11.chapter 5

Chapter 5
Data Analysis and Interpretation

Chapter 5: Data Analysis and Interpretation
Chapter preview
Results of this study are presented in eight sections in this chapter. The first section provides
the results of the study used for determining the factors that drive / hinder the usage of the
internet. The second section provides the results of the evaluation of bank websites. The third
section presents the details of an elicitation study conducted using semi structured interviews
with bank senior managers, technology service providers, employees, users and non-users of
internet banking, to understand their views about internet banking and to confirm the
existence of the latent constructs discussed in existing literature. The fourth section presents
data related to internet banking obtained by filing applications under the RTI Act, 2005. The
fifth section presents the results obtained about the perceptions of bank employees towards
internet banking. The sixth section presents the results of the study used to find a relationship
between web traffic and bank financial performance. The seventh section provides details
about the satisfaction of bank customers about internet banking. The last section highlights
the results of the model used to understand internet-banking adoption.
5.1 The state of the internet in India
Internet usage being a prerequisite for internet banking adoption, it was felt that some
attention needs to be given to find the reasons that drives internet use, with special reference
to India. Literature review revealed that government actions such as creating infrastructure,
framing internet related laws, affect internet usage. Governments’ role as a regulator and
support provider emerged as two important factors responsible for internet growth. A
secondary source of data collected by the Internet Society, which covered 10000 internet
users in 20 countries aimed at finding the role of Government regulation and the attitudes
towards internet. As the primary objective of our study was not to study, internet growth, it
was decided to use secondary data in this phase. 535 responses from India were selected and
questions pertaining to the four hypothesized constructs: government support, government
control, attitude and usage were identified and used. The results of the study are presented
here.
136
5.1.1 Demographic profile of the respondents
Females and males constitute 35.3% and 64.7% of the sample. India being a male dominated
society there appears to be a male bias even in the current survey. Only 6.8% of the
respondents were above the age of 50. Majority of the respondents in the sample were either
teenagers or young adults. Table 5.1 shows the summary of the respondents’ gender and age.
Table 5.1: Sample Demographics (phase 1)
Frequency Percentage
1 Gender
Female 189 35.3
Male 346 64.7
Total 535
2 Age
18-21 70 13.1
22-24 76 14.2
25-29 96 17.9
30-34 87 16.3
35-39 58 10.8
40-44 72 13.5
45-49 39 7.3
50-54 12 2.2
55-59 13 2.4
60-64 8 1.5
65+ 4 0.7
5.1.2 Data screening and preparation for analysis
Data screening for out of range values, missing data, outliers, checks for normality and
multicollinearity was done prior to proceeding with statistical analysis.
5.1.2.1 Missing Data

The missing values in the data set were less than 2 percent. A list wise deletion approach was
used.
5.1.2.2 Outliers
Multivariate outliers were detected using Mahalanobis D2. There were 19 outliers with the
probability of D2 less than 0.001. None of these outliers had a Cook’s distance greater than 1.
(Stevens, 1984), reported that not all outliers need to be deleted. They found that only outliers
with Cook’s distance greater than 1 were influential and worthy of further investigation to
examine if they can be deleted. The Mahalanobis D2 and Cook’s distance for all cases are
137
reported in Table D1, (see Appendix D). In this study, all the outliers had a Cook’s distance
less than 1 and therefore none of the outliers were deleted.
5.1.2.3 Normality
Skewness effects, test of means and kurtosis effects, variance and covariance. Non-normality
was checked by inspecting the Skewness and Kurtosis of the univariate distribution and the
Mardias multivariate Kurtosis value. Skewness greater than three and kurtosis greater than
ten are potential problems, (Kline, 2005; West et al., 1995). The skewness and kurtosis values
of all the items in the scale were examined and reported in Table D2, (see Appendix D). The
univariate skewness and kurtosis statistic are below the cut-off for the data in this study.
5.1.2.4 Multicollinearity
The methods used to detect multicollinearity are discussed in chapter 4. The correlation
matrix for the independent variables was calculated and is shown in Table D3, (see Appendix
D). The correlation between the variables does not exceed 0.8, the cut-off prescribed by (Hair
et al., 1998; Cooper & Schindler, 2003; Sekaran, 2006). Each independent variable was
regressed against the other independent variables, the tolerance and VIF was calculated. The
tolerance values were above 0.5 and VIF values were below 2 and are shown in Table D4,
(see Appendix D). The data meets the cut-off prescribed in literature for correlation
coefficients, tolerance and VIF. Therefore, it was reasonable to assume that the data was not
multicollinear.
5.1.3 Exploratory Factor Analysis
An Exploratory Factor Analysis (EFA) was done to determine distinct constructs. EFA
revealed three different factors having Eigen values greater than 1 (as per Kaiser’s criterion)
which accounted for 51.531% of the total variance. The factor loading of each item was
greater than 0.5. The Kaiser-Meyer-Olkin (KMO) measure of sampling adequacy was 0.904,
which is well above the recommended 0.6 or higher, (Sharma, 1996), indicating good
factorability and Bartlett’s test for sphericity was significant. The principal axis factoring
method, with varimax rotation was used. Table 5.2 shows the three distinct factors obtained
after factor analysis.
138
Table 5.2: Rotated Factor Matrix
How much do you agree or disagree with the following Factors

statements 1 2 3
People need to have access to better and cheaper training .621
opportunities.
Governments need to place a higher priority on expanding the .709
internet and its benefits in my country
Local universities and technical institutes need to offer basic and .754
advanced computer and internet technical training.
Tax reductions need to be given to small and medium-sized .672
businesses that are using the internet to conduct business.
Governments should consider ways to provide easier access to .734
cheaper computers.
Governments should consider ways to create or encourage .664
competition amongst internet service providers.
Government control would put limits on the content I can access. .655
Government control would make me fearful that my actions were .626

under surveillance
Government control would limit my freedom of expression .752
Government control would make the internet too controlled .710
Government control would inhibit the growth of the internet .688
Government control would make me use the internet less .670
Freedom of expression is guaranteed on the internet. .535
The internet is essential for my access to knowledge and education .503
The internet does more to help society than it does to hurt it. .615
My life has improved due to using the internet. .669
The following underlying themes were identified as factors.

Factor 1 Government Support
Factor 2 Government Control
Factor 3 Attitude towards Internet Usage
139
5.1.4 Measurement and revised structural model
The measurement model and the revised structural model are illustrated in Figure 5.1. All the
unobserved (latent) variables used were obtained using Exploratory Factor Analysis. The
measurement model shows the interrelationship between the indicators and the unobserved
(latent) variables. Most of the indicator variables have a standardized regression weight either
above or very close to 0.7.Although some indicators did not meet this criteria they were
retained, because most of them were above .5, except us1 and us3, which were retained to
meet the requirements of the minimum three indicators per construct required in structural
equation modelling, (Hair et al., 1998). By convention these weights have to be .7 or higher.
After establishing the model fit and validity of the measurement model, the proposed research
model was tested for the relationship between the latent constructs. On examining the
structural model for significance of the estimated coefficients or paths, it was found that the
paths from Government control to usage had a critical ratio of .485 and p value of .628.
Government support to usage had a critical ratio of -.121 and a p value of .904 and therefore
was not significant.
Figure 5.1: The measurement and revised structural model (phase 1)
140
5.1.4.1 Construct reliability and validity
The reliability and validity of the constructs were established using Cronbach’s alpha,
composite reliability and Average Variance Extracted. AVE calculations are reported in
Appendix E.
Table 5.3: Summary of the reliability and validity measures
Cronbach’s Composite
Construct AVE
alpha Reliability
Government
0.883 .8834 .5586
Support
Government
0.852 .8799 .551
Control
Attitude
0.786 .8005 .50122
AVE was greater than 0.5 for each construct and Composite reliability was greater than 0.7
and this therefore meets the cut-off suggested in (Bagozzi & Yi, 1988, Byrne, 2001; Fornell
& Larcker, 1981). Cronbach's alpha is one of the most popular methods of measuring internal
consistency of the scales. Cronbach’s alpha for the constructs in the study are shown in Table
5.3. A Cronbach’s alpha greater than 0.7 is considered to be a good indicator of internal
reliability in case of exploratory research a Cronbach’s alpha of 0.6 is also acceptable, (Hair
et al., 2006).
The diagonal elements of Table 5.4 are the AVE, the elements below the diagonal are the
correlation of the constructs, and the values above the diagonal are the square of the
correlations. AVE is greater than the squared inter-scale correlation and therefore
discriminant validity was established.
Table 5.4: Correlation among constructs, AVE and Squared Inter-construct
Correlation (SIC)
Construct Government Government Attitude

Support Control
Government
.5586 .1780 .4900
Support
Government
-.422 .551 .2304
Control
Attitude
.700 -.480 .50122
141
A two-step modelling approach was used. The measurement model was tested for model fit
and validity followed by the structural model. (Schumacker & Lomax, 2004; Hair et al.,
2006) strongly recommend the two-step approach for model development.
5.1.4.2 Measurement and revised structural model fit

The results of measurement model and revised structural model fit indices are shown in Table
5.5. The Measurement model showed an acceptable overall model fit.
Table 5.5: Model fit indices for the measurement and structural models
Statistic Measurement Revised Recommended References

model structural value
model
χ2 399.073 399.227 ----------- -----------
degrees of freedom 146 148 ----------- -----------

(df)
ρ 0.000 0.000 >.05 (for sample (Bagozzi & Yi, 1988)

size greater than
400 it will almost
always be
significant)
χ 2 /df 2.733 2.697 <5 (Wheaton et al, 1977)
Root Mean Square 0.013 0.013 < 0.07 (Steiger, 2007)

Error of
Approximation With 90 percent With 90 percent
(RMSEA) confidence confidence
interval (.011, interval (.011,
.014) and .014) and
PCLOSE=1 PCLOSE=1
Normed Fit Index 0.916 0.916 > 0.9 (Bentler & Bonnet,1980)
(NFI)
Comparative Fit 0.944 0.945 > 0.9 (Bentler, 1990)

Index (CFI)
Incremental Fit 0.945 0.945 > 0.9 (Bollen, 1990)

Index (IFI)
Tucker Lewis Index 0.928 0.929 > 0.9 (Sharma et al., 2005;
(TLI) McDonald & Marsh,
1990)
142
5.1.4.3 Structural model maximum likelihood estimates
The Maximum Likelihood (ML) estimates for the structural model are shown in Table 5.6
Table 5.6: Regression weights of the revised model
Path Standardized Unstandardized S.E. C.R. P

Estimates Estimates
Government Attitude .658 .554 .051 10.928 ***
Support
Government Attitude -.200 -.141 .034 -4.192 ***
Control
Attitude Usage .230 .235 .095 2.465 .014
*** ρ < 0.001
It was found that all the path coefficients for the revised model had a p < 0.05 and therefore
were statistically significant. The standardized estimates shown in the Table 5.7 indicate the
strength of the direct paths in the revised model as indicated.
Table 5.7: Standardized direct and total effects
Government Support Government Control Attitude

Direct effects
Attitude .658 -.200 .000
Usage .000 .000 .230
Indirect effects
Attitude .000 .000 .000
Usage .152 -.046 .000
Total effects
Attitude .658 -.200 .000
Usage .152 -.046 .230
Table 5.7 shows the direct, indirect and total effects of the constructs on one another. It was
found that attitude towards the internet had the highest impact on usage of the internet (.230).
Government support had the highest impact on attitude towards the internet (.658).
Government control had a direct negative effect on attitude (-.200). Government support had
an indirect positive effect through attitude on the usage (.152). Government control had an
indirect negative effect through attitude on usage (-.046).
H1. Government support will have a direct positive effect on attitude towards the internet.
The standardized regression weight for this path was found to be .658 and ρ < 0.001.
Therefore, this hypothesis was supported.
H2. Government support will have a direct positive effect on usage of the internet.
As the statistical significance of the path was not established, this path was dropped in the
revised structural model and this hypothesis is not supported.
143
H3. Government control will have a direct negative effect on attitude towards the internet.
The standardized regression weight for this path was found to be -.200 and ρ <
0.001.Therefore, this hypothesis was supported.
H4. Government control will have a direct negative effect on usage of the internet.
As the statistical significance of the path was not established, this path was dropped in the
revised structural model and this hypothesis was not supported.
H5. Government support will have an indirect positive effect on usage of the internet through
attitude towards the internet.
The standardized regression weight for this path was found to be .658 x .230 =.152 and the
paths are statistically significant. Therefore, this hypothesis was supported.
H6. Government control will have an indirect negative effect on usage of the internet through
attitude towards the internet.
The standardized regression weight for this path was found to be -.200 x .230 = -.046, and the
paths are statistically significant. Therefore, this hypothesis was supported.
H7. Attitude towards the internet will have a direct positive effect on usage of the internet
The standardized regression weight for this path was found to be .230 and ρ = 0.014.
Therefore, this hypothesis was supported.
5.1.5 Findings
Results indicate that Government support had a positive impact on the attitude towards the
internet and an indirect effect on internet usage. Whereas, Government control of the internet
negatively affects attitude towards the internet, which indirectly affects usage, albeit in a
weak sense. It was proposed that there would be a direct effect of government support and
government control over internet usage, but empirical evidence indicated otherwise. This
indicates that government actions do not directly create abhorrence towards the internet.
5.2 Evaluation of Internet Banking Sites in India Based on the

Functionality Dimension
The purpose of this study is to evaluate the internet banking websites of public, private and
foreign banks operating in India. A model proposed by (Diniz et al., 2005) to evaluate the
websites from the user’s viewpoint based on the functionality dimension is used in this study.
The internet banking websites of 26 public, 20 private, and 6 foreign banks in India were
144
investigated by manually accessing them. The parameters used for evaluation were
multilingual support, dissemination, transactional and relational dimension.
5.2.1 Comparison of bank websites based on multilingual support provisions
Table 5.8 illustrates the languages supported by the banks websites. After viewing the
websites of public sector, private sector and foreign banks, it is found that most of the public
sector banks have websites in English and Hindi, whereas private and foreign banks have
only English language websites. However, there were a few exceptions such as Yes Bank,
ING Vysya bank and Barclays bank. (Nantel & Glaser, 2008) found that the perceived
usability of the website increases if the website was conceived in the native language of the
user. (Hillier, 2003) show that a relationship exists between language, culture context and
usability. It has been argued that since most of the content on the internet is in English, it
cannot be used by people who do not understand this language, (Keniston, 1997; Wei &
Kolko, 2005; Roycroft & Anantho, 2003; Ono & Zavodny, 2008). (Andrés et al., 2007) argue
that if the language barrier is removed by making the content available in local languages the
adoption rate of the internet will increase. Therefore, it is reasonable to argue that language
creates a barrier to the use of internet banking in India, as most of the banks websites do not
support local languages.
In India, the official figure shows that there are 22 languages and many dialects (Constitution
of India, Eighth Schedule, Article 344 (1) and 351). The language diversity leads to
difficulties in providing content in all languages. Machine translation of English text to
different Indian languages can be one solution to overcome this problem. The Machine
translation method also has inherent drawbacks, as not all words in a language have
equivalent words in another, including other ambiguities. (Recabarren et al., 2008) state that
the study and comprehension of usability of a website must be extended beyond the level of
national cultures to subcultures that live amongst them, as various traits across inhabitants of
a country can be similar, but there are differences in language, experiences and behaviour.
The websites of public, private and foreign banks were accessed during the first week of
March 2009 and the findings are tabulated in Table 5.8
145
Table 5.8: Languages supported by the bank websites
Name of the Bank Languages supported by Name of the Bank Languages supported by the website Name of the Bank Languages supported
(Private sector) (Foreign banks) by the website
(Public sector) the website
Andhra Bank English, Hindi, Telugu Axis Bank English, Hindi RBS English
Allahabad Bank English, Hindi Catholic Syrian Bank English Barclays Bank English, Hindi
Bank of Baroda English, Hindi City Union Bank English, Hindi Citi Bank English
Bank of India English, Hindi, Marathi Development Credit Bank English Deutsche Bank English
Bank of Maharashtra English, Hindi, Marathi Dhanalakhsmi Bank English HSBC English
Canara Bank English, Hindi, Kannada Federal Bank English Standard Chartered Bank English
Central Bank of India English, Hindi HDFC English
Corporation Bank English, Hindi, Kannada ICICI Bank English, Hindi
Dena Bank English, Hindi IndusInd Bank English
Indian Bank English, Hindi ING Vysya Bank English, Hindi, Kannada, Telugu
Indian Overseas Bank English, Hindi Jammu and Kashmir Bank English
IDBI Bank English, Hindi Karnataka Bank English, Kannada
Oriental Bank of Commerce English, Hindi Karur Vysya Bank English
Punjab & Sind Bank English, Hindi Kotak Mahindra English
Punjab National Bank English, Hindi Lakshmi Vilas Bank English
State Bank of India English, Hindi South Indian Bank English
State Bank of Bikaner & Jaipur English, Hindi Tamilnad Mercantile Bank English
State Bank of Patiala English, Hindi Yes Bank English, Hindi, Bengali, Marathi, Punjabi
State Bank of Travancore English, Hindi, Malayalam
State Bank of Hyderabad English, Hindi
State Bank of Mysore English, Hindi, Kannada
Syndicate Bank English, Hindi, Kannada
Union Bank of India English, Hindi
UCO Bank English, Hindi
United Bank of India English, Hindi
Vijaya Bank English, Hindi
146
5.2.2 Comparison of bank websites based on dissemination, transactional and
relational dimension
The websites were investigated and the services and products that were offered were
categorized into basic, intermediate and advanced. A score of one was assigned for the
presence of the feature and zero for the absence of the feature. A maximum score of 8 for
dissemination, 9 for transaction and 9 for relationship was possible with such an assignment.
Based on these scores, the most well developed website will have a maximum score of 26.
The results of the evaluation are presented below in Table 5.9, Table 5.10 and Table 5.11.
Table 5.9: Public sector bank website evaluation scores
Dissemination Transaction Relationship Overall
NAME OF THE Bas Int Adv Bas Int Adv. Bas Int Adv Dis. Trans Rel Overall
BANK
Andhra Bank 3 2 2 3 4 1 2 1 1 7 8 4 19
Allahabad Bank 3 3 2 1 2 1 3 3 0 8 4 6 18
Bank of Baroda 3 3 1 3 4 2 3 3 1 7 9 7 23
Bank of India 3 3 2 3 3 3 2 3 2 8 9 7 24
Bank of Maharashtra 2 2 1 2 2 1 3 3 0 5 5 6 16
Canara Bank 2 3 2 3 4 2 2 3 0 7 9 5 21
Central Bank of India 2 3 2 3 4 2 3 2 0 7 9 5 21
Corporation Bank 2 3 2 3 4 2 2 3 1 7 9 6 22
Dena Bank 3 3 2 1 4 1 3 3 0 8 6 6 20
Indian Bank 3 3 2 1 2 2 2 3 1 8 5 6 19
Indian Overseas Bank 2 3 2 1 3 1 3 2 0 7 5 5 17
IDBI Bank 2 2 2 2 3 1 3 3 2 6 6 8 20
Oriental Bank of 3 3 2 1 4 2 3 3 2 8 7 8 23
Commerce
Punjab & Sind Bank 2 2 2 0 3 1 3 2 0 6 4 5 15
Punjab National Bank 3 3 2 3 4 2 3 2 3 8 9 8 25
State Bank of India 3 1 1 3 3 2 3 2 3 5 8 8 21
State Bank of Bikaner 1 2 1 3 2 1 3 1 1 4 6 5 15

