The adoption of internet banking: An institutional theory perspective
Received (in revised form): 30th October, 2007

Weihua Shi* is a visiting assistant professor at the University of Southern Mississippi. She received her MA degree in Economics from Vanderbilt University and her PhD in Finance from the City University of Hong Kong. Her research interests include behavioural finance, risk management and banking.

Narcissus Shambare is an assistant professor of Business Information Systems (BIS) and Director of BIS programmes at the College of Saint Mary, Omaha, Nebraska, USA. He has more than 20 years experience in business management in manufacturing, wholesale and retail, services and banking industries. His research interests include information security, strategic management and business education.

Jian Wang is a professor of International Business at the University of International Business and Economics. His research interests include e-business, international business and international trade law and practice. Dr Wang is Director of the International Business Bridge (IBB). He has extensively published in the areas of international business, electronic business and others.

Abstract Several models have been proposed in the literature to understand technology acceptance in which social environmental factors are not primarily addressed. Drawing upon the social contagion theory and institutional theory, this study proposes a model to examine three social environmental factors of normative, coercive and mimetic pressures within the internet banking (IB) context. The model is tested using survey data from 124 respondents. The results reveal that normative and coercive pressures significantly influence the attitude and intention of adopting IB, while mimetic pressures appear not to. Attitude plays a mediating role between institutional pressures and IB adoption. This study spearheads an institutional analysis at the individual level within the IB context. The findings also provide valuable insights for bank managers to manage the IB services. Journal of Financial Services Marketing (2008) 12, 272–286. doi:10.1057/fsm.4760081 Keywords Internet banking, institutional theory, social contagion

INTRODUCTION The increasing popularity of the internet has produced a paradigm shift in the method of conducting business and has created a vast number of challenges for businesses to conduct transactions using this new distribution channel.1 Banking is a very responsive industry that can find value in the use of internet technologies, so an increasing
*Correspondence: College of Business, University of Southern Mississippi – Gulf Coast, Long Beach MS 39560, USA. Tel: + 1 228 865 4505; Fax: + 1 228 865 4588; e-mail: Weihua.shi@usm.edu

number of banks and financial services companies have been evolving their approach from a ‘bricks and mortar’ to a ‘clicks and mortar’.1–3 Thus, for the success of most banks, it has become critical in attracting existing bank customers to adopt the options of internet banking (IB). IB provides both banks and consumers with benefits. On one side, by offering IB services, banks can benefit from having fewer staff and physical branches. It also creates huge cost savings through scale effects in bank operations.4 On the other side, IB can be extremely beneficial to consumers by providing cost and time savings, reduced

272

Journal of Financial Services Marketing

Vol. 12, 4 272–286 © 2008 Palgrave Macmillan Ltd 1363-0539 $30.00 www.palgrave-journals.com/fsm

The adoption of internet banking

dependency on time and location, quick responses to complaints, more service variety and improved services quality.5–7 Owing to the enormous benefits for both parties, IB has motivated both banks and consumers to offer and use IB services, respectively. In fact, IB adoption in the US has averaged 15 per cent year-to-year growth during the 2001–2006 period.8 However, as internet technology is a rather recent innovation, it is still a great challenge of overcoming the resistance of existing bank customers to adopt this new technology in handling their financial transactions.4 It is critical, therefore, to understand what factors facilitate or inhibit IB adoption. A considerable number of studies in the literature have examined IB adoption. Most of these studies were from technical and customer personal perspectives. For example, Shih and Fang1 and Liao and Shao9 examined the influences of attitude and norm on behavioural intention and actual usage of IB using the Theory of Reasoned Action (TRA) and the Theory of Planned Behavior (TPB). Chau and Lai,2 Cheng et al.,7 Wang et al.,10 and Sukkar and Hasan11 investigated perceived usefulness and perceived ease of use of IB using the Technology Acceptance Model (TAM). Tan and Teo12 investigated the impact of consumer’s attitude, norms and perceived behavioural control on IB adoption. Hernandez and Mazzon13 proposed and tested a more comprehensive model in which the consumer’s innovation characteristics (eg relative advantage, demonstrability and visibility), norms and individual characteristics (eg education, age, gender and income) were proposed to influence the intention and actual usage of IB. However, due to the exposure to the social environment, social actors (eg individuals, groups and organisations) are likely to develop their beliefs, attitudes and behaviours consistent with those of their environments.14 Therefore, we speculate that social factors also play an essential role in influencing IB

