiple constructs compared to TRI 1.0 (Blut & Wang, 2020; O’Hern & Louis, 2023). Therefore, TRI 2.0 being a more robust forecasting tool of technology-related behavioral intentions and actual behaviors is deployed for the study.

Actual Usage Conceptualization

The real test of proliferation and success of technology is not only the first usage but its continued consumption. Several researchers have pointed out that actual use is the real indicator of the effectiveness of any technology (Marion & Fixson, 2021). Postadoptive behaviors and continued usage may be considered true tests of the success of any technology.

Straub et al. (1995) suggested that “Usage should be reformulated as two separate constructs: perceived usage and actual usage and that the TAM may require substantial reformulation.” The actual usage literature is majorly seen to evaluate two forms—perceived usage/subjective (self-reported) usage and objective (actual usage or frequency recorded by the digital system) measures. Scholars have peculiarly witnessed that studies that are dependent on the mere subjective measures of usage which may portray substantial effects, do not capture the “depth of use” which embodies more details about the features deployed (Burton-Jones & Straub, 2006; Oertzen & Odekerken-Schröder, 2019; Rabaa’i & AlMaati, 2021; Shaw, 2011).

Thus, follow-on postadoption studies that employ depth-of-use measures of the technology would be an important step in further defining the research gap:

Research Gap

While substantial literature is available on the factors influencing of adoption of technology (Naruetharadhol et al., 2021; Parasuraman, 2000), there is a dearth of studies on factors for sustained use (Huh & Kim, 2008; Shih & Venkatesh, 2004). Since the survival of any technology depends on recurring, sustained, and widespread use, it is important to study the factors for the same. Table 1 portrays the studies that have incorporated frequency of actual usage or depth of use (specific features of technology) and not measured it on an agreement/disagreement scale. Hence, delineating a gap between defining the actual usage construct and operationalizing it in its true facet as a formative construct. This study follows the recommendation of Shaw (2011) of conceptualizing self-reported usage as the level of use. Further, this endogenous formative construct is evaluated by way of partial least square-structural equation modelling (PLS-SEM) using SmartPLS V3.3.3 software, which has the capability of evaluating reflective and formative construct in a single model (Hair et al., 2011).

Table 1

Insights from “Actual Usage/Depth of Use” Literature

Research study

Outcome of research

Industry

Base concept/theory

Definition, operationalization of usage construct

Shih and Venkatesh (2004)

Categorize respondents into four types of users using two dimensions, the rate of use and the variety of usage

Home technology

Use-diffusion (UD) theory and model

Self-reported rate and variety of use
Regression analysis, multinomial logit analysis

Burton-Jones and Gallivan (2007)

Defines usage as a multilevel construct by defining guidelines for function of usage, structure of usage, and context of usage

System usage in organizations

Multilevel research

Qualitative research on multilevel theory development with a functional analysis

Huh and Kim (2008)

Classify users’ product usage behavior into basic and innovative function usage

Cell phone purchases, particularly the use of innovative features

Use-diffusion model, Shih and Venkatesh (2004)

Hypothesis testing for the reflective model framed—with intention as endogenous variable

Son and Han (2011)

Identifies three types of usage pattern as the usage rate of basic functions, the usage rate of innovative functions, and the usage variety of innovative functions

Internet Protocol TV users

Technology readiness (Parasuraman, 2000)

Hypothesis testing for the reflective model framed

Shaw (2011)

proposed that the construct of level of use be measured as a formative construct.

Systems usage

Technology acceptance model (TAM) by Davis (1989) and theory of reasoned action (TRA) by Fishbein and Ajzen (1975)

Principal component analysis for level of use

Oertzen and Odekerken-Schröder (2019)

Model provides a multifaceted view of the postadoption phase, including actual usage

Online banking

TAM by Davis (1989)

Conceptualized actual usage as reflective though measurement items are formative

Hypothesis Development

Technology brings multifaceted benefits to customers. Regardless of the potential assistance offered by these products and services, consumers still choose to avoid, refuse, or postpone their adoption (Deloitte Switzerland, 2021; Laukkanen et al., 2008; Mick & Fournier, 1998; Seth et al., 2020; Talwar et al., 2020).

The new refined 16-item TRI scale (Parasuraman & Colby, 2015) used in the study, demonstrated sound psychometric properties found from numerous reliability and validity tests and from studies incorporating scale replications using varied sample sets. The efficacy of the newly developed scale has increased due to improved simplicity and effortlessness in application across demographic classes, industry sectors, and cultures providing greater strength of TRI 2.0 (Shahid Iqbal et al., 2018).

Technology Readiness and Its Dimensions

Existing theory and literature state technology readiness as a multidimensional construct comprising four dimensions:

Innovativeness

Hirschman (1980) has talked about inherent novelty seeking which is the desire of users for novelty, acting as a catalyst for usage. Studies have found that optimism and innovativeness were found to be significant in impacting continued intentions and usage in the context of self-service banking and smart banking (Chen & Chen, 2009; Khashan et al., 2023). Whereas, innovativeness is insignificant in its impact on adoption intentions toward internet of things (IoT) products and services for online learning (Negm, 2023).

Optimism

Optimism, one of the key constructs of the TR index is about having a positive outlook and believing that technology will better one’s life and make it easier (Parasuraman, 2000). Optimism also reduces any negative thoughts about usage and thus the user embraces technology with an open and flexible mindset (Walczuch et al., 2007). From another perspective, a Finnish study explores how boosting user optimism and reducing insecurity, can influence the intention to use conditionally automated vehicles (O’Hern & Louis, 2023). Whereas, Tavera-Mesías et al. (2023) found optimism to be more relevant, indicating that the users are less insecure about technology.