& Jaipur
State Bank of Indore 1 2 1 2 2 2 3 1 1 4 6 5 15
State Bank of Patiala 0 2 1 2 2 2 3 1 1 3 6 5 14
147
Table 5.9 (Continued)
State Bank of 1 2 2 2 2 2 3 2 1 5 6 6 17
Travancore
State Bank of 1 2 2 2 2 1 3 1 1 5 5 5 15
Hyderabad
State Bank of Mysore 0 2 1 2 2 2 3 1 1 3 6 5 14
Syndicate Bank 2 3 2 3 4 1 3 3 1 7 8 7 22
Union Bank of India 2 3 2 3 4 2 3 3 1 7 9 7 23
UCO Bank 3 3 2 3 3 1 3 2 0 8 7 5 20
United Bank of India 2 3 0 1 3 1 3 2 0 5 5 5 15
Vijaya Bank 2 3 2 3 4 1 3 2 2 7 8 7 22
Table 5.10: Private sector bank website evaluation scores
NAME OF THE Bas Int Adv. Bas Int Adv. Bas Int Adv Dis Tran Rel Overall
BANK . s
Axis Bank 3 2 2 3 3 2 3 2 2 7 8 7 22
Bank of Rajasthan 3 3 2 1 2 1 3 2 1 8 4 6 18
Catholic Syrian Bank 2 3 1 1 1 0 3 2 0 6 2 5 13
City Union Bank 2 3 2 1 3 0 3 2 0 7 4 5 16
Development Credit 2 2 2 3 4 1 3 2 0 6 8 5 19
Bank
Dhanalakhsmi Bank 2 3 2 1 2 0 3 1 0 7 3 4 14
Federal Bank 2 3 1 3 4 1 3 2 1 6 8 6 20
HDFC 3 3 2 3 4 2 3 3 2 8 9 8 25
ICICI Bank 3 2 2 3 4 2 3 3 3 7 9 9 25
IndusInd Bank 2 1 2 2 3 2 3 2 2 5 7 7 19
ING Vysya Bank 3 3 2 3 4 1 3 2 0 8 8 5 21
Jammu and Kashmir 2 3 2 3 4 1 2 2 1 7 8 5 20

Bank
Karnataka Bank 2 3 2 2 4 0 2 2 2 7 6 6 19
KarurVysya Bank 2 2 2 3 3 0 3 1 1 6 6 5 17
148
Table 5.10 (Continued)
Kotak Mahindra 3 1 2 2 3 1 3 3 1 6 6 7 19
Lakshmi Vilas Bank 3 3 2 2 2 0 3 2 1 8 4 6 18
South Indian Bank 3 3 2 3 3 0 3 3 2 8 6 8 22
Tamilnad Merchantile 3 3 2 2 3 0 3 2 0 8 5 5 18
Bank
Vijaya Bank 2 3 2 3 4 1 3 2 2 7 8 7 22
Yes Bank 2 2 2 2 4 2 2 1 2 6 8 5 19
Table 5.11: Foreign sector bank website evaluation scores
NAME OF THE
Bas. Int Adv Bas. Int. Adv. Bas Int Adv Dis. Trans Rel Overall
BANK
ABN-Amro Bank 3 1 2 2 4 1 2 1 1 6 7 4 17
Barclays Bank 1 2 2 3 4 0 3 2 0 5 7 5 17
Citi Bank 2 2 2 2 4 0 3 3 1 6 6 7 19
Deutsche Bank 2 2 2 2 3 2 3 2 2 6 7 7 20
HSBC 3 3 1 3 4 1 3 3 0 7 8 6 21
Standard Chartered 3 2 2 3 4 2 3 3 3 7 9 9 25
Bank
The averages of the dissemination, transaction, relationship and overall has been taken from
the tables of the public sector, private sector and foreign banks as shown in Table 5.12.
Table 5.12: Averages of Dissemination, Transaction, Relationship and Overall scores

Public Sector
6.26 6.76 5.96 19
Private Sector
6.9 6.35 6.05 19.3
Foreign Sector
6.16 7.33 6.33 19.83
149
5.2.3 Findings
Results indicate that the banks with high transactional scores also have high overall scores.
Although the individual overall score of the Punjab National Bank, a public sector bank,
HDFC bank and ICICI both private sector banks and Standard Chartered bank a foreign bank
are the highest. The average performance of foreign banks followed by private banks is
higher than the public sector banks. Seven public sector banks have a score of less than 60%.
Punjab & Sind Bank 15,State Bank of Bikaner and Jaipur 15,State Bank of Indore 15,State
Bank of Patiala 14,State Bank of Hyderabad 15, State Bank of Mysore 14 and United Bank of
India 15. Two private sector banks have a score of less than 60%, the Catholic Syrian Bank
13, and Dhanalakshmi Bank 14. In foreign sector, banks there are no banks with a score of
less than 60%. This indicates that there are more laggards in the public sector banks. The
private sector banks and foreign banks were the first movers in adopting technologically
innovative delivery channels, i.e. internet-banking and therefore have an advantage over
public sector banks.
5.3 Interviews with bank senior leaders, technology service providers, bank
employees, users and non-users of internet banking
An elicitation study was conducted to confirm whether the factors found in extant literature
hold even in the Indian context. The perceptions of senior bank management, employees,
users and non-users of internet banking were captured through semi-structured interviews and
questionnaires. Pre-planning of questions prior to the interview resulted in a broad outline of
questions. The pre-determined paper-based interview guide used during these interviews is
appended in Appendix B.
5.3.1 Interviews with bank senior leaders and technology service providers
As it is a known fact that senior managers do not have the time and inclination to fill a survey
questionnaire a more pragmatic approach was taken, a road map of questions were prepared
and posted to them wherever possible before meeting them in person for the interview this
made them comfortable while responding. The interviews were recorded and later the
transcript of these was analysed. The commonly appearing themes were then identified. Table
5.13 illustrates a brief summary of the transcript of interviews conducted as a part of the
elicitation study.
150
Table 5.13: Excerpts from the transcripts of the semi-structured interviews
Details of the Interviewee Q1. What in your opinion is the reason for underutilization of the internet-banking channel?
Shri. O.K. Kaul, “Security and Risks are not the reasons for the low penetration of internet banking. The primary reason for low
Deputy General Manager, penetration of internet banking is the lack of infrastructure in the rural and semi-urban areas. This is also one of
E-Business, Bank of the reasons why private sector and foreign banks having an urban presence have a higher percentage of internet
Baroda,
banking users as compared to public sector banks that have presence in rural areas.”
B.K.C., Bandra, Mumbai
Shri. S.K. Goyal, “In India one cannot expect very high internet banking usage. If the access device for internet is changed to a
Deputy General Manager, mobile phone then there may be a slight increase in the percentage of internet banking users. But the best way
Retail Banking, Bank of would be to make a limited number of banking transactions available on the normal mobile hand set which are
Baroda, B.K.C., Bandra, used by a majority of the people in the country. The objective for us as bankers would be to make Drafts and
Mumbai Cheques be seen only in museums.”
Manager, Retail “ICICI invested about 50 lacs for internet banking solutions during the period 1995 to 1999. The target
Technology Group, ICICI audience of the internet banking services was initially NRIs, but slowly other customers were also included and
Bank the number of internet banking users by the year 2000 almost touched 300,000. The major challenge is to
change the mind set of many a customer who still prefers personal interactions with bank staff at the branch.”
Dhiraj Bhatia “I have worked with three banks in my career Centurion Bank, HDFC, IDBI and now at Kotak Bank. From my
Manager, experience I find that the new generation banks have a slight advantage over the old banks as the customer
Retail Business, segment they cater to are different and internet banking usage figures are higher than old banks. However I feel
Kotak Bank
that there is still scope for banks to fully utilize the potential of the internet banking channel by adopting a
channel maturity strategy where the customers who are beginners can be motivated to perform advanced
operations to ensure stickiness to this channel.”
Deputy General Manager, “The familiarity with new technologies is low in India and this could be one of the reasons that hinder usage of
Retail banking, internet banking which is directly linked to the lack of infrastructure particularly in the rural areas.”
United Bank, Mumbai
151
Head, Marketing, Digital “Among the new channels internet banking seems to be one of the most mature ones with over 60% of all
Channels, SCB. transactions occurring through this route.”
Manager, Digital Business, “We were the first bank in India to launch multi-factor authentication. Our easy to use functionally loaded
Citi Bank website allows almost every possible banking transaction, which differentiates us from other banks. Almost 40%
of our banking customers use internet banking on a regular basis.”
Deputy General Manager, “The branch remains the preferred channel of banking in India due to the human connects and personal
State Bank of India relationship it provides. The problems of commuting to the branch and the other services like bill pay which
internet banking supports will gradually decrease foot falls at the branch.”
Viraj Sawant, “The reason behind not using internet banking is not due to security or perceived risks. If people can use the
CEO, internet for booking rail tickets for the sake of convenience without any fear then why do they not use internet
Idealake banking? It is simply because they enjoy visiting the bank and want the convenience of the employee doing the
(Leading Technology work on their behalf rather than use self-service technology. The moment the banks impose even minor charges
provider to BFSI),
Hotel Lalit, Sahar, Mumbai for using services at a branch most of the customers will switch to the internet channel.”
Vivek Kumar Dwivedi “We see India as a huge potential market for our products. In many branches of the Punjab National Bank, the
Consultant, Technical bank teller has been replaced with our unmanned transaction systems, which validates and accepts cash. In
Support, India, people are used to paying by cash. If one looks at the queues at the automated ticket vending machines
MEI, (leading global
requiring a smart card and the train ticket coupon punching machine, we find that people prefer multiple ticket
manufacturer of
unattended payment punching on coupon vending machines as compared to using technologically superior automatic ticket vending
systems) machines. These observations show that in India people are initially hesitant to use technology. Therefore we
feel that in India only a select segment of people will use internet banking although it may be superior in many
aspects.”
Consultant, Finacle, “Banks need to invest in creating awareness among employees and customers, incentivize early adopters,
Infosys develop methods to evaluate performance and analyse feedback and improve the internet channel experience so
as to increase adoption rates.”
152
Q2. What are your views on non-bank competition particularly with respect to peer to peer lending that
happens over the internet in developed countries?
Shri O.K. Kaul, “Non-Bank competition is not a threat to us because RBI regulations do not permit banking activity without a
Deputy General Manager, license. The websites which are cropping up offering peer to peer lending schemes may be mere money lenders.”
E-Business, Bank of
Baroda, B.K.C., Bandra,
Mumbai
Shri S.K. Goyal, “Non-Bank competition in India is not going to be a threat for many years to come as there is no credit rating
Deputy General Manager, for the entire population and moreover there are instances of a single person holding multiple PAN cards. Once
Retail Banking, the Aadhar card which ensures biometric validation and credit rating for the entire population will be in place
Bank of Baroda, B.K.C.,
then may be such business is possible.”
Bandra, Mumbai
Head, Retail Technology “Non –Bank competition will not be a threat to the retail banking sector particularly because unlike developed
Group, ICICI Bank countries which have peer-to-peer lending schemes promoted on the web, the Indian ecosystem is totally
different. One can borrow from close relatives, friends and borrowing amongst members of a community is quite
prevalent particularly amongst the Kutchi and Marwari communities and most of the times the loan is interest
free. Another problem in the Indian context with peer-to-peer lending, would be recovery if there is a default as
the legal process is not fast track.”
Dhiraj Bhatia “At present due to strict regulations there is no immediate threat from online lenders. But it is essential that
Manager, banks keep an eye on these activities because in the future these small players will eat into the profits of big
Retail Business, banks.”
Kotak Bank
Q3. Does Internet banking give opportunities for service differentiation?
Shri S.K. Goyal, “There is scope for differentiation even though almost all banks have websites by improving the functionality
Deputy General Manager, and making the interface and processes user friendly, it is possible for the banks to help customer’s reduce time
Retail Banking, and lower costs while fulfilling their banking needs.”
Bank of Baroda, B.K.C.,
Bandra, Mumbai
153
Deputy General Manager, “It is possible to differentiate our services by adding new functionalities, quality reporting and making internet
Oriental Bank of banking available on handheld devices.”
Commerce, Mumbai
Manager, Retail “Every bank has offered bank branching, but still they are able to differentiate themselves. In the same way
Technology Group, ICICI although all banks offer internet banking today there is scope for differentiation based on user experience,
Bank transactional capabilities, improved security features and many more.”
Dhiraj Bhatia “I believe that internet banking provides opportunities to tailor make our services to a particular customer
Manager, segment like personal banking or business customers, and can have different interfaces and transactional
Retail Business, capabilities.”
Kotak Bank
Q4. Is the internet banking only an additional service delivery channel or does it have strategic
importance for your bank?
DGM, Indus Ind Bank “The relationship between the bank employee and the customer is going to stay. The anytime, anywhere, access
provided by the internet channel is going to make this relationship more fruitful as the customer finds it easier to
operate their account and improve efficiency.”
AGM, Indian Bank “Our strategy is to give the customers an option to select their banking channel as per their need. The synergy
between these different channels will be of prime importance to be successful in the future.”
Haresh Amre “The internet banking channel is an alternate channel for service delivery, but from a strategic point of view it
AVP and Head, offers ample opportunities for the bank to cross sell their products, expand their geographical reach, co-create
Process Group, products like trading online which can be delivered only via the internet route and not via the branch.”
Infosys Limited
Ms. Veena Jahagirdar, “From a strategic point of view internet banking will enable our bank to handle high transaction volumes
without increasing infrastructure costs. It will also enable bank employees to have more productive engagements
Senior Manager,
with the customers.”
Karnataka Bank
154
Q5.What in your opinion will be the challenges for internet banking service delivery channel in the
future?
Chief Technology Officer, “The functionalities and feature of the website should live up to the expectations of the customers and there
ICICI Direct would be a need to scale up using new technology to meet the increasing volume of transactions.”
Relationship Officer, “As bank employees will not have personal interactions with customers knowing the customer will be the
HSBC greatest challenge as the process is going to become more complex.”
DGM, “We foresee new technologies and applications which will allow customers to access their bank accounts
Central Bank of India, without using a web browser.”
Chandermukhi, Nariman
Point, Mumbai
Viraj Sawant, “The biggest challenge will be to create similar end-user experiences across multiple channels like desktop,
CEO, Idealake mobiles and tablets. The internet banking product offering should be treated as a unified product which is multi
(Leading Technology device compatible rather than developing a disconnect in online banking services.”
provider to BFSI),
Consultant, Capgemini “The banks are operating each channel as silos. In the future, developing analytics and to integrate and
Consulting, Mumbai
streamline information flow across all the departments will be challenging.”
Consultant, BFSI, “In the future, internet bank access in India will be mostly using mobile devices. The banks need to foresee the
TCS, Mumbai future and use systems that their internet-banking website should work across major platforms and mobile
devices. The banks also need to watch the environment for the latest offerings that technology offers like
analytics, Big Data and cloud technology to maintain a competitive advantage.”
Q6. How can your bank mitigate security risks involved in internet banking?
Satyendra Narayan Singh “The media hype around internet banking fraud is the cause of hindrance towards use of internet banking by
DGM, certain segments of customers. Dual authentication seems to be one way of protecting the internet banking
Allahabad Bank user.”
155
Consultant, “Creating customer awareness through campaigns, regulation, technology that addresses the vulnerabilities
TCS involved in internet banking would go a long way in reducing the risks.”
DGM, Karur Vysya Bank “We offer SSL encryption technology, one time pass word for third party fund transfer, on screen keyboard for
protection from key loggers.”
DGM, State Bank of India “Create customer awareness about spyware, phishing and pharming. Advise customers about the risks of using
public computers to access internet banking account, implementing multi-factor authentication mechanisms”
M.M. Tyagi, GM, Andhra “Legal mechanism to quickly penalize fraudsters can act as a deterrent to people who commit cybercrimes. This
Bank will also create confidence among bank customers using internet banking.”
Q7. Does internet-banking usage reduce costs for the bank?
Head, Retail banking, “ When internet banking usage volumes increase it will not only translate into cost savings for the bank but also
HDFC bring in higher volumes of sales as the services of the bank employees can be utilized for cross-selling other
products”
Senior Manager, Retail “The internet channel has facilitated instant approval of car loans and credit cards, thereby increasing sales
banking, AXIS Bank and reducing operational costs.”
Girija Attavar “With internet banking it is possible to service thousands of customers at the same time. The expense towards
General Manager, paper forms, slips and stationery are almost zero and the ability to cross-sell over this channel can help in
Karnataka Bank Ltd. increasing the profit margins of the bank”
156
5.3.1.1 Factors identified after interviews with bank senior managers
The content analysis of these interviews led to the emergence of the following factors:
Cost reduction and increase in sales