adoption. However, there is a paucity of studies that have adequately addressed the social factors in IB adoption context. To fill this gap, drawing upon the social contagion theories and institutional theories, this study postulates that coercive, normative and mimetic social forces are also significant determinants of IB adoption. In the following sections, the theoretical background, conceptual model and research hypotheses are presented, followed by a discussion of the research methodology. This is followed by the analyses and results. Finally, the managerial and theoretical implications, limitations and conclusions of the study are presented. THEORETICAL BACKGROUND AND CONCEPTUAL MODEL Theoretical background The social contagion theory argues that the social actors’ (eg individuals, groups and organisations) beliefs, attitudes and behaviours are related to the beliefs, attitudes and behaviours of others in the network to which they are linked. To be more specific, within the technology acceptance context, social contagion refers to the social actors’ attitudes of technology adoption that are significantly influenced by other actors (eg family and colleagues for individuals, customers, suppliers and partners for companies) who have direct connections or share similar social networks. The underlying logic of the influences of social ties on social actors’ attitudes and behaviours towards innovation adoption is that the social ties may have built up a collection of implicit rules, which may be both imposed on and upheld by the actors’ attitudes and behaviours. In the literature, there are several innovation adoption theories that consider these implicit and implied rules. The TRA, proposed by Fishbein and Ajzen,15 posits that behavioural intentions are determined by an

© 2008 Palgrave Macmillan Ltd 1363-0539 $30.00 Vol. 12, 4 272–286

Journal of Financial Services Marketing

273

Shi, Shambare and Wang

individual’s attitude toward the behaviour and subjective norms surrounding the performance of the behaviour. The TPB, proposed by Ajzen,16 is an extension of TRA. It posits that behavioural intentions are a function of an individual’s attitude toward the behaviour, the subjective norms and the individual’s perception of behavioural control. The TAM is an adaptation of TRA in the information systems (IS) field and posits that perceived usefulness, perceived ease of use and subjective norm determine an individual’s intention to adopt an innovation.17 The Model of Personal Computer Utilization (MPCU) proposed such factors as technology complexity, facilitating conditions and social factors as influencing innovation adoption.18,19 In Roger’s and Moore and Benbasat’s Innovation Diffusion Theory (IDT),20,21 factors such as relative advantage, ease of use and image are postulated to influence individual technology acceptance. Parasuraman22 proposed the Technology Readiness Index (TRI) to measure readiness to embrace technologybased products and services. In this model, customers’ propensity to adopt new technologies is from ‘an interplay between drivers (optimism, innovativeness) and inhibitors (discomfort, insecurity) of technology readiness’ (p. 317). However, only personal factors, rather than any social factors, are considered in this model. Although most of these models consider various social influence factors, such as subjective norm in TRA, TPB and TAM, social factors in MPCU, and image in IDT, the discussion was generally fragmented and there is a lack of specific focus on such issues.23 Therefore, in the present study, we apply institutional theory to organise these factors and expand their depth and breadth. The institutional theory posits that the beliefs, attitudes and behaviours of individuals and groups are strongly influenced by various networks and interactions.24 It also addresses the role of institutions in explaining the behaviour of social actors, and provides a