Insecurity

Security concerns are one of the major impacting factors in user’s resistance to use. Privacy is an important part of overall security. User characteristics and earlier experiences in technology concerning security and privacy violations also have an impact (An et al., 2021; Huang et al., 2023). Insecurity may form a strong inhibitor to the intention to use, especially so in the case of technology (Parasuraman, 2000). There are many studies in the information systems area focusing on informational privacy concerns, which is a feeling or an attitude originating from privacy violations (Rana & Arora, 2022) and impacts users’ behavior negatively (Hsu & Lin, 2018; Merhi et al., 2019). These concerns critically affect the buying as well as acceptance or resistance to use of technology (Dupuis & Ebenezer, 2018).

Discomfort

The discomfiture of use of technology can be defined as the tangible and intangible costs associated with the inability to comprehend (Mukherjee & Hoyer, 2001; Priporas et al., 2024). Sometimes the feeling of being overwhelmed (Parasuraman, 2000) reduces the comfort level with technology in use. All this can lead to disappointment and frustration (Mick & Fournier, 1998) reducing usage and its frequency. Meuter et al. (2003) theorized that technology anxiety reduces usage and leads to avoidance of technology.

The present research intends to explore the determinants of technology readiness for IT-enabled banking services in the Indian setting. Thereby postulating the first hypothesis:

Hypothesis 1: Technology readiness is a multidimensional construct comprising of four dimensions—optimism, innovativeness, discomfort and insecurity.

TR and Behavioral Intention

Behavioral intention (BI) as a construct was formerly developed in the theory of planned behavior and theory of reasoned action, which are extensively used in successive frameworks associated with technology acceptance. Behavioral intention construct is projected as the most significant predictor in the theory of planned behavior.

Usage intention for any newly introduced product or service entails prospective customers possess mixed perceptions of technology to be able to comfortably use it. Meuter et al. (2003) theorized that technology anxiety is positively associated with word-of-mouth intentions (referral intention), customer trial of self-service technologies and as well as repeat usage intention (repurchase). Zeithaml et al. (2002) suggested that TR had a significant effect on e-shopping behavior. Yousafzai and Yani-de-Soriano (2012) posited that the beliefs–intention relationship shows different results with respect to difference in the levels of technology readiness, different age groups and gender roles toward internet banking adoption. Technology readiness level is even examined for higher education students to explain adoption intention toward educational IoT needed for online learning (Negm, 2023). Therefore, it is assumed that technology readiness would influence customers’ behavioral intentions, leading to the hypothesis:

Hypothesis 2: Technology readiness has a significant relationship with behavioral intentions toward ITeBS.

Behavioral Intention and Actual Usage

Analyzing future intentions has been common practice in TAM based research. The analysis of numerous TAM studies revealed that only 7.5% of them included actual usage, while the others relied on use intention or behavioral intentions (Caldeira et al., 2021; Yousafzai et al., 2007). This has also been reported as their limitation, that is to not being able to find the real usage and consider only an endogenous construct which is an approximation to actual use (Hernandez et al., 2009), ignoring the fact that the ultimate aim in consumer marketing is prediction and comprehension of individual consumer behavior (Dimitrova et al., 2022; Oertzen & Odekerken-Schröder, 2019), thus the present study measures actual behavior.

The study supports the importance of considering the intention to use as a dependent variable when measuring continued usage, as the behavioral outcome varies depending on the study’s purpose and measurement method. Whereas intention to adopt has to be incorporated when prospective usage or adoption is of concern (Karahanna et al., 1999; Xu et al., 2021).

Therefore, both the aspects, relative and actual behavior have been considered for the study thereby, measuring consumer acceptance through intention as well as actual usage of ITeBS. Such mutually connected behavior is also supported by prior studies, which have confirmed that intended behavior relates to the actual behavior of customers (Bölen & Özen, 2020; Venkatesh et al., 2012; Yiu et al., 2007).

This study therefore lays special emphasis on frequency of use by way of measuring actual usage construct along with behavioral intention. Thereby, the following hypothesis is proposed:

Hypothesis 3: There is a significant relationship between behavioral intention and actual usage.

Development of Conceptual Model

A conceptual model as represented in Figure 1 is established based on logically ordered review of literature and hypothesis development. The conceptual framework includes technology readiness as a latent construct and optimism, innovativeness, discomfort, and insecurity as its various dimensions. The framework considers technology readiness as an independent variable and two constructs of behavioral outcome (behavioral intention and actual usage) as a dependent variable. Since, the study aims to measure real facts of frequency of usage through a 20-item scale (refer Appendix), the actual usage construct is conceptualized as a formative endogenous construct.

Figure 1

Conceptual Model

Research Methodology

The study used an integrated research methodology for developing and validating the proposed research model. The research model has been developed based on the extant literature review and the insights from in-depth interviews with industry and academia experts in the area of the subject. The model was then tested empirically by a self-administered research instrument developed for the study using the survey method of data collection.

Measures

The items used in this survey were adaption from the prior studies published in peer-reviewed journals. These are technology readiness and its determinants (TRI 2.0, Parasuraman & Colby, 2015; 16-item scale); behavioral intention (Davis, 1989; three items). The items for actual usage construct (20 items) were the outcome of the related technology-based services literature (Lam et al., 2008) along with semistructured interviews and content analysis of the mobile banking applications and bank websites. Bank mobile applications and websites of winners and runners-up of an award category “Technology Bank of the Year” by the Indian Banks’ Association during 2016–2021 were considered for evaluation. Nine commercial banks namely, Bank of Baroda, South Indian Bank, Citibank, Industrial Credit and Investment Corporation of India Bank, YES Bank, IndusInd Bank, State Bank of India, Union Bank, and Vijaya Bank from all three segments of large, medium, and small banks were selected for website analysis. The services provided via ITeBS were looked for on these websites and mobile applications. Relevant content was extracted and used as keywords. These items were validated through the opinions of banking and industry experts as well as discussions with customers by visiting bank branches.