Replies from senior leaders of banks show an optimistic view about internet banking with
regard to the cost factor. They feel that not only will there be a cost reduction, but profit
margins will increase as this channel provides ample opportunities for cross-selling other
products and frees bank employees from routine tasks to make them available for more
profitable activities.
Service Differentiation
Based on the replies to the question of differentiation, almost all the interviewees were
equivocal about ample scope for differentiation in spite of all banks offering internet-banking
service to customers.
Risk
There was no agreement in the replies to the question about risk. Many interviewees felt that
security and risks were not the factors that dissuaded potential adopters from staying away
from internet banking, whereas there were some senior managers who felt that the perceived
risk could be one of the factors that were a hindrance to internet banking adoption.
Non-Bank competition
None of the interviewees were worried about non-bank competition arising out of increased
internet usage as they felt that the regulator (RBI) would not permit this activity. Absence of
credit rating for all individuals, strong social network and several other factors were
considered deterrents to non-banks.
The factors for low adoption rates that were identified are: lack of infrastructure particularly
in rural areas, customer mindset, lack of personal contact, and low familiarity with
technology. The factors, which drive internet banking, were: convenience, usefulness, ease of
use, banks initiative, trust, government support, and strategic importance
5.3.2 Interviews with Branch employees
Interviews with bank employees led to the identification of factors, which were of concern to
the employees and are enlisted below
157
Decreased number of employees
One employee of a bank said “Pehle hum logo ko lagaki internet banking logo ko bank teller
se bacha ne keliye banaya gaya hein”. (Initially we felt that internet banking was started with
the intent to save the customers from the bank tellers).
Many of the employees did endorse this view expressed by this employee. The employees
were not initially sensitized and informed about the benefits of the new channels, and that
may be the reason that initially there was a lot of resistance from the employees. But,
progressively the employees have begun to feel that the new channels are just another way of
conveniently handling customers.
One branch manager of a new private sector bank who had previous experience in a public
sector bank described how his job profile changed. As more customers have started using
channels like ATM and internet, banking, the roles of the branch manager has also undergone
a drastic change. Earlier it was a practice to do everything from the branch personal loans,
home loans, credit cards. Now all these activities are handled by customer service executives.
The branch manager now has to go door knocking to get business and close deals. Slowly my
role will become a sales job.
Customers’ alienation
A branch manager of a public sector bank observed that with increasing use of alternate
channels, the customers who regularly visited banks were: small businessmen who needed to
deposit cash and cheques, older retired individuals who come for renewing their fixed
deposits and people belonging to the lower strata of society, who use withdrawal slips for
cash withdrawal, and who are from the least profitable category. The bank staff does not have
face-to-face interactions with majority of the account holders on a regular basis and therefore
find it difficult to cross-sell products like insurance. Another bank employee from the branch
added to these observations of the branch manager, that with customers keeping away from
the bank branch due to other channels such as ATM and the internet, the mode of interaction
with the customers will increase through email.
Queue minimization
“Earlier the bank had hired a consultant to propose solutions to make the process of waiting
in queues as pleasant as possible. The consultant proposed a single line that takes the
customer to multiple service employees. This resulted in all lines moving at the same speed as
158
compared to lines moving at different speeds. Then came the automated queue management
systems, which issued tokens to customers and customers, could relax until their token
numbers were displayed. I feel Internet Banking service and ATMs together can in the long
run do away with queues in the bank and thus reduce the pressure on front desk employees.”
This statement from a bank employee of a new private sector bank shows his concern for
queue minimization at the branch. Many other employees expressed the same sentiments.
This is one reason that the branch employees feel the need to promote internet banking, as
this will rid them of unnecessary tensions, which build up leading to quarrels in the branch.
5.3.3 Interviews with users and non-users of internet banking
When customers at the bank were approached with questionnaire, the response from them
was lukewarm and many of them refused to take part in the study. The reason for refusal to
participate in most cases was that they were in hurry to finish their banking activities and
attend to some important task. The other reasons can be attributed to lack of trust when
approached by a stranger in the bank premises. It was found that instead of administering a
formal questionnaire, customers felt easy when asked questions in an informal way after
explaining to them the intent of the research. The bank customers were told that the research
was for academic purposes only was and this is evident from the openness in their
communication. The questions asked during these interviews are provided in Appendix B.
Non-Users comprised of 75 participants. Table 5.14 provides a summary of the reasons and
the identified constructs based on the interviews with 75 internet-banking non-users. Content
Analysis identified eight factors for not using internet.
159
Table 5.14: Summary of the content analysis of internet banking non-users
Reasons for not using internet banking N=75 N%
Banks support Lack of awareness about how to apply for internet 6

and initiatives banking
(20) Lack of information about cost of internet banking 2 (26.6%)
Lack of support in case of any problems 12
Infrastructure Lack of access to computers 1

(4)
Lack of access to internet 2 (5.33%)
Internet connection is very slow 1
Computer usage Lack of confidence in using computers 6

efficacy (10.66%)
(8) Lack of confidence in using the internet 2
Risk Lack of security 12

(15) (20%)
Lack of privacy 3
Trust (5) Lack of trust 5 (6.66%)
Legal issues Lack of fast and efficient legal mechanism and 5

(5) (6.66%)
support
Incompatibility Deposits and Withdrawals not possible 5

with the need (10.66%)
(8) Multiple Bank accounts leading to difficulty and 3
effort to use Internet banking
Individual Inertia (Resistance to change) 6
characteristics (13.33%)
(10) Lack of personal interaction with the bank staff 4
Besides these factors, other factors that existed in literature were also identified after
interviews with internet banking users and non-users. A few of these factors are presented
below.
Usefulness
“The main reason why I use internet banking is because of the convenience it offers. As part
of my job responsibilities, I need to travel at least for fifteen days a month. With internet
banking I can settle all my utility bills without incurring any late fee charges even when I am
travelling.”
160
This user of internet banking is not only using internet banking for the convenience it offers
but is also using it because his lifestyle does not permit him to use the traditional method of
making payments. Moreover, he feels that there is substantial savings, because if the bills
were paid late he would have to pay additional penalty.
“The main motivation for using internet banking was to avoid queues in the branch.”
This statement from a user of internet banking highlights that time savings, which result due
to the use of internet banking, is another motivation to use internet banking.
Users of internet banking were quick to point out the advantages of internet banking such as
viewing account balance when required, ability to transfer funds across multiple accounts,
pay utility bills, availing Letter of credit, applying for pre-sanction of loans, etc.
No perceived need
“I am a small business man. I need to deposit cheques and cash at the bank on a daily basis,
since I visit the bank almost every morning I do not feel the need for subscribing to internet
banking.”
Many other bank customers endorsed the same view. The customers who own small
businesses visit the bank almost every day either to deposit cheques or cash and for
withdrawals. They felt that internet banking does not satisfy their needs of depositing and
withdrawals and hence did not feel the need for internet banking.
“I have multiple current accounts with the bank. I feel comfortable to visit the bank and get
details about the balance in each account. If I use internet banking it will be inconvenient as
it would require me to remember multiple usernames and passwords and it will take a lot of
time and effort to check each account. If there is an urgent need to know the balance in an
account I usually get it by contacting the bank by phone.”
This statement form a non-user demonstrates that he is content with the service offered at the
branch and felt that using internet banking will not only be time consuming but also complex
for him.
“When I shop online for books on flipkart they have cash on delivery mode of payment option
and there is no need for a credit card or internet banking for making payment. Cash on
delivery assures me peace of mind as the process is simple and does not have inherent risks
of losing money due to fraud or non–receipt of goods after making the payment.”
This statement from a non-user shows that she is risk averse and believes that the process of
paying by credit card or internet banking will be complex.
161
Ease of use
“Unlike branch banking which involves waiting in queue, filling bank slips, counting
currency notes, internet banking is easy to use.”
Users of internet banking felt that using this facility is much easier than going through the
process of traditional banking, which is complex requiring myriad of forms approvals, and
waiting in queues.
Trialability
“If one overcomes the initial inertia then internet banking is really simple.”
This shows that offering internet banking on a trial basis or demonstrating the same at the
branch in which the customers would have hands on experience to use internet banking will
go a long way in overcoming the initial inertia, which acts as an obstacle. There were many
other users who confirmed that the initial procedures to be completed when logging in for the
first time is the cause of hindrance to people who are not proficient in using computers and
the internet.
Facilitating Conditions
“I have access to computers only at the office but they have a hardware firewall which blocks
almost all the sites except ones that are required for official purpose. I have a tight work
schedule and spend most of the time at the office and therefore it is difficult to use internet
banking.”
This comment from a non-user shows that even though he has access to computers and the
internet, the access is limited and does not enable him to use internet banking. The work
place policies on restrictive access to the internet is another problem, which many non-users
felt, was acting as a barrier in internet banking usage. The significance of this comment is
indirect, indicating a need for cheap hand held devices with internet access.
“I have internet access at home but the speed is really slow. I have a fixed free usage
download plan and most of the times it is used for completing my child’s school assignments
and projects.”
Many non-users did not use internet banking because of limited internet access due to cost-
based reasons. There were several customers who did not have easy access to computers.
Privacy
“I have a combination of savings and Demat account with ICICI bank. I was surprised when
a relationship management executive from ICICI direct contacted me saying that my trading
162
account shows very few transactions and he advised me to redeem or switch some of the
mutual funds I had purchased and invest in more lucrative mutual fund schemes on offer. I
did not expect the bank would monitor my transaction history. This is a clear invasion of
privacy. I therefore instructed my bank immediately not to link my bank and trading
account.”
This statement from a user of internet banking indicates the bank’s attempt to cross sell
products by invading on the privacy of the customers is actually dissuading customers from
using internet banking. This also indicates a link between privacy and trust.
“I find that there are a lot of calls from insurance companies claiming to be my banks
associates or direct selling agents particularly in the afternoon when I take a short nap. This
is very annoying. I feel the bank shares my personal details with third parties. However, I do
not know whether my personal information was obtained from the branch or from the
Information Technology department, which handles internet banking. In spite of these issues I
do use internet banking because of the convenience it offers.”
This statement from a housewife who is a user of internet banking highlights the fact that
users have some tolerance when they find the benefits outweigh the potential harm.
Security
“I was a regular user of internet banking but my husband persuaded me to stop using it. He
showed me media reports about the frauds occurring in internet banking transactions.”
This statement highlights the fact that media reports create a negative effect about internet
banking in the customer’s mind.
“I use internet banking just to check my account balance. I have not signed up for
transactional features as I fear that my account will be compromised.”
This statement provides insights into the customers’ fears about security breaches if he
performs transactions. This category of customers use only basic features of internet banking
and as far as the bank is concerned a less profitable segment of users who need to be taken to
the next level.
Trust
“Our traditional process of payments by cheque involves authorized employees to approve,
verify and check before making payments. If we make payments by using internet banking
then it happens just by a click and may lead to financial loss.”
163
This statement from a corporate customer shows his apprehension in using internet banking
to make external payments, as there are no laid down procedures for sanction and approval,
which may lead to financial loss due to mistakes. This shows lack of trust in the system.
“If unauthorized employees come to know the user name and password, it can lead to
financial risk to my business.”
This is also a major apprehension that users had about internet banking. The banks should
educate the customers about the multiple authorization facilities that can be availed for
corporate internet banking accounts. Many people also felt that if problems occur during
transactions they do not trust the bank to back them in resolving the same.
Cost
“I feel that initially the banks may offer internet banking facility and then when the number of
customers using this service increases they may start levying additional charges for usage.”
The relevance of this statement by a non-user of internet banking is twofold. On one hand,
the customer feels that the additional cost will be levied on the customer once a critical mass
of people begin using internet, and on the other hand the customer shows lack of trust in the
bank as he may have had some prior bad experience.. Many customers voiced their concerns
about the bank first selling credit cards saying that it is absolutely free and then charging a
fee later, due to which they are hesitant to use any service that is offered free by the bank.
This clearly shows lack of trust in the bank.
“As internet banking does not attract any charge. The process of payment of bills using
internet banking actually turns out to be cheap as I save on the transport cost to visit the
utility providers’ office.”
These two statements seem to be contradictory. Some customers feel that the bank may
charge for this service in the future, whereas others feel that internet banking is a cost saver
as incidental charges for making bill payments is nil if internet banking is used.
Self-Efficacy
“I use email to communicate with my children who have settled in the United States. But, I
am not very proficient in using the computers for other purposes. I feel that using internet
banking will require a lot of effort and skills which I do not possess.”
This statement by a senior citizen shows lack of confidence in using computers and the
internet for performing banking operations.
164
“I am already an avid user of the internet particularly social media sites. Using internet
banking was very easy for me.”
This statement highlights that internet self-efficacy influences individuals’ to use internet
banking.
Lack of personal interaction with the bank staff

“I retired from the Reserve bank of India. I often visit the bank as this activity keeps me busy
and I also to spend time talking to bank employees, as I have developed some bonding and
friendship with the staff of the bank. If I use internet banking it will increase my loneliness.”
A few customers expressed the same sentiments. They mentioned that internet banking did
not have the human touch.
Subjective Norm
“I started using internet banking when I found that many of my colleagues at the work place
were using it to pay insurance premiums, utility bills and credit card bills. They encouraged
me to use this channel.”
Many users said they started using internet banking due to the direct or indirect influence of
friends, family members and office colleagues. In India, most of the decisions happen by
consensus. The decision usually happens after elders, parents and spouse agree together.
Many non-users particularly those who were young and recently employed stated that their
parents advised them not to use internet banking due to the inherent risks of losing hard
earned money.
Banks initiative
“My colleagues at my work place told me to get the internet banking password and user id
from the branch. They also helped me with the initial login process and in fact guided me
about how to use all the different facilities it supports. I found that the website was not
intuitive and I had difficulty using it. But due to the support available from my colleagues, I
continued using this service. There was absolutely no support to use this channel from the
bank.”
The feeling of the customer was due to the attitude most banks have, that customers will
come to this channel automatically without any efforts or initiative from the bank.
165
“The benefits of using internet banking are not advertised by my bank. There are no
brochures, pamphlets or notices in the branch which give neither information about how to
avail these services, nor any information about how to use the same.”
This response by a non-user of internet banking indicates that banks have not aggressively
marketed this offering. A number of non-users also mentioned that they did not use internet
banking because they thought that it would be a complex process requiring a lot of effort.
This indicates that in addition to marketing there is a need to offer this service on a guided
trial basis at the branch.
“I work as an office boy and most of the times I do not have sufficient balance in my account.
I have not availed of the cheque book facility and withdraw money using withdrawal slip. I
do not know whether the bank will give me internet banking facility. I have internet access in
my office.”
This statement from a non-user of internet banking shows that he is unaware whether he can
avail of this service. This shows that the banks have not been able to create awareness as to
which type of accounts will be eligible for internet banking facility.
Government support
“I feel that if financial losses occur due to frauds, errors or disputes, jurisdiction of the
courts will be an issue and the process can be very time consuming.”
The view of the non-user highlights the fact that without the government providing a proper
legal framework to address the disputes arising from internet based transactions in a timely
manner, many will not use internet banking.
Image
“I use the latest gadgets and am techno savvy. My friends call me a geek. I use internet
banking to keep up with this image.”
There were many such individuals who felt that using internet banking would give them a
higher status.
5.3.4 Findings
Interviews with bank senior management revealed that these individuals felt that the factors
for low adoption rates of internet banking were lack of infrastructure, particularly in rural
areas, customer mindset, lack of personal contact, low familiarity with technology and the
factors, which drive internet banking, were convenience and usefulness.
166
A few positive factors such as queue minimization emerged from the interviews with bank
employees, but the employees were worried about the decrease in employees and customer
alienation, amongst other negative factors resulting due to mass adoption of internet banking.
Several factors, which contribute to adoption of internet banking, and the factors that hinder
adoption were identified after content analysis of the interviews with bank customers. The
most important factor was that the banks have not created awareness about this service and
do not provide adequate support to sort out problems associated with this channel.
Usefulness, No perceived need, Ease of use, Trialability, Facilitating Conditions, Privacy,
Security, Trust, Cost, Self-Efficacy, Subjective Norms, Banks initiative, Government support,
Image and Lack of personal interaction with bank staff were some of the factors that emerged
from interviews with bank customers.
5.4 Investigation about financial implications and operational issues

pertaining to internet banking
The purpose of this investigation was to collect information about actual internet banking
usage, capital investments made towards internet banking, expenditure towards promotion of
internet banking, frauds related to internet banking and growth of internet banking users.
Banks when approached with specific questions to collect information, the information was
not readily available with one department. Some data was with the planning and development
department and some with the department of information technology, both of which were at
different locations. Senior Bank Managers of the public sector banks pointed out that the
required information can be easily obtained by filing applications under the Right To
Information Act; consequently, applications with 13 questions, (see Appendix C) were made
and posted to the Public Information Officers of all public sector banks. The data obtained is
presented below in summarized form in Table 5.15 and findings are discussed.
167
Table 5.15: Summary of the information about internet banking collected from the banks
Bank Total Year of No. of No. of Percentage No. of IB Percentage of Name of No. of
Assets inception Accounts registered IB of logins IB logins per Technology IB
of IB users registered Per day day out of the service frauds
IB users total provider Till date
registered
users
Corp 1,63,560.42 2001 12988467 606049 4.66 54000 8.91 Vayana India 10
Crore Ltd.
UCO 1,91,047 2006 157.66 lacs 1.97 lacs 1.24 13329 6.76 All India 2
Crore Technologies
Ltd.
IDBI 2,73,199 2001 7027537 33 lacs 46.95 78000 2.36 --------- -------
Crore
United 2006 15343125 97874 0.63 4000 4.08 Hewlett 1
Packard
Canara 374160.19 2006 35255332 648001 1.83 48000 7.40 SIFY 0
Crore Technologies
PNB 470013.06 2003 540.68 lacs 20 lacs 3.6 113500 5.67 SIFY 156
Crore Technologies
BOB 447321 Crore 2006 407.7 lacs 43.38 lacs 10.64 90456 2.08 CHIC Infotech 156
Syndicate 193690.50 2003 22108857 701912 3.17 35700 5.08 Oracle FSS 18
Crore
BOI 4,15,966.05 2005 45962500 1979832 4.30 85673 4.32 CHIC -------
Crore Infotech
Dena 87387.92 2008 10299000 137313 1.30 10000 7.28 WIPRO 1
Crore
Union 284595.28 2004 38354080 817005 2.13 70000 8.56 CHIC 3
Crore Infotech
Indian 2005 23444582 535156 2.28 ------- ------- TCS -------
BOM 119464.57 2007 16500000 2.3 lac 1.39 26000 1.13 TCS -------
Crore
Punjab and 78624.36 2010 5711975 12358 0.21 39 0.31 Planet E-Com nil
Sind Crore
OBC 192727 Crore 2006 13900000 5.49 lac 3.9 41500 7.55 Infosys 38
168
From the Table 5.15 the following inferences can be drawn.
5.4.1 Inferences drawn from the data obtained.
5.4.1.1 Internet Banking usage