perspective which can help to assess the formal and informal rules (ie institutions) that can strongly shape the beliefs, attitudes and behaviours of social actors.25–27 The institutional theory holds that the institutions’ influences on the beliefs, attitudes and behaviours of social actors are subtle but pervasive. Scott24 defined institutions as ‘social structures that have attained a high degree of resilience’ (p. 48). Once internalised, that is encoded into actors through a socialisation process, institutions transform to a pattern of attitudes and behaviours, which will shape actors’ future attitudes and behaviours and provide stability and meaning to social life. After institutions are established, they become authoritative guidelines for social behaviours and are taken for granted.28 Afterwards, the actors may not even realise that their behaviours are in fact partly shaped by institutions. The institutional theory focuses on social actions taken to gain legitimacy rather than for monetary or utility optimisation.29 It suggests that once institutions are established, they create constraints that are locally rational in an economic sense, but collectively they may be suboptimal.30 From this view, institutional theories are quite different from the rational economic perspective, which emphasises individual self-interest, conscious decision making, and economic optimisation. Although the institutional theory has been primarily applied at the organisational level,30–32 it is nonetheless applicable at the individual level. Scott24 pointed out that institutions can operate at the level of ‘localized interpersonal relationships’ (p. 48). Hughes33 also argued that ‘institutions exist in the integrated and standardized behavior of individuals’ (p. 319). In fact, early institutional theory and analyses, especially in economics areas, were at the individual level. For example, Veblen34 argued that ‘not only is the individual’s conduct hedged about and directed by his habitual relations to his fellows in the group, but these relations,

274

Journal of Financial Services Marketing

Vol. 12, 4 272–286

© 2008 Palgrave Macmillan Ltd 1363-0539 $30.00

The adoption of internet banking

being of an institutional character, vary as the institutional scheme varies’ (p. 629). Based on this argument, his later study35 defined the institutions as ‘settled habits of thought common to the generality of man’ (p. 239). Similarly, Cooley36 argued that ‘the individual is always cause as well as effect of the institution’ and ‘in the individual the institutions exist as habit of mind and of action’ (p. 314). Conceptual model and hypotheses Drawing upon the social contagion theory and institutional theory, we propose a conceptual model (see Figure 1) to investigate how institutional forces influence the acceptance of IB by individuals. In the proposed model, we postulate that three institutional forces influence both the attitudes and the intention of adopting IB. Attitude and intention Attitude and adoption intention are two widely examined variables in the literature of technology acceptance. Attitude is defined as the individual’s feelings about performing a behaviour. It is determined through assessing beliefs towards the consequences arising from a behaviour and the desirability of these consequences.16 It is a common supposition that individuals’ social behaviours are motivated by their behavioural attitudes. For example, TAM postulates the attitudinal explanations of intention to use a specific technology or service.17 In addition, the TRA and TPB models also posit that attitude (together with subjective norm and perceived behavioural control), is an essential determinant of adoption intention.15,16 Empirically, the literature has illustrated that attitude has a significant influence on the
Institutional Forces • Coercive • Normative • Mimetic Figure 1 Conceptual model

intention to adopt IB.1,2,7 Therefore, we postulate: H1: Customers’ attitudes positively influence their intention to adopt IB.

Institutions In TAM, perceived usefulness and perceived ease of use are determinants of attitude, while no determinant is incorporated to explain attitude in the original TRA and TPB. Shih and Fang 1 extended TRA by proposing information quality, transaction speed, userfriendliness, and security as determinants of attitude in the IB context. However, as discussed already, those proposed determinants of attitude are technical and personal level factors. In the present study, we postulate institutional forces as determinants of attitude. The institutional theory sheds light onto the importance of institutional environments to attitudes and behaviours of social actors. The theory argues that in modern societies, social actors are typified as systems of rationally ordered rules and behaviours.37,40 Therefore there are widespread social conceptions of appropriate structures, attitudes and behaviours. The institutional theory posits that social actors face pressures to conform to these shared notions of appropriate structures, attitudes, and behaviours, which suggests that social actors are subject to pressures to be compatible with the shared notions.38 DiMaggio and Powell30 and Scott 24 suggested three institutional pressures — coercive, normative and mimetic. These three elements move ‘from the conscious to the unconscious, from the legally enforced to the taken for granted’ (p. 36).39