The study variables were measured using the 9-point Likert scale with 1 representing strongly disagree and 9 signifying strongly agree. Researchers suggest that a 7-point is better than a 5-point Likert scale and a 9-point is better than a 7-point to achieve measurement of a continuous variable (Norman, 2010). Actual usage construct could not be measured on an agreement/disagreement scale. Literature also suggests that measuring the frequency of ITeBS would be useful (Venkatesh et al., 2012; Yiu et al., 2007). Thus, it has been measured on a 9-point scale with anchors framed on a duration of frequency (Moon & Kim, 2001). Further model estimation was done through testing the measurement model and structural model using SmartPLS V3.3.3 software. This study employs SmartPLS software for measuring the formative construct, actual usage on the recommendation of Sarstedt et al. (2022) which states the five most frequently mentioned reasons to use SmartPLS software.

Analyzing Semistructured Interviews and Website, Mobile Application Analysis

Semistructured interviews were conducted with 31 regular customers of IT-enabled banking services, who majorly used ITeBS for their banking needs (specifically, approx. 75% of their banking transactions are performed through one or the other ITeBS component). The interviews were conducted to analyze various dimensions of actual usage. Customers were asked about various transactions made using IT-enabled banking including internet banking, mobile banking, automated teller machines, telephone banking, electronic fund transfers, electronic clearance services, and point-of-sales terminal (Kiosk banking).

Interview responses were then analyzed using content analysis to identify items for actual usage. A rational form of content analysis is utilized in the current context for analyzing interview responses. Where, after defining differentiated classes, the interviewees’ answers were coded into different themes which were then confirmed by validating previous studies’ findings and expert opinions. The items that did not induce a comprehensive understanding of the variables or that were repeated were eliminated.

During website and mobile application analysis, the relevant content was extracted and used as keywords. Subsequently, the responses of semistructured interviews and website analysis were considered together to frame the items.

Based on the analysis of semistructured interviews and the banks’ websites the items were finalized to be used for the study.

Data Collection

Since every segment of prospective users is important to evaluate, users and nonusers of ITeBS from Tier-1 cities of India were chosen as the sampling frame. It is important to evaluate every segment of prospective users to know their overall perception including their inhibitions and motivation for using/not using ITeBS (Nagdev et al., 2021).

The reason for considering Tier-1 cities was based on the reports that suggest that Tier-1 cities are foremost in case of digital acceptance as they enjoy better infrastructure, education, awareness, and digital penetration. Statistics also state that out of total Unified Payment Interface users across the nation, 99% of Tier-1 citizens of India embraced Unified Payment Interface in its first year of launch in January 2018 (Kumar, 2019). Therefore, it was essential to consider these cities to know their perception as well as their inhibitions and attitudes toward not using ITeBS.

The final questionnaire was distributed to bank customers aged 18 years or older, and the data were collected from October 2022 to January 2023 from the Tier-1 cities of India that is, Hyderabad, Delhi, Bengaluru, Mumbai, Chennai, and Kolkata using a snowball sampling method. Probability sampling techniques are ideal for research, but they only work when the total population size is known. In India, the exhaustive number of bank customers cannot be determined in a given sample frame, and accessing the complete database is not feasible. Hence, the nonprobability sampling method was considered for data collection (Patsiotis et al., 2012).

Two-stage approach of snowball sampling was followed, in the first stage, convenience sampling method was used followed by the referral method of the snowball approach. Initially, 95 bank customers across Tier-1 cities were contacted, which possessed the requisite qualities and satisfied the specific purposes to be fit for the research being undertaken thus considering them appropriate for the first stage study (Saunders et al., 2019). Thereafter, these 95 customers were asked to refer further respondents in the given sample frame.

Later, the questionnaire was circulated to 465 bank customers, out of which 421 filled questionnaires were received, leading to a response rate of 90.53%. Out of the responses received, 22 invalid responses were discarded from the data set.

Due to diverse data collection requirements, about 60% of data was collected manually and the rest 40% was collected through a questionnaire developed using Google Forms. This method is recognized as one of the securest forms of data collection in the industry. Major benefits of Google form-based surveys are ease of use, widespread availability, and better security (Rayhan et al., 2013)

Data Cleaning and Reliability Assessment

A preliminary questionnaire was administered as a pilot study and distributed to 124 respondents. The data collected were further assessed for reliability and validity. The final questionnaire was then used for data collection, and data from 399 respondents collected, was analyzed using exploratory factor analysis (EFA).

The data set was imported into Excel and checked for missing values and reliability. The missing completely at random test was applied to handle the missing values. The test was found to be insignificant which implies that data were missing randomly and therefore can be directly deleted (Li, 2013). This led to the deletion of 11 responses and resulted in a total sample of 388. The demographic characteristics are shown in Table 2. All the statements were assessed for reliability and each statement was found to have Cronbach’s α in the range of .80–.86.

Table 2
Demographic Details

Demographic variable

Frequency

Percentage value (%)

Gender

 Male

181

46.56

 Female

207

53.45

Age

 18–25

43

11.08

 26–33

89

22.94

 34–41

95

24.48

 42–49

84

21.65

 50 and above

77

19.85

Occupation

 Own business

69

17.78

 Homemaker

59

15.21

 Government employee

63

16.24

 Private sector employee

61

15.72

 Student

32

8.25

Which bank do you mostly transact with? Mention the name of one bank

 Axis Bank

65

16.75

 Bank of Baroda

69

17.78

 Bank of India

52

13.40

 Housing Development Finance Corporation bank

64

16.49

 Industrial Credit and Investment Corporation of India bank

51

13.14

 State bank of India

68

17.53

 Others (Union Bank of India, Yes Bank, Kotak Mahindra, Citi, Central Bank of India, Andhra Bank, etc.)

19

4.90

Note. Where n = 388.