The number of users who have registered for the internet banking facility is quite low, and in
most cases less than 5%. In the case of IDBI, the figure is around 47%, which is an outlier.
On investigating further, employees of the bank said that in the recent past they were
instructed to give a kit comprising of multicity cheque book, ATM card and self-user creation
of login id and password for internet banking. This may be the reason why the percentage of
registered internet banking users was high. This may have led to the bank customers who did
not need internet banking facility to register as users. When the average numbers of logins on
a day out of the total number of registered users was compared, they were found to be the
lowest among all the banks. The number of registered users is inflated and does not give the
correct indication of internet usage for the banks that automatically register the customers as
internet banking users the moment they open an account. The average number of users
logging into the internet banking website can provide an indication of the usage. It is found
that the percentage of registered internet banking customers logging on to the internet
banking website is dismal, indicating that most of the internet banking registered users are
either dormant or have been given this facility as a matter of procedure by their banks.
5.4.1.2 Capital investments to make internet banking operational

IDBI Bank, Indian Overseas Bank, Bank of India, Dena Bank, Corporation Bank, Canara
Bank, Bank of Baroda, and Syndicate Bank, did not maintain separate records of investment
made for internet banking applications. The investments made in internet banking were a part
of the technology implementation. UCO Bank incurred an expense of approximately Rs. 2.8
crores towards the internet banking channel. United Bank incurred about Rs.2.59 crores as
capital and other investments towards making the internet baking channel operational. Punjab
National Bank invested around 4.70 crores towards securing an enterprise license for internet
banking.
5.4.1.3 Expenditure towards promotion of Internet Banking

IDBI Bank, Indian Overseas Bank, Bank of India, Dena Bank, Corporation Bank, Bank of
Baroda, Syndicate Bank, and Punjab National Bank did not have data on the promotional
expenses incurred for internet banking as these expenses form a part of the business
169
development expenses and no segregation is done for the internet-banking channel. Most of
these banks just mentioned the internet banking channel as an alternative delivery channel in
their advertisements. Canara Bank incurred a sum of Rs. 3,00,785 towards promoting internet
banking during the year 2012.
5.4.1.4 Updation of information on the website

All the banks stated that they update their websites as and when the need arises and mostly on
a daily basis.
5.4.1.5 Frauds related to internet banking

Many banks refused to provide correct figures of the number of frauds, citing that this
information was of commercial confidence and disclosing the same would impede the
process of investigation or apprehension or prosecution of the offenders.
5.4.2 Findings
Based on the data, it is evident that the usage of internet banking is extremely low. The banks
have not paid much attention on promoting internet banking. It was also observed that many
banks were not transparent in reporting the number of frauds related to internet banking.
5.5 Investigation to understand the perception of bank employees’ towards

internet banking
The purpose of the study was to find the attitude of bank branch employees’ towards internet
banking. Extant literature reveals that some employees resist technological changes, whereas
others embrace these innovations. When the primary intent of the bank to facilitate internet
banking was to enhance customer satisfaction by enhancing channel experience, improve
work processes, increase operational efficiency while bringing down costs, the negative
attitude of employees towards internet banking could percolate to the customers with whom
they interact. The study investigated the role of age, gender, education, work experience,
hierarchy in the branch, bank category and bank size on the perception of positive and
negative attitude towards internet banking. The study used an 11-item survey instrument, (see
Appendix B) to investigate bank employee’s perception about internet banking. Factor
analysis on these eleven items revealed the existence of three underlying themes, which were
named as positive factor, negative factor and strategic advantage factor. The items whose
factor loading were more than 0.5 were converted to standard ten (sten) scores and analysed
170
with respect to the dimensions, (viz. employee and bank profile). A paper-based
questionnaire was used, as it was difficult to get email addresses of bank employees and
motivate them to participate in this study electronically. A total of 170 responses were
obtained for analysis. The data analysis and findings resulting from the study are reported
here.
Females and males constitute 52.4% and 47.1% of the sample. Almost 67% of the
respondents were less than the age of 40 indicating that the sample comprised of young or
middle aged respondents. Table 5.16 shows the summary of the respondents’ gender and age.
1 Gender
Female 89 52.4
Male 80 47.1
Total 170
2 Age
20-30 71 41.8
31-40 42 24.7
41-50 25 14.7
>51 32 18.8
multicollinearity was done, prior to proceeding with statistical analysis.

The missing value in the data set was less than 10 percent. Little’s MCAR test resulted in
Chi-Square=104.557, degree of freedom=120 with significance p value=.841. This non-
significant Chi-square indicated that the hypothesis: the missing values are not completely at
random stands rejected. In this data set, the missing values are completely at random. The
regression imputation method was used in this study for missing data imputation
5.5.2.2 Outliers
Multivariate outliers were detected using Mahalanobis D2. Outliers were not found in the data
set. The Mahalanobis D2 and Cook’s distance for all cases are reported in Table D5, (see
Appendix D).
171
5.5.2.3 Normality
Skewness effects test of means, and kurtosis effects variance and covariance. Non-normality
was checked by inspecting the skewness and kurtosis of the univariate distribution and the
Mardias multivariate kurtosis value. Skewness greater than three and kurtosis greater than ten
are potential problems (Kline, 2005; West et al., 1995). The skewness and kurtosis values of
all the items in the scale were examined and reported in Table D6 (see Appendix D). The
univariate skewness and kurtosis statistic are below the cut-off for the data in this study.
The correlation matrix for the independent variables was calculated and is shown in Table
D7, (see Appendix D). The correlation between the variables does not exceed 0.8, the cut-off
prescribed by (Hair et al., 1998; Cooper & Schindler, 2003; Sekaran, 2006). Each
independent variable was regressed against the other independent variables, the tolerance and
VIF was calculated. The tolerance values were above 0.9 and VIF values were below 2 as
shown in Table D8, (see Appendix D). The data meets the cut-off prescribed in the literature
for correlation coefficients, tolerance and VIF. Therefore, it was reasonable to assume that
the data does not suffer from multicollinearity.
5.5.3 Reliability and Validity of the instrument
The instrument was tested for reliability using Cronbach’s alpha indices. Cronbach's alpha is
one of the most popular methods of measuring internal consistency of the scales. In case of
exploratory research Cronbach's alpha greater than 0.7 is considered a good indicator of
internal reliability. A Cronbach's alpha of 0.6 is also acceptable, (Hair et al., 2006). The scale
satisfies the internal consistency requirements, as Cronbach's alpha is 0.646 for the 11-item
scale.
5.5.4 Results
Descriptive statistics of all the 11 items used in the study is as shown in Table 5.17. All items
have above the middle value of the 5-point Likert scale that was used. There were some items
like decrease in the number of employees, increased risk for banks and forced customer
alienation, that were very close to the middle value.
172
Table 5.17: Descriptive statistics of the 11 items used in determining the perceptions of
internet banking
No. Questions Mean Std. Variance

Deviation
1 Create cost reduction for the banks. 4.26 .835 .697
2 Improve the bank’s image. 4.46 .587 .344
3 Reduce queuing in branches 4.41 .796 .633
4 Increased sales 3.91 .952 .907
5 Forced customer alienation (isolation) 3.20 .960 .922
6 Improve customer service and satisfaction. 4.24 .730 .533
7 Decreased number of Employees. 3.02 1.097 1.204
8 Increased competition with non-banks 3.75 .976 .953
9 Give opportunities for service differentiation 3.87 .806 .649
10 Improve market transparency 4.06 .785 .616
11 Increased risks for the banks. 3.21 1.150 1.323
Factor Analysis of the 11 items reveals existence of four factors as shown in Table 5.18.
Table 5.18: The rotated component matrix using Principal Components and varimax
rotation
To what extent do you agree that the adoption
of internet banking will ….. Component
1 2 3
Reduce queuing in branches .725
Create cost reduction for the banks. .699
Improve the bank’s image. .613
Improve customer service and satisfaction. .561
Improve market transparency .556
Increase sales .555
Increase risks for the banks. .698
Force customer alienation (isolation) .675
Decrease number of Employees. .670
Give opportunities for service differentiation .774
Increase competition with non-banks .761
The following underlying themes were identified after the factor analysis
Positive Factors (PF) included questions 1, 2, 3, 4, 6, 10 explained 22.2 percent variance and
included question related to cost reduction, bank’s image, reduced queuing, increase of sales,
customer satisfaction and market transparency.
Negative Factors (NF) included questions 5, 7, 11 explained 16.249 percent of variance and
included questions related to customer isolation, decreased number of employees and
increase in risks for the bank.
173
Strategic Advantage Factors (SAF) included questions 8, 9 explained 13.887 percent of
variance and included questions related to competition with non-banks and service
differentiation.
These three factors explained 52.335 percent of the total variance. Factor scores for all these
factors were in the range 0.555 to 0.774. Factor mean scores were converted to standard ten
scores (sten) using the mathematical formula sten=z*2 + 5.5, and is as shown in Table 5.19.
Table 5.19: Respondents profile & factor mean scores converted to standard ten scores
Respondents Number of Positive Factors Negative Factors Strategic
Profile respondents Advantage
Factors
Age 20-30 5.3692 5.1796 5.4644
71
31-40 5.4346 5.7479 5.7826
42
41-50 5.4701 6.0208 5.2639
25
more than 51 5.7215 5.5069 5.2940
32
Gender Female 5.5327 5.6461 5.4968
89
Male 5.3875 5.3567 5.4753
81
Education SSC (std. X) 5.2336 5.4926 5.2671
5
HSC (std. XII) 6.1118 5.6894 5.2110
4
Bachelors 5.4892 5.4566 5.3903
120
Masters 5.3426 5.5186 5.8614
39
Work less than 9 years 5.4645 5.3015 5.4326
84
Experience
10-20 years 5.2403 5.7644 5.8458
34
more than 20 5.6303 5.5903 5.2767
49
years
Hierarchy Sub staff 5.2837 5.6362 4.6215
6
Clerk 5.4794 5.5803 5.5193
41
Cashier 5.6284 5.8511 4.8465
13
Teller 6.6886 6.3431 5.7203
6
Officer Cashier 5.4424 5.5514 5.0365
20
Assistant 5.2902 5.2160 5.6352
32
Manager
Deputy Manager 5.4367 6.3136 6.1025
7
Manager 5.6048 5.2176 5.9252
22
Senior Manager 5.5302 5.1994 5.4523
13
Chief Manager 5.7606 5.9674 6.4217
5
Bank’s Nationalized 5.6989 5.4575 5.3474
83
Category Bank
Old Private sector 5.7604 5.0027 5.8267
29
New Private 4.7397 5.2888 5.5884
34
sector
Co-operative 5.2792 6.8580 5.3144
23
Bank
Bank size Big 5.6126 5.3072 5.6047
112
Medium 5.3110 6.0884 5.2586
49
Small 4.0793 3.4314 4.2800
2
174
5.5.5 Findings
Results indicate that the attitude about internet banking varies across employee and bank
profile as discussed below:
Age
Employees who were young and in the age group of 20-30 and 31-40 felt that internet
banking offers positive and strategic advantages as these factors outweigh the negative
factors. Employees in the age group 41-50 felt that the negative factors of internet banking to
be more prominent than the others are. An interesting observation is those employees who
were more than 51 years believed that positive factors are more important than other factors.
Gender
Female employees seem to be more concerned about the negative effects of internet banking.
In the case of male employees, the strategic advantages followed by positive factors outweigh
the negative effects of internet banking. The findings are in line with (Sacks et al., 1994; Jehn
et al., 1999; Rose & Straub, 1998; Morahan-Martin, 2000), where the study found that males
are more positive about computers regardless of the familiarity, in contrast to the female
attitude, which becomes positive only as familiarity increases. In India this observations may
also be due to women not getting adequate opportunities to adopt and use new technologies.
Education
Educational qualification plays a major role in the attitude about use of technology.
Employees with a high school, bachelor’s degree had high scores for the positive factors, and
employees with a master’s degree believed that strategic advantage is the most important
factor.
Work experience
Employees who had less than nine years of experience in the banking sector felt that positive
factors followed by strategic advantages to be important. Employees who had experience
between ten to twenty years considered negative effects of internet banking to be more
important than other factors. Employees who had more than twenty years’ experience found
that positive factors outweigh other factors.
Hierarchy
Employees belonging to the managerial level felt that internet banking offers strategic
advantages compared to non-managerial positions. Employees belonging to the lower cadre
were more concerned about the negative factors of internet banking.
175
Bank category
Employees of nationalized banks felt that positive factors outweigh other factors. Employees
of new private sector banks felt that strategic advantages were more crucial than other
factors. Employees of co-operative banks seem to be risk averse and negative factors appears
to overrule other factors for them.
Bank size
The analysis based on a banks size revealed that the fluctuation in factor scores for big banks
is the least as compared with medium and small banks. The factor scores for all the factors in
case of employees of small banks were below the middle level. The employees of big banks
and small banks felt that the positive and strategic advantages far outweigh the negative
factors. The employees of medium sized banks were more concerned about the negative
factors associated with internet banking.
5.6 Relationship between website traffic and financial performance of the

banks
The primary aim of this study was to determine whether there is an association between the
website traffic the bank attracts and the financial performance of the banks. The study
hypothesizes that the higher the website traffic, the better the performance of the bank. The
study proposes and empirically tests six hypothesis using financial performance measures of
25 public sector banks, 6 new private sector banks, 10 old private sector banks and 4 foreign
banks operating in India from their yearly audited results for the year ending March 2011,
and website traffic statistics from Alexa, (a web traffic reporting company). Linear regression
was used to test the hypothesis. Equations (5.1) and (5.2) describe the linear regression.
Table 5.20, Table 5.21, Table 5.22, Table 5.23 and Table 5.24 summarize the results obtained
for linear regression on the dependent and independent variables for each category of the
banks. The hypothesis was tested using linear regression
Performance = β0 + β1 (log of global rank) + ε (5.1)
Performance = β0 + β1 (log of India rank) + ε (5.2)
Where β0 and β1are the regression weights and ε is the error in the approximation
176
The Hypothesis was tested on the following groups:
1. All banks
2. Public sector banks (Including State Bank of India and its subsidiaries)
3. Private sector (new banks)
4. Private sector (old banks)
5. Foreign Banks
Table 5.20: Linear Regression results of all banks
All banks
Independent Dependent R R2 F Sig.
variable variable β0 Sig. β1 Sig.
Log of L_Tot_asset .451 .204 10.996 .002 20.742 .000 -.788 .002
Global rank L_Tot_inc .448 .200 10.772 .002 18.146 .000 -.779 .002
L_opprofit .455 .207 11.207 .002 16.790 .000 -.712 .002
Log of L_Tot_asset .443 .196 10.498 .002 17.795 .000 -.673 .002
India rank L_Tot_inc .439 .193 10.273 .003 15.229 .000 -.665 .003
L_opprofit .449 .201 10.847 .002 14.152 .000 -.612 .002
Table 5.21: Linear Regression results of public sector banks
Public sector banks