Attitude

Intention

© 2008 Palgrave Macmillan Ltd 1363-0539 $30.00 Vol. 12, 4 272–286

Journal of Financial Services Marketing

275

Shi, Shambare and Wang

Coercive pressures are defined as formal or informal pressures exerted on social actors to adopt the same attitudes, behaviours and practices, because they feel pressured to do so by more powerful actors.30 Previous empirical evidence suggests that, at the organisational level, coercive pressures may stem from a variety of sources, like resourcedominant organisations (eg suppliers, customers), regulatory bodies and parent corporation.40 In general, there are two types of coercive pressures — competition and regulation. Competitive pressure arises from the threat of losing competitive advantage. Regulatory pressure may rise from government agencies and professional regulatory bodies.29 The literature has cited the influence of coercive pressures on innovation acceptance. For example, Wang and Cheung41 found that coercive pressure was positively related to travel agencies’ adoption of e-business. Zhu et al.42 indicated that the regulatory environment plays an important role in e-business diffusion. At the individual level, it seems unlikely that there are coercive pressures from regulatory bodies and resource-dominant organisations identified at the organisational level. However, at the individual level, IB individuals may still face coercive pressures from other sources. For example, to use such financial services such as mortgage, loan and retirement account management, banks may require users to embrace IB. Employees may be required to use IB for payroll checks. Therefore, we hypothesise: H2: Individuals experiencing higher coercive pressure are more likely to use IB.

actors are then compelled to adopt certain behaviour due to their desire for legitimacy and not necessarily for suitability.43 However, this imitation or copying is not coerced by any powerful actors, neither is it conscious. On the contrary, attitudes, behaviours and practices demonstrated for a long time by most actors in the same social context become so legitimised as the ‘right’ way things are done that individuals often come to believe that these practices and behaviours represent the ‘only’ way to do things.29 The normative pressures may lead social actors who have not adopted the innovation to experience dissonance and hence discomfort when peers whose approval they value have adopted the innovation.30,44 Previous studies have demonstrated this imitation. For example, Granovetter45and Krassa46 suggest that decisions to engage in a particular behaviour depend on the sheer number of similar others in the environment that have already done likewise. In the context of IB, normative pressures suggest that individuals will be more likely to adopt IB if they perceive that a large number of other individuals in their social network have already adopted IB. In many cases, individuals may be afraid that they will be deemed ‘old fashioned’ or ‘laid back’ if they do not follow the current trend. These phenomena have been described before as Bandwagon theories.47 Therefore, we hypothesise: H 3: Individuals experiencing higher normative pressure are more likely to use IB.

Normative pressures occur when social actors voluntarily, but unconsciously, copy other actors’ same attitudes, behaviours and practices. The institutional theory suggests that social actors are more likely to imitate a certain action if that action has been taken by a large number of other actors. Social

Mimetic pressures force social actors to behave by seeking examples of established practices and behaviours to follow through voluntarily and consciously copying the same practices and behaviours of other successful and high-status actors.30 They mimic the course of action of such successful and high-status actors due to the belief that actions taken by successful actors will be more likely to yield positive

276

Journal of Financial Services Marketing

Vol. 12, 4 272–286

© 2008 Palgrave Macmillan Ltd 1363-0539 $30.00

The adoption of internet banking

outcomes. In addition, through imitating the actions of other successful and high-status actors, actors can economise on search costs and minimise experimentation costs, and avoid risks inherent from being the first-movers.40 In the IB context, individuals may selectively imitate the attitudes and behaviours that have been adopted by higher-status individuals. Individuals may believe that they can acquire higher status or associate with higher status individuals by imitating attitudes and behaviours taken by elite groups and high-status individuals. Therefore, it suggests that individuals will be more likely to adopt IB if their social ties who they perceive as high-status have already adopted IB. Therefore, we hypothesise: H4: Individuals experiencing higher mimetic pressure are more likely to use IB.