Results

Factor Assessment

In the present study, the variables were defined through literature and found to be higher in number. In addition, sample demographics differ largely from the sample of originally validated constructs. Hence, it was necessary to apply EFA on all the variables of each construct individually to verify the original priori structure. For these items developed to measure self-reported actual usage, the study accepts the proposition of Shaw (2011) to identify types of usage by way of EFA. Where, the study suggests that omnibus metrics like frequency and duration have overly broad definitions, and EFA is used to make them lean to get a granular understanding of each dimension derived. For instance, if two users are compared in the current context, one may spend many hours each day using only one feature (say, trading of shares) while the other may spend less time but utilize numerous other functions. Hence, to assess how much an organization, group, or person is utilizing the features provided by the IT system, the dependent variable of actual usage must be defined (Burton-Jones & Straub, 2006).

Researchers suggest conducting EFA in the initial part of the analysis which presents the factors that are further confirmed through confirmatory factor analysis (Fabrigar et al., 1999; Sharma & Prasad, 2018). Confirmatory factor analysis is estimated through a measurement model in PLS-SEM. For this purpose, a random sample of 253 is extracted using simple random sampling from the complete sample of 388 using the university edition of Statistical Analysis System. This subsample of n = 253 (Sample A) is subjected to EFA and measurement and the structural model is then tested on a complete sample of n = 388 (Sample B) to ensure the reliability of the complete data set. Exploratory factor analysis was applied separately to 20 variables measuring actual usage as it was a formative construct.

Actual usage was found to be a second-order formative construct, with four first-order reflective dimensions. The items with lower factor loadings AU2, AU, AU9, AU10, AU18, and AU24 were dropped as the values were less than 0.7. Complete statistics have been presented in Table 2. All the items used for measuring actual usage were found to be combined into four composite factors and were named core banking services, regular transactions, digitalization transactions, and augmented level transactions explaining 24%, 19%, 15%, and 13%, respectively, from the total variance of 71% approximately.

Model Estimation

A PLS path model entails two broad components. First, the measurement model which is also called an outer model in PLS-SEM. These represent the relationship of the construct (circles) and its respective indicator items (rectangles). Later, a structural model called the inner model is constructed in PLS-SEM. This structural model indicates the relationship or paths from one construct to another. Figure 2 attempts to conceptualize the proposed model of the study to be tested, demarcating measurement and structural model.

Figure 2

Conceptualizing Model Estimation for the Study

Evaluation of Measurement Model

The measurement model was evaluated by assessing all constructs of the model including the reflective and formative models of measurement.

In the final run of the measurement model through the PLS Algorithm as represented in Figure 3; it was found that innovativeness (INN) had low outer weight. Hence iteratively, augmented level and core banking services were removed from the model, which helped in improving overall indices of actual usage construct. The improved values can be seen in the following model output.

Figure 3

PLS Algorithm Output (Drawn in SmartPLS Software V3.3.3)
Note. TR = technology readiness; OPT = Optimism; INS = Insecurity; DIS = Discomfort; BI = behavioral intention; AU = actual usage.

Assessment of Reflective Measurement Model

While assessing the reflective measurement model, four aspects of the model results are observed, namely, indicators’ outer loadings for testing individual indicator’s reliability; Composite reliability to assess the internal consistency of parameters and constructs; to examine convergent validity, average variance extracted (AVE) is observed, lastly to measure discriminant validity, Fornell–Larcker criterion and cross-loadings are observed.

Outer loadings for indicators of reflective constructs show individual indicator’s reliability. It is observed from Table 3 that the outer loadings of all indicators of reflective constructs were more than the minimum acceptable value (0.7). The items with lesser values, INN1 and INN2 (innovativeness) were dropped from the model. Thus, the innovativeness construct had to drop completely as there was no dimension to measure it.

Composite reliability of the reflective constructs is presented in Table 3. It was found that the reflective constructs portrayed a fairly good composite reliability, all above 0.7 cutoff values demonstrating great levels of internal consistency and reliability among all eight reflective constructs.

Table 3
Individual Indicator Reliability

Item

Augmented transaction

BI

Core transaction

Digitalization transactions (digi push)

Discomfort

Insecurity

Optimism

Regular transactions

AU1

0.720

AU17

0.931

AU20

0.941

AU16

0.778

AU19

0.923

AU6

0.987

AU7

0.772

AU8

0.773

BI1

0.834

BI2

0.884

BI3

0.909

DIS1

0.784

DIS2

0.813

DIS3

0.796

DIS4

0.820

INS1

0.832

INS2

0.855

INS3

0.849

OPT1

0.894

OPT2

0.872

OPT3

0.805

OPT4

0.759

Note. Sample B, n = 388. AU = actual usage; BI = behavioral intention; DIS = discomfort; INS = insecurity; OPT = optimism.

Convergent validity of the model is evaluated through the AVE value and shown in Table 4. All the reflective constructs depicted satisfactory levels of convergent validity as all the AVE values were above 0.5 for all reflective constructs.

Table 4
Composite and Convergent Reliability

Construct

Cronbach’s α

rho_A

Composite reliability

AVE

AU_

.849

0.852

0.898

0.689

BI_

.848

0.849

0.908

0.768

Discomfort

.817

0.818

0.879

0.645

Insecurity

.801

0.801

0.883

0.715

Optimism

.853

0.860

0.901

0.696

TR

.886

0.888

0.907

0.593

Note. Sample B, n = 388. TR = technology readiness; BI = behavioral intention; AU = actual usage; AVE = average variance extracted.

Discriminant validity is assessed by observing the values of Fornell–Larcker criterion and cross-loadings. The criterion states that the construct’s highest correlation with any other construct in the model should be lower than the square root of the AVE of each construct.

In Table 5, the Fornell–Larcker criterion results are shown. Discriminant validity was assessed by observation of cross-loading values, where it was found that all constructs were seen to be distinguished by respondents as the values accorded to the set threshold and rules of the criterion, compiling the given criteria.

Table 5
Fornell–Larcker Criterion

Construct

AU_

BI_

Discomfort

Insecurity

Optimism

TR

AU_

0.830

BI_

0.102

0.876

Discomfort

−0.027

0.440

0.803

Insecurity

−0.001

0.575

0.651

0.846

Optimism

0.103

0.978

0.472

0.586

0.834

TR

0.032

0.810

0.831

0.875

0.818

0.702

Note. Bold values indicate all constructs were found to be significant. Sample B, n = 388. TR = technology readiness; BI = behavioral intention; AU = actual usage.