Log of L_Tot_asset .390 .152 4.123 .054 16.840 .000 -.426 .054
L_opprofit .459 .210 6.123 .021 13.998 .000 -.447 .021
Log of L_Tot_asset .390 .152 4.122 .054 15.000 .000 -.338 .054
L_opprofit .460 .211 6.169 .021 12.074 .000 -.356 .021
Table 5.22: Linear Regression results of old private sector banks
Old private sector banks

Log of L_Tot_asset .714 .510 8.324 .020 16.557 .000 -.555 .020
L_opprofit .636 .405 5.436 .048 13.792 .001 -.575 .048
Log of L_Tot_asset .628 .395 5.221 .052 13.748 .000 -.392 .052
L_opprofit .560 .314 3.664 .092 10.886 .000 -.407 .092
177
Table 5.23: Linear Regression results of new private sector banks
New private sector banks

Log of L_tot_assets .912 .832 19.819 .011 15.624 .000 -.475 .011
Global rank L_tot_income .936 .877 28.495 .006 11.670 .000 -.449 .006
L_op.profit .892 .795 15.542 .017 11.888 .000 -.396 .017
Log of India L_tot_assets .908 .825 18.862 .012 14.162 .000 -.457 .012
rank L_tot_income .936 .877 28.495 .006 11.670 .000 -.449 .006
L_op.profit .889 .790 15.065 .018 10.673 .000 -.381 .018
Table 5.24: Linear regression results of foreign banks
Foreign banks
Global rank L_tot_income .920 .846 10.947 .080 20.585 .002 -.291 .080
L_op.profit .791 .626 3.343 .209 18.591 .004 -.243 .209
India rank L_tot_income .938 .881 14.753 .062 19.638 .001 -.265 .062
L_op.profit .821 .674 4.141 .179 17.823 .002 -.225 .179
178
Table 5.25: Hypothesis test results of different categories of banks.
Bank Category Hypothesis

All Banks H1a. The banks Global website rank will be negatively associated with the banks total assets. Supported
H1b. The banks Global website rank will be negatively associated with the banks total income. Supported
H1c. The banks Global website rank will be negatively associated with the banks operating profit. Supported
H2a. The banks India website rank will be negatively associated with the banks total assets. Supported
H2b.The banks India website rank will be negatively associated with the banks total income. Supported
H2 c. The banks India website rank will be negatively associated with the banks operating profit. Supported
Public Sector Banks H1a. The banks Global website rank will be negatively associated with the banks total assets. Not Supported
H1b. The banks Global website rank will be negatively associated with the banks total income. Not Supported
H2a. The banks India website rank will be negatively associated with the banks total assets. Not Supported
H2b.The banks India website rank will be negatively associated with the banks total income. Not Supported
H2c. The banks India website rank will be negatively associated with the banks operating profit. Supported
Old Private Sector Banks H1a. The banks Global website rank will be negatively associated with the banks total assets. Supported
H2a. The banks India website rank will be negatively associated with the banks total assets. Not Supported
H2c. The banks India website rank will be negatively associated with the banks operating profit. Not Supported
New Private Sector Banks H1a. The banks Global website rank will be negatively associated with the banks total assets. Supported
H2c. The banks India website rank will be negatively associated with the banks operating profit. Supported
Foreign Banks H1a. The banks Global website rank will be negatively associated with the banks total assets. Not Supported
H1b. The banks Global website rank will be negatively associated with the banks total income. Not Supported
H1c. The banks Global website rank will be negatively associated with the banks operating profit. Not Supported
H2b.The banks India website rank will be negatively associated with the banks total income. Not Supported
H2c. The banks India website rank will be negatively associated with the banks operating profit. Not Supported
179
5.6.1 Findings
Results indicate that the association between the web traffic and the performance of the bank
is partially supported, and there is no conclusive evidence across all categories of banks
indicating an association between the web traffic and the bank performance. The study has
empirical evidence to illustrate that the new private sector banks have utilized the internet-
banking channel optimally and therefore all the plausible hypothesis was supported for this
category of banks.
5.7 Measurement of the internet banking users’ satisfaction
The purpose of the study was to determine the major factors that contribute to the level of
satisfaction of the internet banking users in India. The End User Computing Satisfaction
(EUCS) model was used for this purpose. A survey questionnaire was administered to
internet banking users and 387 responses were collected. A factor analysis on the 12 items
used in the EUCS model with oblique (non-orthogonal) rotation and five fixed factors
revealed the existence of the same latent constructs hypothesized in the original EUCS
Model. Confirmatory Factor Analysis (CFA) was then used to test and validate the four
hypothesized models for model fit.
society there appears to be a male bias even in the current survey. Almost 65% of the
middle aged respondents. Table 5.26 shows the summary of the respondents’ gender and age.
1 Gender
Female 157 40.6
Male 346 59.4
Total 387 100
2 Age
=< 35 84 21.7
35-50 168 43.4
=>50 135 34.9
180
multicollinearity was done prior to proceeding with statistical analysis.

Data from the paper based questionnaire and online forms were combined and saved as a
single file. A univariate statistics (Missing Value Analysis) revealed that the percentage of
missing values for all the variables was less than 10%. Diagnosis of the nature of the missing
data using Little’s MCAR test gave a Chi-Square=185.368, degree of freedom=187 with
significance p value=.520. This non-significant Chi-square indicated that the hypothesis that
the missing values are not completely at random stands rejected. In this dataset, the missing
values are completely at random. This makes it possible to impute the missing data using any
method. The regression imputation method was used in this study for missing data
imputation, as (Byrne, 2001) pointed out that this means the imputation is based on variance
and covariance and may lead to biased standard errors in SEM.
5.7.2.2 Outliers
Multivariate outliers were detected using Mahalanobis D2. There were 11 outliers with a
probability of D2 less than 0.001. None of these outliers had a Cook’s distance greater than
one. (Stevens, 1984) reported that not all outliers need to be deleted. They found that only
outliers with Cook’s distance greater than one were influential and worthy of further
investigation to examine if they can be deleted. The Mahalanobis D2 and Cook’s distance for
all cases are reported in Table D9, (see Appendix D). In this study, all the outliers had a
Cook’s distance less than 1 and therefore none of the outliers were deleted.
5.7.2.3 Normality
Skewness effects test of means and kurtosis effects variance and covariance. Non-Normality
are potential problems, (Kline, 2005; West et al., 1998). The skewness and kurtosis values of
all the items in the scale were examined and reported in Table D10, (see Appendix D). The
univariate skewness and kurtosis statistics are below the cut-off for the data in this study.
181
matrix for the independent variables was calculated and is illustrated in Table D11, (see
Appendix D). The correlation between the variables does not exceed 0.8, the cut-off
prescribed by (Hair et al., 1998; Cooper & Schindler, 2003; Sekaran, 2006). Each
independent variable was regressed against the other independent variables, the tolerance and
VIF was calculated. The tolerance values were above 0.2 and VIF values were below 4 and
are shown in Table D12, (see Appendix D). The data meets the cut-off prescribed in the
literature for correlation coefficients, tolerance and VIF. Therefore, it was reasonable to
assume that there was no multicollinearity in the data.
5.7.3 Factor Analysis
In the current study, it was expected that the items would load on five factors, which were
identified earlier as content, accuracy, format, ease of use and timeliness. The existence of the
well-established theory that supports the contention that these twelve items will lead to five
factors suggests, that there was no need for factor analysis. However, to avoid blind faith in
the instrument, a factor analysis using principal axis factoring with non-orthogonal (oblique)
rotation and forcing the fixed five factors supported the existence of the same latent
constructs.
A factor analysis on the twelve items used in the EUCS model with oblique (non-orthogonal)
rotation with five fixed factors revealed the existence of the same latent constructs as
hypothesized in the original EUCS Model. These five factors accounted for 79% of the total
variance.
The Kaiser-Meyer-Olkin (KMO) measure of sampling adequacy was 0.849 which was well
above the recommended 0.6 or higher (Sharma, 1996) indicating good factorability. Bartlett’s
test for sphericity was significant (sig. =0.000), which indicated that the variables correlated
with each other. Table 5.27 shows the five distinct factors obtained after factor analysis.
Coefficients having absolute value of less than 0.5 were suppressed in the display.
182
Table 5.27: The rotated component matrix using Principal axis factoring and oblique
rotation
Construct Items Code

Name Component
1 2 3 4 5
Does internet banking provide the C1 .739
precise information you need?
Does information content on the C2 .919
Content internet-banking website meet your
needs?
Does the internet-banking website C3 .594
provide reports that meet your need?
Does the internet-banking website C4 .912
provide sufficient information?
Is internet banking accurate? A1 .936
Accuracy
Are you satisfied with the accuracy of A2 .936
internet banking website?
Format Is the information clear from the F1 .933
internet banking website?
Is the website lay out and format of F2 .796
providing good information?
Is internet banking user friendly? EU1 .916
Ease of use
Is internet banking easy to use? EU2 .917
Timeliness Do you get the information you need T1 .898

in time?
Does the internet-banking website T2 .840
provide up-to-date information?
5.7.4 Tests on competing factor structures
In the existing literature, four alternative models for EUCS have been described.
1. EUCS items load on a single factor (A single first-order factor model).

2. EUCS items load on to three uncorrelated or orthogonal first-order factors
3. EUCS items load on to three first-order factors, which are correlated with each other
4. EUCS items load on to three first-order factors, which do not inter-correlate but form a
single second-order factor.
The four models are as shown in Figure 5.2
183
Model: 1 Model: 3
Model: 2 Model: 4
Figure 5.2: The four hypothesized End User Computing Satisfaction models
5.7.5 Model fit
(Marsh & Hocevar, 1985) recommend χ 2/df to be between 2 and 5 for a reasonable fit.
(Bentler & Bonett, 1980) recommend that NFI should be greater than 0.9 for a good fit. CFI
close to 1 indicates a good fit, (Bentler, 1990). IFI values very close to 1 indicate a very good
fit (Bollen, 1990). TLI close to 1 indicates a good fit,(Sharma et al, 2005; McDonald and
Marsh, 1990). RMSEA in the range of 0.05 to 0.1 is considered a fair fit and beyond 0.1, a
184
poor fit, (MacCallum et a1., 1996) Table 5.28 shows the Goodness of Fit Index (GFI) for the
four alternative models.
Table 5.28: Goodness of fit index for the four alternative models
Mode χ 2 (df) χ 2 /df (NFI) (GFI) (AGFI) (RMSR) (CFI) (TLI) (IFI) (RMSEA)
l
1. 1569.382 (54) 29.063 .567 .594 .414 .076 .574 .480 .576 .270
2. 814.436 (58) 14.042 .775 .683 .573 .216 .788 .758 .788 .184
3. 96.356 (44) 2.190 .973 .960 .928 .021 .985 .978 .985 .056
4. 138.684(49) 2.830 .962 .943 .909 .032 .975 .966 .975 .069
2
Chi-square (χ ), degrees of freedom (df), Root Mean Square Error of Approximation (RMSEA), Normed Fit
Index(NFI), Comparative Fit Index(CFI), Incremental Fit Index(IFI), Tucker Lewis Index(TLI), Goodness of Fit
Index (GFI),Adjusted Goodness of Fit Index(AGFI), Root Mean Squared Residual (RMSR)
All four models showed a significant chi-square. In this study, the sample size was 387. The
χ2 is appropriate only when the sample size is between 75 to 200 (Kenny, 2011). The χ2
statistic is not appropriate for large sample size, (MacCallum, 1990; Jöreskog & Sörbom,
1993; Byrne, 2001) and therefore these results are not unexpected. While comparing
competing models, smaller the value of χ 2 fit the model better.
Empirical results indicate both model 1 and model 2 do not fit the data, model 3 had an
excellent fit, and model 4 had a good fit. These findings are consistent with a prior study,
which compared these four competing models, (Doll et al., 1994). Thus, model 3 and model 4
can be considered competing models for the EUCS instrument. (Doll et al., 1994) calculated
the target coefficient, keeping model 3 as the target model. Target coefficient is the ratio of
the chi-square of model 3 to the chi-square of model 4. Target coefficient was used to test the
existence of a higher order satisfaction construct. In this study, the target coefficient was
found to be 0.69, which indicates that 69% of the variation in the five first-order factors in
model 3 is explained by the second-order EUCS construct. Model 4 happens to have an
advantage over model 3 and as extant literature shows substantial evidence of satisfaction as
a single second-order construct. This study uses model 4 for testing the validity and reliability
of the constructs and items.
185
Reliability and validity of the instrument was tested using model 4 as the basis. Table 5.29
shows the Squared Multiple Correlation (SMC), Cronbach’s alpha, Composite Reliability and
AVE for the constructs in the study. The procedure adopted for calculation of AVE is shown
in Appendix E.
Construct R2 Cronbach’s Composite AVE

alpha Reliability
Content .453 .866 .882 .657
Accuracy .479 .936 .936 .881
Format .708 .912 .914 .842
Ease of Use .516 .928 .929 .868
Timeliness .619 .868 .869 .768
5.7.6.1 Convergent Validity
AVE should be greater than 0.5 and Construct Reliability should be greater than 0.7, (Byrne,
2001; Fornell & Larcker, 1981). Construct reliability should be greater than 0.6 and AVE
should be equal to or greater than 0.5, (Bagozzi & Yi, 1998). In this study, Cronbach’s alpha
and composite reliability of all constructs were found to be high, thereby indicating adequate
convergence or internal consistency. AVE of all the constructs was found to be greater than
0.5, which suggests adequate convergent validity.
The validity of the measurement has already been established in a previous study, (Doll et al.,
1994; Pikkarainen et al., 2006). Extant literature also supports the use of these items for
measurement. Table 5.30 shows the standardized loading estimates of all the items in model
4. The factor loading for all the items are more than 0.7, which indicates adequate convergent
validity.
186
Table 5.30: Factor loading of all items used in the study
Path Standardized R2 S.E. C.R. ρ

Estimates / (Reliability)
Factor
Loading
C1 CONTENT .842 .709 .041 21.005 ***
C2 CONTENT .890 .791 .043 22.786 ***
C3 CONTENT .607 .369 .063 12.999 ***
C4 CONTENT .871 .758 Parameter fixed at 1
A1 ACCURACY .924 .854 .040 23.618 ***
A2 ACCURACY .952 .907 Parameter fixed at 1
F1 FORMAT .887 .787 .042 23.520 ***
F2 FORMAT .946 .896 Parameter fixed at 1
EU1 EASEOFUSE .947 .897 .048 23.093 ***
EU2 EASEOFUSE .916 .839 Parameter fixed at 1
T1 TIMELINESS .899 .807 .059 17.852 ***
T2 TIMELINESS .854 .729 Parameter fixed at 1
Standard Error (S.E.), Critical Ratio (C.R.), *** ρ < 0.001
5.7.6.2 Discriminant Validity

In Table 5.31, the diagonal elements are the AVE. The values below the diagonal are the
implied correlation of the constructs and the values above the diagonal are the square of the
correlations between the constructs.
Discriminant validity is established if the AVE from the construct is greater than the variance
shared between the construct and the other constructs in the model (Chin, 1998). (Hair et al.,
2006) state that if the AVE is higher than the squared inter-scale correlation of the constructs,
then discriminant validity is supported. Discriminant validity of the model was established by
using the criteria, that the AVE estimates for the construct should be larger than the square of
the inter-construct correlation estimates. In this study, there were five constructs in all. In this
model, the AVE is greater than the square of the inter-construct correlation in all cases and
hence discriminant validity was established.
187
Table 5.31: Correlation amongst the constructs, AVE and Squared Inter-construct
Correlation (SIC)
Construct Content Accuracy Format Ease of Use Timeliness

Content .6567 .217 .321 .234 .281
Accuracy .466 .8805 .339 .247 .297
Format .567 .582 .8415 .366 .438
Ease of Use .484 .497 .605 .8680 .320
Timeliness .530 .545 .662 .566 .7680
(Hair et al., 2006) state that if the AVE is higher than the squared inter-scale correlation of
the constructs, then discriminant validity is supported.
5.7.7 Results
The estimation of the regression weights of model 4 is illustrated in Table 5.32. Estimates
with Critical Ratios (C.R.) greater than 1.96 are significant at the .05 level, (Garson, 2004).
All the paths were significant and had a ρ< 0.001. The standardized estimates shown in the
table indicate the strength of the direct paths in the revised model as indicated. The regression
weights along the paths in the model provide useful insights as to the importance of each
factor that contributes to satisfaction. In the study it was found that the factor “format” with
factor loading 0.842 was the highest followed by “timeliness” with a factor loading of 0.787
and “ease of use” 0.719.
Table 5.32: The regression weights of the variables in the model
Path Standardized S.E. C.R. ρ

Estimates
Content EUCS .673 .106 10.339 ***
Accuracy EUCS .692 .085 10.986 ***
Format EUCS .842 .089 12.583 ***
Ease of Use EUCS .719 .105 10.951 ***
Timeliness EUCS .787 .106 10.339 ***
Standard Error (S.E.), Critical Ratio (C.R.), *** ρ < 0.001
188
Table 5.33 shows the mean and standard deviation of the items used in the study.
Table 5.33: Mean and Standard Deviation of the items

Construct Items Code Mean Standard
Name Deviation
Does internet banking provide precise C1 3.73 .789
information you need?
Does information content on the internet- C2 3.60 .862
Content banking website meet your needs?
(3.465) Does the internet-banking website provide C3 3.05 1.078
reports that meet your need?
Does the internet-banking website provide C4 3.48 .900
sufficient information?
Is internet banking accurate? A1 4.06 .672
Accuracy
(4.035) Are you satisfied with the accuracy of the A2 4.01 .692
internet-banking website?
Is the information clear from the internet F1 3.86 .726
Format banking website?
(3.895) Is the website lay out and format of providing F2 3.93 .683
information good?
Is internet banking user friendly? EU1 3.61 .896
Ease of Use
(3.66) Is internet banking easy to use? EU2 3.71 .841
Do you get the information you need in time? T1 3.77 .718