with higher incomes are more likely to adopt IB. METHODOLOGY Measures To enhance the content validity of the measurements, the questionnaire was designed based on an intensive review of the literature. The questionnaire was then reviewed by academic researchers with expertise in survey methodology, IB, and innovation adoption. Wherever possible, existing measures that had been used in previous studies were adopted. The questionnaire was also pilot-tested using bank customers who had IB experience. The feedback from the pilot test was used to improve the readability and the quality of the questions in the instrument. The questionnaire is shown in the appendix, and the design for the measurement items is described below. The measures for the institutional forces (ie normative, coercive and mimetic) were adapted from Liang et al.32 and Teo et al.40 Specifically, for normative forces, the respondents were asked to indicate the degree to which others in their social network were using IB. For coercive force, the respondents were asked the degree to which IB was required for their financial transactions. For mimetic force, the respondents were asked the degree to which they agreed that individuals who were using IB had a high status. In all these measures, a Likert-type scale of 1–7 was used. A score of 1 indicates ‘strongly disagree’ with the statement, whereas 7 indicates ‘strongly agree’ with the statement. The scale with three items for adoption intention of IB was adapted from Cheng et al.,7 Venkatesh and Morris48 and Hernandez and Mazzon.13 The scale with four items for attitudes was adapted from Cheng et al.7 and Venkatesh et al.,23 Similarly, a Likert-type scale of 1–7 was used for these measures.

Controls Demographic variables may have the potential to influence IB adoption, especially when social factors are considered. We include gender, age and income in our research model as control variables. The literature suggests that gender and age are important variables when examining social factors. For example, women tend to be more sensitive to others’ opinions and thus more salient to social influences.23,48 Similarly, older individuals are more likely to be salient to social influences.49 In addition, wealthy individuals appear more likely to use IB.50 Individuals with high income may subscribe to more banking services, thus the use of IB may bring more benefits to those individuals. Therefore, this may suggest that they may be more inclined to adopt IB. Tan and Teo12 found that higher banking needs lead to higher IB adoption. Roger20 suggested that IB is more likely to be adopted if the innovation meets a felt need. Therefore, we speculate that individuals

© 2008 Palgrave Macmillan Ltd 1363-0539 $30.00 Vol. 12, 4 272–286

Journal of Financial Services Marketing

277

Shi, Shambare and Wang

The measures for controls of age, income and gender were adapted from Cheng et al.7 and Tan and Teo.12 The respondents were asked to identify the pre-defined groupings of their age, income and gender. Data collection The participants used in the study were nontraditional students who were all full-time employees with several years of working experience. Following Chau and Lai,2 we consulted several experienced bankers who were in charge of the IB services. Similar to Chau and Lai,2 these banking professionals indicated that their target customers of IB services would be college-educated, computer-literate, 20–50-year-olds and employed individuals. This segment of target customers is also consistent with the previous studies.50,51 Therefore, our sampling of nontraditional students was considered appropriate. A total of 152 questionnaires were distributed during class. Individuals were informed that the participation in the study was voluntary and their responses would be kept confidential and would be analysed only at the aggregate level. A total of 145 questionnaires were returned. Twenty-one of them were partially completed and consequently excluded from the data analysis, resulting in a total of 124 usable responses (82 per cent). The respondent profile is presented in Table 1. Among the 124 respondents, 92 per cent respondents were using various IB services, and 8 per cent were not using any IB service. Regarding gender and age, 41 per cent were male, 59 per cent female, 39.1 per cent were under 24 years old and 57.6 per cent were between 25 and 50 years old. As for income, most of them had annual income of less than $70,000, among which 32.6 per cent were less than $20,000, 19.6 per cent were between $20,000 and $30,000, and 35.9 per cent were between $30,000 and $70,000.