Assessment of Formative Model

AU is a formative construct in the PLS path model, and it is assessed for validity for the collinearity issue and the significance and relevance of indicators (by checking outer weights and outer loadings).

Collinearity Issue

All indicators of AU have a variance inflation factor value <5 as shown in Table 6. Hence, there is no collinearity issue present between the indicators.

Table 6
Outer VIF Values

Item

VIF

AU1

1.310

AU17

1.145

AU17

2.107

AU20

1.353

AU20

1.811

AU25

1.294

AU27

1.294

AU27

1.797

AU6

1.310

AU6

2.593

AU7

1.145

AU8

1.353

Note. Sample B, n = 388. VIF = variance inflation factor.

Assessment of Significance and Relevance of the Formative Indicators

Table 7 represents the outer weight of each indicator (an item’s relative contribution to the given construct), and outer loading (an item’s absolute contribution to the given construct) was examined and used the method of bootstrapping results to assess their significance. We observed that only four out of eight indicator weights were significant whereas indicators’ outer loadings were found substantial (where, the value of outer loadings >0.5).

Table 7
Significance Values for Outer Weights and Loadings

Item > Construct

Outer weight

Outer loading

p

Significance

AU1 > Augmented

0.183

0.720

.000

***

AU17 > Digi push

0.792

0.931

.000

***

AU20 > Core

0.739

0.941

.000

***

AU16 > Regular

0.438

0.778

.000

***

AU19 > Regular

0.714

0.923

.000

***

AU6 > Augmented

0.898

0.795

.000

***

AU7 > Digi push

0.390

0.987

.000

***

AU8 > Core

0.395

0.772

.000

***

Note. Sample B, n = 388. AU = actual usage.

***Indicator’s outer weight and loadings are significant at 95% confidence interval.

The thumb rule in this context of suggests that if any indicator’s outer weight and loadings are insignificant, it should be dropped. Second, if the outer weights are not significant but their respective outer loading is significant (i.e., higher than 0.50), in such a situation the item is mostly retained as it is interpreted as absolutely important but not relatively important (Hair et al., 2011). Therefore, in the run-1, AU11, AU13, AU15, AU14, AU18, AU10, AU12, AU5 were dropped, whereas AU1, AU17, AU20, AU16, AU19, AU6, AU7, AU8 were retained in the model. Values of outer weights and outer loadings of indicators of actual usage with significance status are shown in Table 6. Outer loadings of all indicators are found relatively high (i.e., >0.70). Moreover, the items of the actual usage were scarce in technology adoption literature, as it has been studied as a reflective construct in the existing studies. Hence, it was crucial to validate the developed items before testing the structural model.

Assessment of Structural Model

In context to PLS-SEM, one of the most important evaluation matrices for measurement and structural model are common to CB-SEM covariance based structural equation modeling such as reliability, convergent validity, and discriminant validity (common in measurement model analysis); for structural model, the important ones like R2 also known as explained variance, f2 as the effect size, Q2 referred as predictive relevance, also including the values of size and the statistical significance of the structural path coefficients. Subsequently, PLS-SEM is governed by variances instead of covariances to define an optimum solution, thus covariance-based model indices or the goodness of fit measures provided in Analysis of Moment Structures software are not relevant in context to PLS-SEM (Hair et al., 2016).

The PLS-SEM-based fit measures are the following:

  1. Collinearity issues of the structural model

  2. The significance values portraying the relevance of the structural model relationships

  3. The level of R2

  4. The effect sizes f2

  5. The predictive relevance Q2 effect sizes

  6. SRMR

Collinearity Issues of Structural Model

The collinearity issue of the constructs was assessed by validating variance inflation factor values which should be less than 5. The variance inflation factors of constructs were found <5; hence, we concluded that the collinearity issue is not present between the constructs.

Assessing the Significance and Relevance of the Structural Model Relationships

PLS algorithm calculation using SmartPLS software which provided the R2 values of the endogenous constructs are shown inside the circles (see Figure 3). The bootstrapping procedure reports significance of path coefficient values. These path coefficient values (in between +1 to −1) are used for analyzing the strength of the hypothesized theoretically formed relationships. The values of path coefficients approximately nearing +1 denote a strong significant relationship whereas a value near 0 symbolizes a weaker relationship.

It also provides empirical t statistics’ (obtained by dividing path coefficient value by standard error) and “p values” (referred to the probability associated with the null hypothesis to erroneously reject it). The empirical t value is then compared to the critical value to see if it is greater than the critical value which is desired. Where, the values 2.57, 1.96, and 1.65 for a significance level of 1%, 5%, and 10%, respectively (two-tailed tests) are, which are universally accepted critical t values. The significance of path coefficients for our model as per the bootstrapping report is presented in Table 8.

The magnitude of the path coefficient provides us with the relevance of that path. Hence, it is found that insecurity has the highest relevance in predicting technology readiness, followed by discomfort and optimism. The study also reveals that if technology readiness changes by one standard deviation, behavioral intention will increase by 45.743. Similarly, if behavioral intention changes by 1 SD, actual usage will increase by 41.254.

Table 8

T Statistics and p Values

Relationship (Construct_ → Item)

T statistics (O/STDEV)(|O/\text{STDEV}|)

p

AU_ → Augmented

74.289

.000

AU_ → Core

49.444

.000

AU_ → Digi push

52.823

.000

AU_ → Regular

41.668

.000

BI_ → AU_

41.254

.000

TR → BI_

45.743

.000

TR → Discomfort

38.202

.000

TR → Insecurity

46.530

.000

TR → Optimism

36.713

.000

Note. The significant relationships are highlighted in bold. Sample B, n = 388. STDEV = standard deviation; AU = actual usage; BI = behavioral intention; TR = technology readiness.