Timeliness
(3.81) Does the internet-banking website provide up- T2 3.85 .724
to-date information?
5.7.8 Findings
The mean scores for all the variables were higher than the midpoints of the scale. This shows
that the respondents are satisfied with the internet banking websites. Comparing the mean
scores of all the factors reveals that internet banking users were least satisfied with the
“content” (mean 3.465) of the internet banking websites. The mean score of the construct
“accuracy” (mean 4.035) was found to be highest, which indicates that the banks are
providing accurate information on their websites.
Table 5.34 summarizes the goodness of fit criteria in four different studies (Pikkarainen et al.,
2006; Doll et al., 1994; Abdinnour-Helm et al., 2005; McHaney & Cronan, 1988), where
EUCS was used as a second-order construct. A comparison of all these models provides good
189
insight about the stability of the EUCS model for measuring satisfaction across different
domains and cultures. The goodness of fit measures obtained in this study is consistent with
other studies, indicating that the EUCS instrument can be used for measuring internet
banking users’ satisfaction.
Table 5.34: Comparison of the goodness of fit measure of the second-order EUCS model
across five studies:
This study (Pikkarainen et (Doll et (Abdinnour- (McHaney &

al., 2006) al.,1994) Helm et al., Cronan, 1988)
2005)
Sample Size 387 268 409 176 411
2
χ (df) 138.684(49) 30.09 (22) 185.81 (50) 61.08 (48) 25.74(5)
2
χ /df 2.830 1.367 3.72 1.27 5.15
NFI 0.962 0.90 0.940 Not reported 0.979
CFI 0.975 0.98 Not reported 0.99 Not reported
GFI 0.943 0.97 0.929 Not reported 0.977
AGFI 0.909 Not reported 0.889 Not reported 0.932
SRMR 0.048 Not reported 0.035 Not reported 0.027
RMSEA 0.069 0.04 Not reported 0.04 Not reported
Table 5.35 summarizes the factor loading on the five constructs of the EUCS model across
four different studies. A look at the highest factor loading across the five constructs illustrates
the construct “content” to be the highest followed by the construct “format” in (Doll et al.,
1994; McHaney & Cronan, 1988) while in (Pikkarainen et al., 2006) it was “ ease of use”. In
this study the highest factor loading was on the construct “format” followed by “timeliness”.
(Doll et al., 1994) used the model for measuring satisfaction of computer application
software. (McHaney & Cronan, 1988) used EUCS for satisfaction measurement of the
decision support system based on computer simulation. (Pikkarainen et al., 2006) used it for
satisfaction measurement of online banking users in Finland. These differences in factor
loading are not a measure of the performance of the EUCS model, as the context of this study
is internet banking and the study is restricted to samples from India, It differs from the study
on general computing systems and studies in other countries where the study settings are
different.
190
Table 5.35: Comparison of the factor loading and reliability of the constructs of the
EUCS model, in which satisfaction was a second-order factor dependent on five first-
order factors.
This study (Pikkarainen et al.,2006) (Doll et al.,1994) (McHaney &

Cronan, 1988)
Construct Factor Factor Factor Factor
Loading R2 Loading R2 Loading R2 Loading R2
(C.R) (C.R) (C.R) (C.R)
Content 0.673 .453 0.81 0.66 0.912 0.68 0.950 0.90
(10.33) (8.83) (17.67) (61.40)
Accuracy 0.692 .479 0.71 0.50 0.822 0.73 0.776 0.60
(10.98) (n.r) (16.04) (24.85)
Format 0.842 .708 * * 0.993 0.53 0.808 0.65
(12.58) (18.19) (27.69)
Ease of 0.719 .516 0.73 0.53 0.719 0.68 0.822 0.68
Use (10.95) (n.r) (13.09) (29.21)
Timeliness 0.787 .619 * * 0.883 0.76 0.791 0.63
(10.33) (13.78 (26.18)
Critical Ratio (C.R) are indicated in parentheses , not reported (n.r), * indicates that these constructs were not
used in the model
In the case of internet banking users’ satisfaction in India, empirical evidence indicates that
the construct “format” followed by the construct “timeliness” are the most important factors.
5.8 Developing an internet banking adoption model
The basic objective was to first determine whether TAM is applicable for internet banking
acceptance in India, and then extend TAM to include antecedents such as subjective norm,
image, government support, trialability, trust and perceived risk, which have been identified
in existing literature. Three new constructs Internet Usage Efficacy, Internet Banking Self
Efficacy and Banks Initiative relevant to internet banking proposed in this study, were also
envisaged to be antecedents of the constructs found in the original TAM in the internet-
banking context. Internet Usage Efficacy and Internet Banking Self efficacy were based on
the SCT. In this part of the study, the SEM was used. The primary reason behind choosing
SEM was that the relationship between constructs was interdependent. (Hair et al., 1998),
state that SEM was found to be more suitable in situations where the dependent variables
become independent variables in subsequent dependence relationships.
191
society, there appears to be a male bias even in the current survey. Almost 98% of the
middle aged respondents. Table 5.36 illustrates the summary of the respondents’ gender, age,
education and income.
1 Gender
Female 70 23.3
Male 230 76.7
Total 300 100
2 Age
20-30 158 52.7
31-40 97 32.3
41-50 40 13.3
more than 51 5 1.7
3 Education
Bachelors 144 48
Masters 151 50.3
PhD or more 2 0.7
4 Income
< 1.6 lac 12 4
1.6 – 5 lac 104 34.9
5 – 8 lac 79 26.5
> 8 lac 103 34.6
The same dataset was used for the TAM and the E-TAM. Data screening for out of range
values, missing data, outliers, checks for normality and multicollinearity was done prior to
proceeding with the statistical analysis.

Data from the paper based questionnaire and online forms were combined and saved as a
single file. A univariate statistics (Missing Value Analysis) revealed that the percentage of
missing values for all the variables was less than 1%.The Little’s MCAR test on the data gave
the following results Chi-Square as 792.998 (586) with significance=0.000. Little’s MCAR
test shows a significant Chi-Square, indicating that the missing values are not completely at
random. (Little & Rubin, 1987) suggested that the nature of the missing data could be
diagnosed by finding the correlation of the items having missing data. The Pearson’s
192
correlation between the variables SN2 and AU1 were significant at the 0.05 level, indicating
that the data was missing at random. The regression imputation method was used in this study
for missing data imputation.
5.8.2.2 Outliers
Multivariate outliers were detected using Mahalanobis D2. There were 10 outliers with
probability of D2 less than 0.001. None of these outliers had a Cook’s distance greater than
one. (Stevens, 1984) reported that not all outliers need to be deleted. They found that only
outliers with the Cook’s distance of greater than one were influential and worthy of further
investigation to examine, if they can be deleted. The Mahalanobis D2 and Cook’s distance for
all cases are reported in Table D13, (see Appendix D). In this study, all the outliers had a
Cook’s distance less than 1 and therefore none of the outliers were deleted.
5.8.2.3 Normality
Skewness effects test of means and kurtosis effects variance and covariance. Non-Normality
are potential problems, (Kline, 2005; West et al., 1995). The skewness and kurtosis values of
all the items in the scale were examined and reported in Table D14, (see Appendix D). The
univariate skewness and kurtosis statistic are below the cut-off for the data in this study. All
the variables are univariate normal, but in this case of the Mardia’s multivariate kurtosis, the
value is 154.627 and the critical ratio is 52.686, much greater than 1.96, indicating significant
kurtosis indicating significant non-normality.
matrix for the independent variables was calculated and is illustrated in Table D15 (see
Appendix D). The correlation between the variables exceed 0.8, the cut-off prescribed by
(Hair et al., 1998; Cooper & Schindler, 2003; Sekaran, 2006) in two cases. Each independent
variable was regressed against the other independent variables, the tolerance and VIF was
calculated. The tolerance values were above 0.2 and VIF values were below 4 and reported in
Table D16, (see Appendix D). The data meets the cut-off prescribed in the literature for
tolerance and VIF.
193
5.8.3 Measurement model assessment and confirmatory factor analysis for the
Technology Acceptance Model
The two-step approach suggested by (Hair et al., 2006; Schumacker & Lomax, 2004) was
used in this study. The measurement model was examined first, and then the structural model.
The measurement model was used to test convergent and discriminant validity and then the
structural model was used to test the nomological validity. The measurement model for the
basic Technology Acceptance Model (TAM) is shown in Figure 5.3.
Figure 5.3: The measurement model for TAM
194
In the measurement, model shown in Figure 5.3, the rectangular boxes with labels are the
observed or manifest variables also called items, and the latent variables are oval. Double-
headed arrows indicate covariance between the latent variables.
5.8.3.1 Measurement Model fit Assessment

The model was assessed using the Confirmatory Factor Analysis (CFA) approach. The
Maximum Likelihood (ML) estimation method for calculating the model parameters were
selected from the many other options available in the analysis properties dialog box. The
model χ2 = 975.345, df = 289, p = 0.000, χ 2 /df = 3.375
The model fit indices for the measurement model showed GFI=.783, AGFI=.737,
NFI=.843, RFI=.824, IFI=.884, TLI=.869, CFI=.883,
Standardized RMR=.0614, RMSEA=.089(LO 90=.083, HI 90=.095) PCLOSE=.000. Most of

the goodness of fit measures was less than the recommended values, which indicated that the
model could be refined.
5.8.3.2 Model Refinement
Table 5.37 illustrates the factor loading and Squared Multiple Correlations for the items used
in the measurement model.
195
Table 5.37: The factor loading and item reliability of the items used in this study to test TAM
Construct Statements Code Factor Item

loading reliability
(SMC)
Perceived Usefulness Using Internet banking enables me to accomplish my banking tasks quickly.
PU1 .878 .771
Cronbach's alpha=
0.923 I find Internet Banking very convenient for managing my finance.
PU2 .810 .656
Internet banking enables me to utilize my time effectively by not having to stand in long
PU3 .859 .738
queues at the bank counter.
I find that Internet banking is useful in conducting banking transactions
PU4 .829 .687
Internet banking enables me to access banking services at any time.
PU5 .815 .664
Internet banking is available 24 X 7 and therefore enables me to carry out banking
PU6 .739 .546
whenever I like.
With Internet banking, I can access my bank account even on bank holidays.
PU7 .763 .582
Internet Banking enables me to reduce banking cost, such as reduced bank charges and
PU8 .638 .407
transportation cost.
Internet banking enables me to get correct information unlike branch banking where I
PU9 .524 .275
get inappropriate response from the bank staff.
Internet banking eliminates geographic limitation and increases flexibility and mobility
PU10 .796 .634
Internet banking is more useful than other existing channels such bank branches, ATMs
PU11 .500
and phone banking. .250
Perceived Ease of It is easy to use Internet banking.
PEU1 .827 .684
Use
Cronbach's alpha= It is easy to learn how to use Internet banking.
PEU2 .805 .648
0.921
My interaction with Internet banking is clear and understandable
PEU3 .849 .721
It is easy for me to become skillful at using the Internet banking. PEU4 .864 .747
196
Using internet banking does not require a lot of mental effort.
PEU5 .757 .572
Using internet banking increases the quality of my banking services output with
PEU6 .770 .592
minimal efforts.
Internet banking is flexible to interact with.
PEU7 .698 .488
Attitude In general, I have a positive opinion about Internet banking.
ATT1 .878 .770
Cronbach's alpha=
0.933 I like the idea of using internet banking.
ATT2 .923 .852
In my opinion, it is desirable to use Internet banking.
ATT3 .928 .860
Behavioural intention If I have the facilities required for using Internet Banking, I intend to use it.
BI1 .839 .704
Cronbach’s alpha=
0.820 I plan to experiment with, or use Internet banking regularly in the next six months.
BI2 .613 .375
In the future, I intend to continue using internet banking.
BI3 .898 .807
Actual Usage How long have you been using the Internet banking facilities?
AU1 .568 .322
R=0.410
On weekly basis, how many times do you use Internet banking?
AU2 .723 .522
197
The factor loading of each item was observed. It was found that the critical ratio was greater
than 1.96 and therefore each item was significant at the 0.05 level. Standardized loading
should be higher than 0.5 and ideally more than 0.7, (Hair et al., 1998). The Squared Multiple
Correlation of all the items needs to be more than 0.5. In this measurement model, the factor
loading was at least 0.5 for all manifest (observed) variables. Items PU8, PU9, PU11, PEU7,
BI2, and AU1 had standardized regression weights less than 0.7. In this measurement model
PU8, PU9, PU11, PEU7, BI2, AU1 had SMCs below the cut off 0.5. Items having high
correlation and high regression weights in the modification index were identified and are as
shown in Table 5.38
Table 5.38: Modification index with high values in error covariance and regression
weights
Error MI Path MI
Covariance Regression
weight
PU5 PU6 17.430
epu6 epu5 40.491
PU6 PU5 12.383
PU6 PU7 14.626
epu7 epu6 37.366
PU7 PU6 16.050
PU11 PU8 14.370
epu8 epu11 25.089
PU8 PU11 18.516
PU11 PU9 39.876
epu9 epu11 55.995
PU9 PU11 41.315
PU8 PU9 25.127
epu9 epu8 35.271
PU9 PU8 20.203
The items not fulfilling the criteria for factor loading, reliability and having high covariance
modification index along with high regression weights in the modification index were
deleted.
The items PU5, PU6, PU7, PU8, PU9 and PU11 were deleted because they had high
modification index both in covariance and regression weight. The items PEU7 and BI2 were
also deleted because the reliability (SMC) was less than 0.5.
The measurement model after deleting these 7 items is shown in Figure 5.4
198
Figure 5.4: The trimmed TAM measurement model
The trimmed measurement model was again subjected to CFA
The trimmed measurement model had χ 2 =348.437, df =125, p=0.000, χ 2 /df = 2.787
Standardized RMR=.0506, RMSEA=.077 (LO 90=.068, HI 90=.087) PCLOSE=.000
199
The model fit improved and the trimmed model was found to fit the data adequately. Most of
the fit measures met the recommended values indicating that the model was acceptable. If
sample size is more than 200, it is commonly found that the Chi-Square statistic would reject
valid models, (Anderson & Gerbing, 1988; Bagozzi & Yi, 1988). The Chi-Square of the
model in the study was found to be significant in the trimmed model. The sample size used in
the study was 300 and this could be the reason behind the significant Chi-Square. In the
trimmed measurement model, the correlation between the latent factors was less than 0.8 as
recommended by (Kline, 2005). Many other fit measures being in the range that was
recommended in previous literature, further adjustments were not deemed necessary.
Reliability and validity of the instrument was established and confirmed for the trimmed
measurement model. Table 5.39 shows Cronbach’s alpha, Composite Reliability and Average
Variance Extracted (AVE) for the constructs in the study.
Construct Cronbach’s Composite AVE

alpha Reliability
Perceived 0.922 0.925 0.7126
Usefulness
Perceived 0.919 0.9212 0.6615
Ease of Use
Attitude 0.933 0.935 0.8276
Behavioural 0.740 0.8531 0.7445

Intention
Actual Usage 0.410 0.5898 0.4215
AVE should be greater than 0.5 and the Construct Reliability should be greater than 0.7,
(Byrne, 1988; Fornell & Larcker, 1981). Construct reliability should be greater than 0.6 and
AVE should be equal to, or greater than 0.5, (Bagozzi & Yi, 1988). In the trimmed
measurement model, all the constructs except for actual usage satisfy the conditions stated
above. Actual usage being a two-item construct, and the inability to remove any items due to
the threat of model un-identification was one of the reasons for these observations.
200
All the above evidences support convergent validity for all the constructs in the measurement
model except for where the construct Actual Usage. Construct validity is concerned with
finding whether the instrument is measuring what it actually intended to measure, (Churchill,
1995). The measure of validity refers to developing correct and adequate operational
measures for the concept being tested (Malhotra, 1996). In this part of the study, construct
validity was examined by finding convergent and discriminant validity.
5.8.4.1 Convergent validity

Convergent validity was examined to determine whether the items of the same construct are
correlated and discriminant validity was used for conclude whether the items of a construct
do not correlate on other constructs.
Table 5.40 illustrates the factor loading and SMC for the items used in the trimmed
measurement model.
201
Table 5.40: Factor loading and item reliability for the trimmed TAM
Construct Statements Code Factor loading Item reliability
(SMC)
Perceived Usefulness Using Internet banking enables me to accomplish my banking tasks quickly. PU1
.918 .843
Cronbach’s alpha=
0.922 I find Internet Banking very convenient for managing my finance. PU2
.815 .664
Internet banking enables me to utilize my time effectively by not having to stand in long queues at the PU3
.888 .789
bank counter.
I find that Internet banking is useful in conducting banking transactions PU4
.852 .726
Internet banking eliminates geographic limitation and increases flexibility and mobility PU10
.735 .541
Perceived Ease of Use It is easy to use Internet banking. PEU1 .828
.685
Cronbach’s alpha=
0.919 It is easy to learn how to use Internet banking. PEU2 .806
.650
My interaction with Internet banking is clear and understandable PEU3 .856
.733
It is easy for me to become skilful at using the Internet banking. PEU4 .866
.749
Using internet banking does not require a lot of mental effort. PEU5 .754
.569
Using internet banking increases the quality of my banking services output at minimal efforts. PEU6 .764
.583
Attitude In general, I have a positive opinion about Internet banking. ATT1
.878 .772
Cronbach’s alpha=
0.933 I like the idea of using internet banking. ATT2
.922 .850
In my opinion, it is desirable to use Internet banking. ATT3
.928 .861
Behavioural intention If I have the facilities required for using Internet Banking, I intend to use it. BI1
.815 .664
R= 0.740
In the future, I intend to continue using internet banking. BI3
.908 .825
Actual Usage How long have you been using the Internet banking facilities? AU1 .326
.571
R= 0.410
On weekly basis, how many times do you use Internet banking? AU2 .517
.719
202
Convergent validity of the trimmed measurement model was established by using three
criteria
1. Factor loading
2. Average Variance extracted (AVE)
3. Construct Reliability / Composite Reliability
Standardized factor loading of all the items were greater than the recommended value of 0.5,
(Byrne, 2001)
The software IBM SPSS AMOS 21.0.0 (Build 1178) does not have provisions to calculate
AVE and Construct Reliability. The formulae and calculations for AVE and Construct
Reliability are as shown
AVE = Sum of the squared factor loading / number of items
Construct Reliability = (Sum of factor loading)2/ [(Sum of factor loading)2 + (Sum of

standardized error variance)]
Tables E1, E2, E3 and E4, (see Appendix E) show the AVE and Construct reliability
calculation for all the constructs.
5.8.4.2 Discriminant validity

Discriminant validity is established if the AVE from the construct is greater than the variance
shared between the construct and other constructs in the model, (Chin, 1998). Discriminant
validity of the trimmed measurement model was established by using the criteria that the
AVE estimates for the construct should be larger than the square of the inter-construct
correlation estimates. In this part of the study, there are in all five constructs.