Table 1

Respondent profile Per cent

Gender Male Female Age Under 24 years 25–34 years 35–49 years 50 or over years Income Less than $20,000 $20,000~$29,999 $30,000~$49,9999 $50,000~$69,999 $70,000~$89,999 $90,000 or more Missing

41 59 39.1 31.5 26.1 3.3 32.6 19.6 25 10.9 3.3 3.3 5.4

DATA ANALYSES AND RESULTS We used partial least squares (PLS) to analyse our data. Unlike LISREL-type structural equation modelling (SEM), which is based on the covariance structure of the latent variables, PLS is a component-based approach,52 thus can avoid two problems: inadmissible solutions and factor indeterminacy.53 PLS examines the significance of the relationships between research constructs and the predictive power of the dependent variable54; thus it is better suited for explaining complex relationships and building theories, as Wold55 argued ‘In large, complex models with latent variables PLS is virtually without competition’ (p. 590). In addition, PLS places minimal restrictions on the sample size and residual distributions.56 PLS analyses simultaneously a measurement model describing the relationships between a research construct and the indicators used to measure the construct (ie factor loadings) and a structural model depicting the relationships between research constructs (ie path coefficients).53 Measurement model We display descriptive statistics for the measurement items in Table 2. Unlike

278

Journal of Financial Services Marketing

Vol. 12, 4 272–286

© 2008 Palgrave Macmillan Ltd 1363-0539 $30.00

The adoption of internet banking

Table 2 The measurement model Construct Normative Item NF1 NF2 NF3 CF1 CF2 CF3 MF1 MF2 MF3 BE1 BE2 BE3 BE4 IN1 IN2 IN3 AGE GEN INC Loading 0.7025 0.9551 0.8938 0.9400 0.8067 0.6470 0.9607 0.9603 0.9097 0.9468 0.9109 0.9660 0.9603 0.9702 0.9599 0.9768 1.00 1.00 1.00 Std. Error 0.1603 0.1063 0.1414 0.0340 0.1022 0.1860 0.1294 0.1200 0.1427 0.0126 0.0328 0.010 0.0076 0.0075 0.0142 0.0105 0.00 0.00 0.00 t-Statistic 4.3835 8.9866 6.3209 27.6813 7.8914 3.4793 7.4256 8.0058 6.3733 75.2334 27.7457 94.4310 126.4765 128.9980 67.5187 93.3523 0.00 0.00 0.00 Composite reliability 0.891 AVE 0.735

Coercive

0.845

0.651

Mimetic

0.961

0.891

Attitude

0.972

0.895

Intention

0.979

0.939

Age Gender Income

NA NA NA

NA NA NA

Note: All t-statistics are significant at the 0.001 level; AVE=Average Variance Extracted; NA: not applicable to single-item measures

LISREL-type SEM which provides global good-of-fitness indices, PLS provides t-statistics for factor loadings. As shown in Table 2, all of the t-statistics of the factor loadings were significant at the p < 0.01 level. Based on the results of the measurement model, we analysed the convergent validity, discriminant validity and reliability of all of the multiple-item scales, following the guidelines in the literature.57 We assessed reliability in terms of item reliability and composite reliability. Item reliability was examined by means of factor loadings of the items of the construct. It is widely accepted that items with loadings of 0.7 or more have adequate item reliability. Table 2 shows that except for item 3 of the construct coercive forces (CF3), all other factor loadings are higher than 0.7, indicating acceptable item reliability. We kept CF3 in the model due to its closeness (0.6470) to the threshold value of 0.7 and its content validity. Construct composite reliability is similar to and superior to Cronbach’s alpha because it considers the

actual factor loadings instead of assuming an equal weight for each item.57 The composite reliabilities in our measurement model ranged from 0.845 to 0.979, which are all above the recommended value of 0.70,58 suggesting adequate construct reliability. We assessed convergent validity in terms of average variance extracted (AVE), which explains the variance that is measured by the construct in relation to the measurement error. Convergent validity requires an AVE of no less than 0.50.57 Table 2 shows that all AVE values were above the recommended value of 0.50 (ranging from 0.651 to 0.939), thus demonstrating adequate convergent validity. Discriminant validity was assessed by comparing the AVE of each individual construct with the shared variances between this individual construct and all of the other constructs. A higher AVE than shared variance for an individual construct suggests discriminant validity.57 Table 3 shows the inter-construct correlations off the diagonal