Coefficient of Determination (R2 Value)

The coefficient of determination (value of R2) depicts the structural model’s projected accuracy and is intended through the squared correlation among a specific endogenous construct’s actual and estimated values (Hair et al., 2014). The R2 gives the joint effects of exogenous variables on the endogenous variable, that is, it signifies the amount of variation within the dependent constructs is explained through all the exogenous constructs associated with it (Hair et al., 2014). The R2 value ranges between 0 and 1, where the values nearer to 1, signify high predictive accurateness.

The R2 value of actual usage (dependent variable) for this study is 0.710, that is, the combined effect of all the independent variables can cause a 71.0% variation in usage of ITeBS. For behavioral intention, the R2 value is obtained as 65.6%. Acceptable R2 indices are dependent on the model framework from simple to complex and differ with various genres of research subjects (Hair et al., 2014).

This implies that prior studies on TAM have reported R2 in the range of 0.4–0.5 (Singh, 2016). The model under study is higher than the values of established models with R2 value of 71.0%.

Effect Size (f2)

The change or the delta of R2 value, when an exogenous construct is absent from the model can be used to estimate whether the missing construct has a noteworthy influence on the endogenous constructs (Hair et al., 2014).

Effect size value is calculated using Equation 1:

f2=(R2includedR2excluded)(1R2included),f^{2} = \frac{({R^{2}\,\text{included} - R^{2}\,\text{excluded}})}{({1 - R^{2}\,\text{included}})},


where the two values R2 included and R2 excluded are the R2 values of the endogenous latent variable when a particular exogenous latent variable is included in or excluded from the model. The acceptable limits for assessing f2 values are 0.02, 0.15, and 0.35, respectively, these values represent small effect (less than 0.02), medium effect (less than 0.15), and large effects (less than 0.35) of the exogenous latent variable (Cohen, 1988).

It is observed by evaluating the below Tables 9 and 10 that the effect size of the technology readiness variable on BI, BI on AU, and all other relationships represent a high effect size (>30%).

Table 9
Coefficient of Determination, R2 Values

Construct

R2

Augmented

0.803

AU_

0.710

BI_

0.656

Core

0.753

Digi push

0.778

Discomfort

0.690

Insecurity

0.766

Optimism

0.669

Regular

0.742

Note. Sample B, n = 388. BI = behavioral intention; AU = actual usage.

Table 10
Effect Size, F2 Values

Construct

Augmented

AU_

BI_

Core

Digi push

Discomfort

Insecurity

Optimism

Regular

AU_

4.078

3.049

3.500

2.882

BI_

0.810

TR

1.906

2.227

3.266

2.024

Note. The significant relationships are highlighted in bold. Sample B, n = 388. STDEV = standard deviation; AU = actual usage; BI = behavioral intention; TR = technology readiness.

Predictive Relevance (Q2)

While the R2 values denote predictive accuracy the predictive relevance Q2 specifies the predictive relevance of the model, also called “Stone-Geisser’s Q2 value” (Hair et al., 2011). The Q2 values should be greater than 0, for assuring that a reflective endogenous latent variable and their associated path model possess predictive relevance for the construct (Hair et al., 2014).

After running the blindfolding procedure, the values of Q2 were found to be greater than zero. Q2 for actual usage and behavioral intention was found to be 0.486 and 0.410, respectively, which indicates that path model in the given study has high predictive relevance.

Standardized Root-Mean-Square Residual and Normed Fit Index

The concept of standardized root-mean-square residual (SRMR) was introduced by Henseler et al. (2014) as a model fit indices measure for PLS-SEM. The SRMR is calculated by finding the difference between the observed correlation and the predicted correlation. It provides an average magnitude of the divergences between the mentioned correlations as an absolute figure to measure (model) fit criterion. The model should at least be 0.08 or at most less than 0.10 is regarded as a good fit criterion (Hair et al., 2014)

Bentler–Bonett index or normed fit index (NFI) is observed for factor models (Bentler & Bonett, 1980). NFI values above 0.90 are considered as acceptable (Byrne, 2013). The PLS bootstrapping procedure provides the SRMR criterion and NFI as shown in Table 11.

Table 11
SRMR and NFI Values

Model fit measure

Saturated model

Estimated model

SRMR

0.086

0.082

NFI

0.905

0.912

Note. Sample B, n = 388. SRMR = standardized root-mean-square residual; NFI = normed fit index.

It is observed that SRMR values are within the range of 0.08–0.10; hence, we can say that the model is meeting the goodness of fit criteria.

Results of Hypothesis Testing

The present research model proposed three hypotheses for predicting the dependent variable (actual usage). The first hypothesis is not supported as technology readiness is a three-dimensional construct, with optimism, insecurity, and discomfort being significant factors. Innovativeness is found to be insignificant.

The remaining two hypotheses, testing direct relations between technology readiness (independent variable) and behavioral intention; and between behavioral intention and actual usage were found to be significant.

Discussion

Managerial Implications

The main purpose of this article was to go beyond adoption and study postadoptive behavior as a compliment to it. Several implications for banks providing ITeBS can be derived from our results. First, this study will provide a basis for future research to explore different usage patterns predicting actual behavior in varying contexts and might be useful across industries integrating technology to offer products and services (Bölen & Özen, 2020; Rabaa’i et al., 2021). It helps marketers to get real, factful insights that could not be obtained in intended behavior measuring studies and suggests that banking marketers focus on strategies of sustained use by customizing for varied types of transactions based on their utility for customers and avoid the traditional push of adoption.

This study also proves that technological readiness (TR) has a significant and positive effect on buying intention. Hence, the banks should provide users with a safe and user friendly online banking service to limit the inhibitors of discomfort and insecurity. They should also assist customers in overcoming difficulties while using ITeBS. Banks could also promote knowledge and usefulness about ITeBS through pull marketing techniques.