Correlation (SIC)
PU PEU ATT BI AU
PU 0.7126 0.5 0.263 0.3648 0.1332
PEU 0.707 0.6615 0.283 0.251 0.1823
ATT 0.513 0.532 0.8276 0.6336 0.1681
BI 0.604 0.501 0.796 0.7445 0.2530
AU 0.365 0.427 0.410 0.503 0.4215
203
In Table 5.41, the diagonal elements are the AVE (shown in green). The values below the
diagonal are the implied correlation of the constructs (shown in red) and the values above the
diagonal are the square of the correlations between the constructs (shown in blue). (Hair et
al., 2006) state that if the AVE is higher than the squared inter-scale correlation of the
constructs, then discriminant validity is supported. In this model, the AVE is greater than the
square of the inter-construct correlation in all cases and hence discriminant validity was
established.
5.8.5 The structural model (TAM)
The measurement model fit and convergent and discriminant validity was established using
the measurement model. (Hair et al., 1995; Kline, 2005; Anderson & Gerbing, 1988)
suggested that after achieving satisfactory measurement model fit and validating all the
constructs a structural model could be tested. The structural model aims to specify the
influence of latent constructs directly or indirectly on the other constructs in the model,
(Byrne, 2001). Following these guidelines in this stage, the structural model was tested in
order to establish nomological validity. Figure 5.5 shows the structural model. The constructs
PU, ATT, BI and AU are endogenous constructs or dependent variables. All the endogenous
constructs have at least one single headed arrow pointing towards it. The construct PEU is the
only exogenous construct. The error terms begin with the alphabet (e) and are indicative of
the measurement error. The residual errors begin with the alphabet (z) and are indicative of
the residual errors in the structural model due to random error influences, which were not
considered in the model.
Figure 5.5 shows the structural model showing the causal relationship between the constructs.
The structural model was tested by using the goodness of fit indices, which indicates the ideal
fit for the model. The path coefficients, which indicate the strengths of the relationship
between the different constructs, were evaluated. The R2 values for the endogenous variables,
indicates the variance explained by the predictor variable, was estimated.
204
Figure 5.5: The structural model of the basic TAM
The structural model had χ 2 =357.601, df =129, p=0.000, χ 2 /df=2.772
Standardized RMR=.0557, RMSEA=.077 (LO 90=.068, HI 90=.087) PCLOSE=.000
Most of the fit measures met the recommended values indicating that the model was
acceptable.
Path coefficients of TAM is shown in Table 5.42
Table 5.42: Path coefficients of the model
Path Standardized Unstandardized S.E. C.R. p

Estimates Estimates
PU PEU .706 .539 .053 10.229 ***
ATT PU .277 .346 .100 3.459 ***
ATT PEU .333 .317 .077 4.119 ***
BI ATT .660 .596 .055 10.937 ***
BI PU .267 .301 .060 5.031 ***
AU BI .515 .717 .144 4.975 ***
205
The SMC values for the constructs are shown in Table 5.43
Table 5.43: Squared Multiple Correlations of the constructs in the TAM for internet
banking
Construct PU ATT BI AU
(SMC) R2 .499 .319 .687 .265
Perceived ease of use had a positive effect on perceived usefulness, with path coefficient
0.706, and explained 49.9% of the variance contained in perceived usefulness. Perceived
usefulness and perceived ease of use, contributed to attitude towards using internet banking.
These factors had path coefficients of 0.277 and 0.333 respectively, and they explained
31.9% of the variance. Perceived usefulness and attitude were associated with behavioural
intention to use internet banking, with path coefficients of 0.267 and 0.660. The construct
explained 68.7% variance contained in behavioural intention. Behavioural intention had a
positive influence on actual usage of internet banking; with path coefficient 0.515. The R2
value for usage was .265, which indicates that 26.5% of variation in usage is explained by its
predictor variable behavioural intention. All the hypothesized paths in the model are
significant as the Critical Ratio (C.R.) was found to be greater than 1.96.
The structural TAM enabled testing of the following hypothesis.
Table 5.44: Hypotheses tested using TAM
Hypothesis
H3. Perceived Ease of Use will positively affect Perceived Usefulness of Supported
internet banking
H1. Perceived Usefulness will positively affect Attitude towards internet Supported
banking
H4. Perceived Ease of Use will positively affect the Attitude towards Supported
internet banking
H33. Attitude will positively affect the Behavioural Intention towards Supported
internet banking
H2. Perceived Usefulness positively influences Behavioural Intention Supported
towards internet banking
H34. Behavioural Intention positively influences Actual Usage of internet Supported
206
A multiple regression approach was used with the endogenous variables in the model as the
independent variable and its predictors as the dependent variable. A comparison of the
regression weights obtained by both multiple regressions and structural equation modelling is
shown in Table 5.45. It was found that the regression weights obtained by both methods are
comparable.
Table 5.45: Comparison of path coefficients obtained by multiple regression & SEM for
TAM
Path Estimates p Estimates p

(Multiple Regression) (Structural Equation
modelling)
PU PEU .653 *** .539 ***
ATT PU .325 *** .346 ***
ATT PEU .328 *** .317 ***
BI ATT .678 *** .596 ***
BI PU .276 *** .301 ***
AU BI .70 *** .717 ***
5.8.6 The Extended Technology Acceptance Model (E-TAM)
The original TAM was then augmented to include the constructs subjective norm and image
found in DTPB, TAM 2 and TAM 3. In this study, three new constructs: Internet Usage
Efficacy, Internet Banking Self Efficacy and Banks Initiative, relevant to internet banking
were also included as antecedents of the constructs found in the original TAM. Internet
Usage Efficacy and Internet Banking Self efficacy were based on the Social Cognitive
Theory (SCT).
For the E-TAM, the same two-step approach suggested by (Hair et al., 2006; Schumacker &
Lomax, 2004) was used to validate the original TAM. Figure 5.6 shows the measurement
model for the extended TAM.
207
Figure 5.6: The measurement model for the extended TAM
5.8.6.1 Measurement Model fit Assessment for the extended TAM

The model was assessed using the CFA approach. The Maximum Likelihood (ML)
estimation method for calculating the model parameters was selected from the many other
options available in the Analysis properties dialog box. The model χ 2 = 4746.908, df =1861,
p=0.000, χ 2 /df = 2.551. The model fit indices for the measurement model showed GFI =
.658, AGFI = 618, NFI = .729, RFI = .706, IFI=.816, TLI = .798, CFI = .814, Standardized
RMR = .072, RMSEA= .072 (LO 90 =.069, HI 90 = .075) PCLOSE=.000
Most of the goodness of fit measures was less than the recommended values, which indicated
that the model can be refined
208
5.8.6.2 Model Trimming
In this measurement model, the factor loading was at least 0.5 for all manifest (observed)
variables. However, items SE4, PU8, PU9, PU11, IUE4, IUE6, IUE7, IUE9, BAI4, BI2, AU1
and AU2 had standardized regression weights less than 0.7.
In this measurement model AU1, AU2, BI2, BAI4, IUE9, IUE7, IUE6, IUE4, PEU7, PU11,
PU9, PU8, SE4, PR1, PR3 had SMC’s below the cut-off 0.5. PR1, PR3, AU1, AU3 were left
as there will be only 1 item left in the Perceived Risk factor which may lead to potential un-
identification.
The items not fulfilling the criteria for factor loading, reliability and having high covariance
modification index along with high regression weights in the modification index were
deleted.
209
Figure 5.7: The trimmed measurement model (E-TAM)
The trimmed measurement model was again subjected to CFA. The trimmed measurement
model had χ 2 = 1977.401, df = 898, p = 0.000, χ 2 /df=2.202
210
The model fit indices for the measurement model showed GFI = .786, AGFI=.742, NFI=.834,
RFI=.809, IFI=.902, TLI=.886, CFI=.901, Standardized RMR=.0520, RMSEA=.063 (LO
90=.060, HI 90=.067) PCLOSE=.000
The model fit improved, and the trimmed model was found to fit the data adequately.
Although the fit was not excellent, it was decided to construct the structural model. For a
model having sample size greater than 250, number of variables greater than 30 having CFI
or TLI greater than 0.90, SRMR less than 0.08, RMSEA less than .07 with CFI of .90 or
higher is considered to have a good fit, (Hair et al., 2006). Many researchers interpret
goodness of fit measures in the .80 to .89 range as representing reasonable fit; scores of .90 or
higher are considered evidence of good fit, (Doll et al., 1994).
In the trimmed measurement model the correlations between the latent factors was less than
0.8 as recommended by (Kline, 2005) and many others. Fit measures being in the range that
was recommended in previous literature, further adjustments were not deemed necessary. The
AVE and Construct Reliability of the latent variables used in the E-TAM are shown in Table
5.46.
Table 5.46: AVE and CR of the latent variables
Construct AVE CR
TRU 0.7542 0.901
PR 0.558 0.789
IBSE 0.721 0.886
PU 0.713 0.925
PEU 0.661 0.921
SN 0.721 0.885
IUE 0.605 0.859
GS 0.782 0.915
BKI 0.738 0.894
IM 0.770 0.931
ATT 0.828 0.935
BI 0.741 0.851
AU 0.411 0.582
TRI 0.715 0.882
211
As recommended by (Byrne, 1988; Fornell & Larcker, 1981), the AVE was greater than 0.5
and the Construct Reliability was greater than 0.7 for almost all the constructs in the trimmed
measurement except actual usage. Actual usage being a two-item construct and the inability
to remove any items due to the threat of model un-identification was one of the reasons for
these observations. All the above evidences support convergent validity for all the constructs
in the measurement model except the construct Actual Usage.

Correlation (SIC) of the latent variables in the extended TAM.
AU BI ATT IM TRI BKI GS IUE SN PEU PU SE PR TRU

AU 0.411 .262 .164 .002 .0001 .151 .0004 .0396 .050 .184 .152 .081 .139 .054
BI .512 0.741 .643 .0005 .036 .481 .035 .204 .0009 .251 .363 .291 .069 .122
ATT .405 .802 0.828 .013 .034 .347 .040 .218 .00001 .283 .263 .25 .135 .178
IM -.049 .024 .116 0.770 .095 .011 .093 .0015 .077 .004 .005 .001 .027 .003
TRI .012 .192 .186 .309 0.715 .019 .002 .0007 .034 .00004 .00002 .0484 .015 .0019
BKI .388 .694 .589 .106 .139 0.738 .179 .297 .0231 .2981 .391 .2981 .052 .269
GS .020 .188 .201 .306 .045 .424 0.782 .091 .199 .035 .037 .042 .003 .075
IUE .199 .452 .467 -.034 .027 .545 .302 0.605 .022 .285 .356 .274 .057 .137
SN -.224 -.031 -.004 .278 .187 .152 .447 .147 0.721 .001 .001 .0002 .011 .003
PEU .429 .501 .532 -.067 .007 .546 .187 .534 -.039 0.661 0.5 .368 .075 .292
PU .389 .603 .513 -.073 .005 .626 .192 .597 .034 .707 0.713 .284 .032 .25
SE .285 .539 .500 .036 .220 .546 .205 .524 .016 .607 .533 0.721 .059 .124
PR -.373 -.263 -.367 .166 -.123 -.230 .058 -.239 .105 -.274 -.181 -.244 0.558 .153
TRU .234 .350 .422 .063 .044 .519 .274 .371 .062 .540 .500 .353 -.392 0.7542
In Table 5.47, the diagonal elements are the AVE (shown in green). The values below the
diagonal are the implied correlation of the constructs (shown in red) and the values above the
diagonal are the square of the correlations between the constructs (shown in blue). (Hair et
al., 2006) state that if the AVE is higher than the squared inter-scale correlation of the
constructs, then discriminant validity is supported. In this model, the AVE is greater than the
square of the inter-construct correlation in all cases, and hence discriminant validity was
established.
5.8.7 The structural model (E-TAM)
After verifying model fit, convergent validity and discriminant validity the hypothesized
structural model was tested. The hypothesized model was described in chapter 3 and was
shown in Figure 3.5. Subjective Norm (β=0.016, ρ=0.420) did not significantly affect
Perceived Usefulness. Subjective Norm (β=0.009, ρ=0.676) did not significantly affect
Behavioural intention. The path from image to perceived usefulness (β=0.008, ρ=0.7) was not
significant. The path from trialability to (β=0.0320, ρ=0.4) other constructs was non-
significant. The path from Government Support to behavioural intention (β=0.0265, ρ=0.40)
was not significant. The path from Government Support to Attitude (β=0.0012, ρ=0.5) was
212
not significant. The path from Image to Perceived Usefulness (β=0.0282, ρ=0.2473) was not
significant. The path from Perceived Risk to Perceived Ease of Use (β=0.01204, ρ= 0.7514)
was not significant, and therefore, these constructs were removed from the model. Guided by
the modification index, some new paths were established (PR to AU, BAI to TRU, BAI to
ISE and ISE to BI). The revised extended structural model is as shown in Figure 5.8.
Banks
Initiative
(BAI)
Internet
Banking
Self-Efficacy
(ISE) Perceived
Usefulness
(PU)
Attitude Behaviour Actual
towards usage Intention Usage
(ATT) (BI) (AU)
Perceived
Ease of use
(PEU)
Internet
Usage
Efficacy
(IUE)
Government
Support
(GS)
Trust (TRU)
Perceived
Risk (PR)
Figure 5.8: Revised version of the Extended TAM after removing the non-significant
paths
The revised extended structural model was tested by using the goodness of fit indices, which
indicates how well the data fits the model. The path coefficients, which indicate the strengths
of the relationship between the different constructs, were evaluated. The R2 values for the
endogenous variables that indicates the variance explained by the predictor variable was
estimated.
213
The structural model had χ 2 =1381.951, df =570, p=0.000, χ 2 /df=2.424
NFI=.848, RFI=.832, IFI=.905, TLI=.894, CFI=.904, Standardized RMR=.0791,

RMSEA=.069 (LO 90=.064, HI 90=.074) PCLOSE=.000. Most of the fit measures met the
recommended values indicating that the model was acceptable. Path coefficients of revised
extended TAM is shown in Table 5.48
Table 5.48: The regression weights of the variables in the revised and extended TAM
Path Standardized Estimates S.E. C.R. P

regression
weights
ISE IUE .307 .315 .070 4.502 ***
ISE BAI .393 .432 .076 5.680 ***
PEU ISE .455 .377 .055 6.806 ***
PEU IUE .312 .265 .053 4.995 ***
PU PEU .521 .395 .047 8.426 ***
PU BAI .372 .257 .038 6.817 ***
TRU PR -.308 -.336 .068 -4.921 ***
TRU GS .122 .117 .057 2.044 .041
TRU BAI .410 .489 .076 6.399 ***
ATT PU .180 .223 .096 2.316 .021
ATT PEU .215 .201 .076 2.663 .008
ATT TRU .161 .115 .041 2.816 .005
ATT IUE .203 .162 .054 2.995 .003
BI ATT .622 .611 .053 11.593 ***
BI PU .236 .286 .070 4.075 ***
BI ISE .110 .084 .042 2.008 .045
AU BI .449 .615 .125 4.920 ***
AU PR -.246 -.258 .089 -2.903 .004
Table 5.49: Squared Multiple Correlations of the constructs in the extended TAM
Internet Trust Perceived Perceived Attitude Behavioural Actual