© 2008 Palgrave Macmillan Ltd 1363-0539 $30.00 Vol. 12, 4 272–286

Journal of Financial Services Marketing

279

Shi, Shambare and Wang

Table 3

Correlations of latent variables X1 X2 0.807 0.268 − 0.111 0.037 0.015 0.392 0.343 X3 X4 X5 X6 X7 X8

Mimetic (X1) Coercive (X2) Normative (X3) Age (X4) Income (X5) Gender (X6) Attitude (X7) Intention (X8)

0.944 0.386 0.282 − 0.116 0.072 − 0.137 0.096 0.077

0.857 − 0.036 0.055 0.115 0.284 0.339

NA 0.414 0.142 − 0.101 − 0.094

NA − 0.081 0.102 0.120

NA 0.032 0.059

0.946 0.825

0.969

Note: Square root of Average Variance Extracted (AVE) is shown on the diagonal of the matrix; Inter-construct correlation is shown off the diagonal; NA: not applicable to single-item measures

of the matrix. A comparison of all of the correlations and square roots of the AVEs on the diagonal indicated adequate discriminant validity. Structural modelling The structural model was assessed by estimating the path coefficients and the R2 values. Path coefficients indicate the strength of the relationships between the independent variables and dependent variable. R2 values indicate the amount of variance explained by the exogenous variables and measure the predictive power of the structural models.59 We calculated path coefficients and t-statistics for hypothesised relationships using a bootstrapping technique. Results of hypothesis testing are presented in Figure 2 and discussed in the following paragraphs. The significant path coefficient from attitude to adoption (b = 0.814, p < 0.01) provided support for H1. As indicated by path coefficients, coercive and normative forces had significant influences on attitude (b = 0.376 and 0.214, respectively; p < 0.01). This result confirmed our theoretical expectation and provided support for H2 and H3. However, as indicated by path coefficient, mimetic forces had no significant impacts on attitude (b = 0.109, NS), suggesting rejection of H4. Regarding controls, the paths from age, income and gender to IB adoption were all insignificant (b = − 0.042, 0.058, 0.044, respectively).

As shown in Figure 2, our model explained 19.8 per cent of the variance in attitude and 68.5 per cent in IB adoption intention. The magnitude of these R2s provides additional evidence in support of the research model. Post hoc analyses As demonstrated in Figure 2, normative and coercive forces relate significantly to attitude, and in turn attitude relates to IB adoption. This causal chain signifies the mediating effect of attitude on the relationship between institutional forces and IB adoption. To test this mediating effect, we followed Baron and Kenny’s procedure60 to examine two more models. The first model (see Figure 3) removes the mediator of attitude and connects all three institutional forces to adoption directly. The second model connects all three institutional forces directly to IB adoption in addition to the mediating links, as shown in Figure 4. Another run of PLS analysis was conducted to test these two models. The model shown in Figure 3 yielded significant links from normative and coercive forces to IB adoption. In addition, the explained variance of IB adoption significantly drops from 0.685 to 0.226. The results of the model shown in Figure 4 demonstrate that the direct effects of normative and coercive forces on IB adoption are insignificant. Meanwhile, the explained variance of IB adoption only

280

Journal of Financial Services Marketing

Vol. 12, 4 272–286

© 2008 Palgrave Macmillan Ltd 1363-0539 $30.00

The adoption of internet banking

Institutional Forces

Coercive
:0 H2

Normative

H3: 0.214*

10

9 ns

.3 76 H
4:

0 -0.
Attitude R2=0.198 H1: 0.814* Adoption R2=0.685

42

ns

Age

*

0.058ns
0.0 44 ns

Income

Gender

Mimetic *p