The study proposes to limit the inhibitions of insecurity and discomfort to enhance the readiness of the customers. The discomfort could be reduced by delivering service simply nine or 10 times better than the offline banking mode (branch banking), providing customers a disproportionate benefit to using ITeBS. This would create a secure environment for ITeBS usage, leading to behavioral change to accept ITeBS channels over their traditional branch banking.

There are also key takeaways for communication and advertising, for example, strategizing to persuade customers to overcome concerns about using technology by positioning their product in a way that reduces the customers’ distrust and inhibitions to use (Sebastian et al., 2023; Son & Han, 2011). Marketers can also think of providing them motivation in the form of cashback and reward points to embrace the technology for their betterment and eventually become comfortable once they start reaping its true benefits.

Limitations and Future Scope

This study comes with a few limitations, the aspects considered during this research making this article restricted in some respects are: First, the proposed research framework was purposefully opted for one services sector, that is, banking. Hence, the model constructs, although realistic enough to form a basis for further research, may not reflect the true objective and model formation in other service domains.

There exists scope for future studies in terms of adapting longitudinal design which further expands the study to test the long-term impact of behavioral constructs like repurchase intention, willingness to recommend, or satisfaction. Also, the model framed can be used for future research to take up payment wallets integrating government digital interface—Bharat Interface for Money app as it is gaining momentum at a very high rate.

The study is also constrained by way of employing EFA and confirmatory factor analysis on identical data sets, instead of employing a split-sample approach, where a better approach is to have a first subsample sample for EFA and another half of sample for confirmatory factor analysis (Byrne, 2013) or either the three-faced construct validation method that follows 20-40-40 method (Kyriazos, 2018).

This study has launched into greenfield areas of usage research and hence opens several avenues of further research. Researchers may incorporate some more variables into the model and further refine the existing one.

The actual usage measure incorporates both the breadth of use which is the number of different applications/features and depth of use which are shown as the frequency of use. According to Burton-Jones and Straub (2006), further research can be developed by adding more structural elements, like those related to user and tasks and even range of use (Eder & Igbaria, 2001) to investigate the explanatory power of behavioral intention and usage. However, the self-reported actual usage data has its inherent limitations, which suggest that it is commonly used to assess self-regulation and is susceptible to reference bias, a systematic error resulting from differences in implicit behavior evaluation standards (Lira et al., 2022).

Further, Son and Han (2011) identified three categories of usage patterns pertinent to high-tech products, that is, the usage rate of basic utilities, the usage rate of innovative utilities, and variation in uses of innovative utilities. Hence, usage dimensions could act as a key moderator between users’ TR and behavioral outcomes.

Last, it would be worthwhile to find the moderating role of age, gender, and other demographic variables influencing the users’ technology readiness, usage, and behavioral outcomes. These insights would provide appropriate recommendations for banking marketers to devise strategies for varied segment characteristics.

Conclusion

Parasuraman developed four constructs determining technology readiness. Out of these four perceptual factors, two are enablers (optimism and innovativeness) and the other two are inhibitors (discomfort and insecurity) to technology (Parasuraman, 2000). This study expresses technology readiness in three factors namely, optimism, insecurity, and discomfort.

The objective to explore the relationship between one’s technology readiness and his/her intention to use ITeBS can be interpreted through these results that out of the total customers ready for using ITeBS, 81% of the total portray their intention toward using it. The results confirm the findings of Chen and Chen (2009); Chen (2012); Curran and Meuter (2005) and J.-S. C. Lin and Pei-ling (2006) which state TR is a critical factor for behavioral intention to use. The method of analysis adopted in this research is unique in establishing a direct relationship between readiness and intention to use. The results also signified that out of the total customers intending to use ITeBS, 70.2% predict actual usage.

Actual Usage and Its Nature

Based on the analysis mentioned in the previous sections, it was found that the dimensions of actual usage are not reflective but formative. It is not a compulsion for a customer to use all ITeBS offered by banks. For instance, a customer may use auto bill payment services but not invest in mutual funds using ITeBS or any other usage category for that matter. Hence, it is justified to be a formative construct. Whereas, in the case of a reflective construct, if a respondent agrees with the construct’s one dimension, he would agree with others as well. Thus, dropping one of the items from the construct would not change its essence, which is a with a formative construct. Therefore, based on the above, this study used a PLS-SEM approach over covariance based structural equation modeling (Dash & Paul, 2021).

Exploratory factor analysis is applied to evaluating the usage depth as recommended by Shaw (2011). Where, the results conceptualize the actual usage measurement model as a second-order formative-reflective construct with four first-order reflective constructs based on a four-dimensional view of actual usage of IT-enabled banking services (ITeBS) comprising core banking services, regular transactions, digitalization transactions, and augmented level transactions. Such a form of multidimensional construct is categorized under “Type II” (Diamantopoulos et al., 2008; Jarvis et al., 2003) of aggregate construct form. According to Wong et al. (2008) in this type of aggregate construct, the dimensions are algebraically merged to create the construct, and the indicators for each dimension are various expressions (reflections) of that dimension. In simple terminology, an aggregate construct is referred to as a construct in which the relationship flow is from the dimensions to the construct. Where, the majority of research models incorporating reflective first-order, formative second-order constructs are tested using PLS (Polites et al., 2012).

Theoretical Contribution and Implications

The present study, firstly incorporated the role of technology readiness instead of TAM within the context of ITeBS as the need was to comprehend the role of individual consumer traits in the acceptance of new technologies. Research outcomes confirm the results of Chen and Chen (2009); Chen (2012); Curran and Meuter (2005) and J.-S. C. Lin and Pei-ling (2006) that TR is a critical factor for behavioral intention to use IT-enabled banking services.