Banking (TRU) Ease of Usefulness (ATT) Intention Usage
Self- Use (PEU) (PU) (BI) (AU)
efficacy
(IBSE)
Estimate .386 .374 .455 .580 .343 .676 .299
Figure 5.9 shows the Hypothesized model. Dashed lines with arrows indicate non-significant
paths.
214
Subjective norm
Image
Banks Initiative
Perceived
usefulness
Internet banking Attitude toward Behavioral Actual Usage
Self-Efficacy Using Intention
Perceived Ease of
use
Internet usage
Efficacy
Trust
Perceived Risk
Trialability
Government
support
Figure 5.9: Extended TAM model 215
The structural revised and extended Technology Acceptance Model enabled testing of the following hypothesis.
Table 5.50: Hypotheses tested using E-TAM
Hypothesis
H22. Subjective norm will positively affect behavioural intention towards use of internet banking Not Supported
H20. Subjective norm will positively affect perceived usefulness of internet banking Not Supported
H21. Subjective norm will have a positive influence on image. Supported
H33. Image will have a direct positive influence on attitude towards using internet banking. Not Supported
H32. Image will positively affect perceived usefulness of internet banking Not Supported
H34. Image will have a direct positive influence on intention towards using internet banking Not Supported
H29. Trialability will have a direct positive influence on intention towards using internet banking Not Supported
H30. Trialability will positively affect perceived ease of use towards internet banking Not Supported
H31. Trialability will have a direct positive influence on perceived usefulness Not Supported
H24. Government support will positively affect attitude towards internet banking Not Supported
H23. Government support will positively affect behavioural intention towards internet banking Not Supported
H25. Government support will positively affect trust towards internet banking Supported
H5. Trust in internet banking will have a positive effect on perceived ease of use. Not Supported
H6. Trust in internet banking will have a positive effect on perceived usefulness. Not Supported
H8. Trust in internet banking will have a positive effect on behavioural intention towards internet banking. Not Supported
H15. Internet banking self-efficacy positively influences the perceived usefulness towards using internet banking. Not Supported
NR. Internet banking self-efficacy positively influences the behavioural intention towards using internet banking Supported
H18. Internet usage efficacy positively influences the perceived usefulness towards using internet banking. Not Supported
H27. Banks initiative will positively influence perceived ease of use towards internet banking. Not Supported
H7. Trust in internet banking will have a positive effect on attitude towards internet banking Supported
H3. Perceived Ease of Use will positively affect Perceived Usefulness of internet banking Supported
H2. Perceived Usefulness will positively affect Attitude towards internet banking Supported
H4. Perceived Ease of Use will positively affect the Attitude towards internet banking Supported
H35. Attitude will positively affect the behavioural intention towards internet banking Supported
H1. Perceived Usefulness positively influences Behavioural Intention towards internet banking Supported
H36. Behavioural Intention positively influences Actual Usage of internet Supported
216
NR. Perceived Risk will negatively affect usage of internet banking Supported
H9. Perceived Risk will negatively affect perceived usefulness towards internet banking Supported
H10. Perceived Risk will negatively affect perceived ease of use towards internet banking Not Supported
H11. Perceived Risk will negatively influence intention towards internet banking Not Supported
H12. Perceived Risk will negatively affect trust towards internet banking Supported
H14. Internet Banking Self-Efficacy will positively affect perceived ease of use towards internet banking Supported
H17. Internet Usage Efficacy will positively affect perceived ease of use towards internet banking Supported
H19. Internet Usage Efficacy will positively affect Internet Banking Self-Efficacy towards internet banking Supported
H16. Internet Usage Efficacy will positively affect attitude towards internet banking Supported
NR. Banks Initiative will positively affect Internet Banking Self-Efficacy towards internet banking Supported
NR. Banks Initiative will positively affect Trust towards internet banking Supported
H26 Banks Initiative will positively affect Perceived Usefulness towards internet banking Supported
NR- New Relationship found in this study
217
A multiple regression approach was used with the endogenous variables in the revised extended
model as the independent variable and its predictors as the dependent variable. A comparison of
the regression weights obtained by both multiple regressions and structural equation modelling is
illustrated in Table 5.51. It was found that the regression weights obtained by both methods are
comparable.
Table 5.51: Comparison of the path coefficients obtained by multiple regression and SEM
for E-TAM
Path Estimates p Estimates p

(Multiple (Structural
Regression) Equation
modelling)
ISE IUE .336 *** .315 ***
ISE BAI .410 *** .432 ***
PEU ISE .372 *** .377 ***
PEU IUE .341 *** .265 ***
PU PEU .537 *** .395 ***
PU BAI .255 *** .257 ***
TRU PR -.292 *** -.336 ***
TRU GS .168 .001 .117 .041
TRU BKI .353 *** .489 ***
ATT PU .238 .002 .223 .021
ATT PEU .212 .001 .201 .008
ATT TRU .103 .018 .115 .005
ATT IUE .195 *** .162 .003
BI ATT .629 *** .611 ***
BI PU .240 *** .286 ***
BI ISE .101 .009 .084 .045
AU BI .636 *** .615 ***
AU PR -.173 .008 -.258 .004
The revised extended TAM for internet banking was then subjected to model invariance test
between groups.
218
5.8.8 Measurement Invariance and moderating effects of the demographic variables
on the variables in the model
Measurement invariance gives evidence that the instrument is working in the same manner
across different groups. (Horn & McArdle, 1992), argue that establishing psychometric
properties of an instrument with just one representative of the overall population does not
guarantee identical measurement properties for population subgroups. According to (Doll et al.,
1988; Klenke 1992) the invariance of the model across different subgroups is important as it
confidence to the researcher about the findings. The seminal work of (Jöreskog, 1971) led to the
development of the process of multi-group invariance testing. The parameters of interest, while
testing for equivalence across groups are usually factor loading, structural regression paths and
factor covariances.
Traditionally, the χ2 difference test had been employed for assessing invariance between groups.
The χ2 difference test is influenced by sample size, (Kelloway, 1998; Brannick, 1995). (Cheung
& Rensvold, 2002) based on a simulation analysis of 20 fit indices proposed that a CFI
difference of less than 0.01 for evaluating multi-group measurement invariance. This alternative
criteria based on the difference in CFI is increasingly being used by researchers. (Byrne, 2001)
points out that researchers can be confronted with diametrically opposite conclusions based on
these two criteria for determining measurement invariance and recommend the CFI difference
approach to be more practical. The χ2 difference and the CFI difference are reported, but the
decision to decide invariance was guided by the CFI difference based on a cut-off of 0.01.
The data based on 4 demographic dimensions gender, age, income, education was divided into
two groups for each of these demographic dimensions. The invariance test was first performed
on the measurement model and then on the structural model. The measurement model and
structural model were subjected to tests of equivalence of parameters across groups. The model
fit for each group separately and multi-group is as reported in Appendix F. The z test value was
used as the test for significance of the difference in factor scores for the two groups formed from
each of the 4 demographic variables.
219
5.8.8.1 Testing for invariance across gender
In this study, there were 230 male respondents and 70 female respondents. Table F1, (see
Appendix F) shows the χ2, df, and fit statistics of the unconstrained model along with the model
with constrained measurement weights, structural weights and structural covariances.
5.8.8.1.1 The measurement model

In the original Technology Acceptance Model (TAM), the CFI difference values for the model
with measurement weights and structural covariance constrained equal was below the cut-off
value of 0.01, which suggested that the constraints associated with metric and scalar invariance
did not significantly degrade the overall fit of the model. On reviewing the Table F1, (see
Appendix F) for individual factors and factor loading non-invariance was found for 3 items
PEU3 (p < 0.05), PEU4 (p < 0.1) and PEU5 (p < 0.1). The non-invariance for these 3 items
indicated that they operate differently for male and female respondents.
The extended Technology Acceptance Model also had CFI difference values less than 0.01.
However, factor-loading non-invariance was found for only 1 item BAI2 indicating that this item
operates differently for both the groups.
5.8.8.1.2 The structural model

In the original TAM, a significant difference in the relationship between Perceived Ease of Use
and Attitude was found between male and female respondents with female respondents
exhibiting a stronger effect.
In the extended TAM the relationship between the seven constructs were significantly different.
The link Perceived Ease of Use and Attitude was strong in the case of females and not significant
for males. The relation between Internet Usage efficacy and Internet banking Efficacy not
significant in females but was significant in case of males. Banks Initiative and Perceived
Usefulness relationship non-significant in the females and was significant in the case of males.
The relation between trust and attitude was not significant for female but was significant in
males. The relation between internet usage efficacy and attitude was not significant for females
but significant for males. Banks initiative and Trust relationship not significant in the case of
females but was strong in case of males. Perceived Risk and Actual Usage is negative and strong
for females and not significant for males.
220
5.8.8.2 Testing for invariance across age
In this study there were 158 respondents less than or equal to 30 years of age, they were
classified as Age group 1 the other 142 respondents who were above 30 years of age were
classified as Age group 2.

The CFI difference for factor covariance exceeds the cut off 0.01 for TAM and extended TAM
indicating that the factor covariances are not equivalent across the groups. Table F2, (see
Appendix F) reveals non-invariance for 2 items PEU3 and PU2 in the original TAM.
In the extended TAM non-invariance was found for the items SE1, PU2, PEU5, SN2, IUE2,
IUE3, BAI2, BAI1 and IM2.

The CFI difference for the original TAM was below the cut-off for factor loading, structural
weights and structural covariance but for the extended TAM, the CFI difference was above the
cut-off for structural weights.
In the original TAM, the relationship of PEU with PU was significantly different with age group
1 showing a stronger influence. The relationship between PU and BI was also significantly
different with age group 1 having a strong influence and this link not significant in the case of
age group 2.
In the extended TAM, the relationship between PEU and PU was also found to be significantly
different across the two groups with age group 1 showing a stronger influence. The relationship
between BAI and PU was found to be stronger in age group 1 compared to age group 2. The
relationship between BAI and TRU was also found to be significantly different across both the
groups with age group 1 exhibiting a stronger influence and age group 2 showing a non-
significant path. The relationship between PU and ATT was not significant for age group 1 but
was significant and strong in age group 2. The relationship between PU and BI was also
significantly different with age group 2 showing a non-significant relationship. The relationship
between PR and AU was found to be negative and this link was not significant in case of age
group 2.
221
5.8.8.3 Testing for invariance across income level
In this study, there were 116 respondents belonging to income level 1 and 182 respondents who
belonged to income level 2. Respondents who belonged to income level 1 had an annual income
less than or equal to 5 lacs and respondents belonging to income level 2 had an annual income
more than 5 lacs.

The CFI difference in the case of the original TAM and extended TAM were found to be less
than the cut-off of 0.01.
In the original TAM, there was no significant difference in the factor loading across the two
groups and all factor loading were found to be significant.
In the extended TAM the factor loading were found to be significantly different for the two
groups for the items BAI1 and AU1.

The CFI difference was more than the cut-off for TAM and extended TAM when structural
weights and structural covariances are constrained.
In the original TAM, the relationship between PEU and PU was stronger for income level 1. The
relationship between PEU and ATT was not significant for respondents from income level 1 but
strong for respondents belonging to income level 2. The ATT and BI relationship was
significantly different across the two groups with the influence being stronger for income level 1.
The PU and BI relationship was found to be not significant for income level 2. The BI and AU
relationship was found to be not significant for income level 1.
For the extended TAM, significant difference was found between PEU and PU with the
influence stronger for income level 1. The relationship between BAI and TRU was found to be
stronger for income level 2. The relationship between BI and AU was strong for income level 2
and not significant for income level 1 respondents. The relationship between PR and AU was
negative and stronger for income level 1, but for income level 2 respondents this relationship was
not significant.
222
5.8.8.4 Testing for invariance across education levels
In this study, there were 144 respondents who had bachelors or lower education and the other
153 respondents had a masters or higher degree. These respondents were classified as having
limited and expanded education.

The CFI difference criteria was below the cut-off when measurement weights and covariances
were constrained for the original TAM but for the extended TAM the CFI difference was equal
to the cut-off when structural covariances were constrained. On investigating further, the original
TAM did not show any significant difference in the factor loading for all the items for both
groups. The extended TAM showed significant difference in factor loading for 9 items TRU3,
SE2, SE1, SN2, SN1, IUE2, GS2, BAI2 and AU1.

The structural model for the original TAM showed significant difference in the relationship
between PU and BI with the respondents belonging to the expanded education group showing a
higher significance.
In the extended TAM, four relationships were significantly different for the two groups. ISE and
PEU had a stronger influence shown by respondents belonging to the limited education group.
IUE and PU had stronger influence shown by respondents from the expanded education group.
BAI and PU were stronger for the respondents with expanded education. PR and AU with a
negative structural weight and respondents with expanded education showing a stronger
influence and was not significant for the limited education group.
5.8.9 Findings
It was found that about 29.9% variation in the usage of internet banking was caused by its
predictor variables. Perceived Risk was found to have a direct negative influence on the usage of
internet banking. Subjective norm and Image on Perceived Usefulness and subjective norm on
behavioural intention was found to be not significant. Government support was found to be not
significant on all other constructs in the study except trust. The construct Banks Initiative on
Perceived Usefulness, Internet banking self-efficacy, Behavioural intention and Trust was found
223
to be positive and significant. The study found that both internet usage efficacy and internet
banking self-efficacy had a significant positive influence on perceived ease of use. Internet
Usage Efficacy was found to have a significant positive effect on attitude towards internet
banking. The construct trialability was found not to be a significant influencer of any of the
constructs. Perceived usefulness and Perceived Ease of Use had a significant positive effect on
attitude towards internet banking, which in turn significantly influenced Behavioural intention.
Perceived Ease of Use had a very high significant influence on Perceived Usefulness indicating
that if banks make the process of internet banking simple customers will find it more useful.
Chapter Summary
In this chapter, the data analysis and results obtained in the eight phases of the research were
presented. Interpretation of the results and key findings were discussed. The extended model for
internet banking acceptance was able to explain the relationship between the variables involved
in influencing internet-banking use. It is essential for banks to understand the strength of the
relationship, so that marketing strategies can be targeted based on the demographic profiles of
the customers. A comprehensive analysis of the findings from different stakeholders’
perspectives reveals that traditional bank branches along with the right mix of other distribution
channels would be essential for banks to remain competitive. The next chapter will focus on
drawing conclusions and will suggest measures, which can lead to increased usage of the
internet-banking channel.
224

11.chapter 5

Uploaded by

Document Information

Original Description:

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

11.chapter 5

Uploaded by

Copyright:

Available Formats

Chapter 5

Data Analysis and Interpretation

5.1 The state of the internet in India

Table 5.1: Sample Demographics (phase 1)

5.1.2 Data screening and preparation for analysis

5.1.2.1 Missing Data

5.1.3 Exploratory Factor Analysis

How much do you agree or disagree with the following Factors

Government control would make me fearful that my actions were .626

Government control would make the internet too controlled .710

Government control would inhibit the growth of the internet .688

Government control would make me use the internet less .670

Freedom of expression is guaranteed on the internet. .535

The internet is essential for my access to knowledge and education .503

My life has improved due to using the internet. .669

The following underlying themes were identified as factors.

Figure 5.1: The measurement and revised structural model (phase 1)

Construct Government Government Attitude

5.1.4.2 Measurement and revised structural model fit

Statistic Measurement Revised Recommended References

χ2 399.073 399.227 ----------- -----------

degrees of freedom 146 148 ----------- -----------

ρ 0.000 0.000 >.05 (for sample (Bagozzi & Yi, 1988)

χ 2 /df 2.733 2.697 <5 (Wheaton et al, 1977)

Root Mean Square 0.013 0.013 < 0.07 (Steiger, 2007)

Comparative Fit 0.944 0.945 > 0.9 (Bentler, 1990)

Incremental Fit 0.945 0.945 > 0.9 (Bollen, 1990)

Path Standardized Unstandardized S.E. C.R. P

Table 5.7: Standardized direct and total effects

Government Support Government Control Attitude

5.2 Evaluation of Internet Banking Sites in India Based on the

5.2.1 Comparison of bank websites based on multilingual support provisions

Dissemination Transaction Relationship Overall

Central Bank of India 2 3 2 3 4 2 3 2 0 7 9 5 21

Indian Overseas Bank 2 3 2 1 3 1 3 2 0 7 5 5 17

Punjab & Sind Bank 2 2 2 0 3 1 3 2 0 6 4 5 15

Punjab National Bank 3 3 2 3 4 2 3 2 3 8 9 8 25

State Bank of India 3 1 1 3 3 2 3 2 3 5 8 8 21

State Bank of Bikaner 1 2 1 3 2 1 3 1 1 4 6 5 15

State Bank of Indore 1 2 1 2 2 2 3 1 1 4 6 5 15

State Bank of Patiala 0 2 1 2 2 2 3 1 1 3 6 5 14

State Bank of Mysore 0 2 1 2 2 2 3 1 1 3 6 5 14

Union Bank of India 2 3 2 3 4 2 3 3 1 7 9 7 23

United Bank of India 2 3 0 1 3 1 3 2 0 5 5 5 15

Table 5.10: Private sector bank website evaluation scores

Dissemination Transaction Relationship Overall

Catholic Syrian Bank 2 3 1 1 1 0 3 2 0 6 2 5 13

City Union Bank 2 3 2 1 3 0 3 2 0 7 4 5 16

ING Vysya Bank 3 3 2 3 4 1 3 2 0 8 8 5 21

Jammu and Kashmir 2 3 2 3 4 1 2 2 1 7 8 5 20

Lakshmi Vilas Bank 3 3 2 2 2 0 3 2 1 8 4 6 18

South Indian Bank 3 3 2 3 3 0 3 3 2 8 6 8 22

Table 5.11: Foreign sector bank website evaluation scores

Dissemination Transaction Relationship Overall

Table 5.12: Averages of Dissemination, Transaction, Relationship and Overall scores

Dissemination Transaction Relationship Overall

Q7. Does internet-banking usage reduce costs for the bank?

Cost reduction and increase in sales

5.3.2 Interviews with Branch employees

5.3.3 Interviews with users and non-users of internet banking

Reasons for not using internet banking N=75 N%

Banks support Lack of awareness about how to apply for internet 6

Lack of support in case of any problems 12