The theory of technology readiness index evolved in the current context, as it was found to have three dimensions unlike the four proposed earlier by the theory. Dimensions of TR have been observed to portray varied forms in varied demographics. Previous research has treated TR as a four-dimensional construct and examined the individual effect of each dimension (Lam et al., 2008; Son & Han, 2011). Also, some studies have used a two-dimensional model to conceptualize TR regarding motivators and inhibitors (Jin, 2013). Liljander et al. (2006) discovered that two of the enablers of technology readiness clubbed into one unique dimension. The present study tests TR in the context of India and found innovativeness to be insignificant and only three dimensions of technology readiness (optimism, discomfort, and insecurity) were found to be significant. This evidence coincides with the results of Negm (2023) study, about intention to use IoT in higher education online learning. The likely reason for this is the enhanced user experience provided by banks’ mobile applications and websites, which helps users navigate their needs comfortably.

As per the diffusion of innovation theory, it is said that “When it comes to adopting new technologies, people with a high level of innate innovativeness (i.e., openness to new ideas) are anticipated to naturally demonstrate an interest in the new technology and end up as innovators or early adopters (Rogers, 1995). Rampant digitalization during the pandemic and previously postdemonetization outlines, that adopting ITeBS had changed the Indian way of life and had been more compelling rather than impulsive (Nagdev et al., 2021; Setia, 2021)”. Hence, it is justified to report that innovativeness is insignificant in explaining technology readiness in India concerning IT-enabled banking services, forming a valuable contribution to the theory of technology readiness.

This study contributes to the already existing theoretical base by linking the TR indices with behavioral intention and also importantly studies sustained and recurring usage. Typically, earlier literature had just concentrated on intentions of usage, this study goes on to investigate postadoptive behavior and offers fine-grained insights. The formation of scale items for measuring actual usage, consecutively finding the four reflective dimensions of usage provides an insightful contribution to technology acceptance literature. Also, the study followed the approach of Burton-Jones and Straub (2006) and Venkatesh et al. (2012), by focusing on technology use as the system element (i.e., breadth and extent/depth of the use) in the current context.

The novelty lies in the major finding of this study. Where, the measurement of actual usage was found to be a formative index of 20 exhaustive items (detailed in Appendix) of usage categories possible through all ITeBS channels, measured on consumers’ usage frequencies. Thus, the measure includes both the breadth of use (i.e., the number of different applications/features) and depth of use (i.e., the frequency of use). The 20 items for measuring self-reported usage were extracted through exploratory analysis structured as a second-order formative construct, with four first-order reflective dimensions. Whereas the technology acceptance literature has majorly measured usage as a reflective measure, the seminal studies TAM and Unified Theory of Acceptance and Use of Technology 2, that incorporate usage have also measured usage by way of two and six dimensions/items, respectively (Oertzen & Odekerken-Schröder, 2019).

Video Summary

Appendix

Factors of Technology Readiness, Behavioral Intention and Actual Usage After Exploratory Factor Analysis

Factor

Variable label

Item

Factor loading

Optimism

OPT1

IT-enabled banking services (ITeBS) simplifies my work, contributing to a better quality of life

0.892

OPT2

ITeBS enables me to do anywhere banking, giving me more freedom of mobility

0.86

OPT3

ITeBS give people more control over their daily lives without any intervention of others.

0.836

OPT4

ITeBS makes me more productive in my personal life

0.823

Innovativeness

INN1

Other people come to me for advice on new technologies

0.794

INN2

In general, I am among the first in my circle of friends to acquire new technology when it is introduced

0.773

Discomfort

DIS1

I do not wish to take support from any bank staff, as I feel that because of my ignorance they may exploit me

0.73

DIS2

Technical support systems are not helpful because they do not explain things in terms I understand

0.703

DIS3

Sometimes, I think that ITeBS are not designed to be used by ordinary people

0.75

DIS4

There is no such thing as a manual for the ITeBS that is written in plain language

0.721

Insecurity

INS1

People are too dependent on ITeBS to do their things.

0.701

INS2

Too much use of ITeBS distracts people to a point that is risky

0.723

INS3

ITeBS lowers the quality of relationships by reducing personal interaction with bank staff

0.798

Behavioral intention

BI1

I intend to conduct banking transactions by using ITeBS in the future.

0.804

BI2

I intend to learn how to use ITeBS in the future

0.801

BI3

I will highly recommend ITeBS to others.

0.724

Core banking services

AU11

Tax payment (income tax/GST)

0.857

AU13

Instant IMPS

0.832

AU15

Open fixed deposit/recurring deposit using mobile banking app

0.802

AU8

Using credit card for all purposes

0.789

AU20

Loan services

0.778

Regular transactions

AU14

Seeking information like balance enquiry or locating nearest ATM/branch

0.852

AU16

Cheque services using mobile and internet banking

0.816

AU19

Fund transfers

0.815

AU10

Viewing and emailing e-statement

0.74

AU18

Withdrawn amount from ATM

0.754

Digitalization transactions

AU7

Online bookings

0.748

AU17

Utility bill payments

0.755

AU12

Redeeming reward points (points accumulated by using ITeBS)

0.812

Augmented level transactions

AU1

Apply IPO shares or trading of shares

0.804

AU5

Upload import–export documents on portal for customs approval

0.801

AU6

Invest in mutual funds using mobile banking app

0.724

AU2a

Installed point-of-sale machine at workplace

0.622

AU3a

Auto bill payment services

0.543

AU9a

Used payment portal of your bank for, for example, SBI Buddy

0.506

AU4a

TDS enquiry

0.521

INN3a

I can usually figure out new high-tech products and services without help from others

0.515

INN4a

I keep up with the latest technological developments in my areas of interest

0.5

INS4a

I do not feel confident doing business with a place that can only be reached online

0.452

Note. IT = information technology; OPT = optimism; DIS = discomfort; BI = behavioral intention; AU = actual usage; GST = Goods and Services Tax; IMPS = Immediate Payment Service; ATM = automated teller machine; IPO = initial public offering; SBI = State Bank of India; TDS= Tax Deducted at Source.

a Depicts for the items dropped with lower factor loadings where the values were less than 0.7.


Received November 7, 2023
Revision received March 6, 2024
Accepted March 11, 2024
Comments
0
comment
No comments here
Why not start the discussion?