Citation
Online credit card bill payment and personality type

Material Information

Title:
Online credit card bill payment and personality type
Creator:
Borkan, Gary L. ( author )
Place of Publication:
Denver, CO
Publisher:
University of Colorado Denver
Publication Date:
Language:
English
Physical Description:
1 electronic file (117 pages). : ;

Subjects

Subjects / Keywords:
Credit cards -- Social aspects ( lcsh )
Adaptability (Psychology) ( lcsh )
Genre:
bibliography ( marcgt )
theses ( marcgt )
non-fiction ( marcgt )

Notes

Abstract:
The internet has become a fully integrated part of the worlds business community and social culture. An important part of this integration i the internets interaction with credit cards, both as a medium of exchange and transaction, and as an interface for credit card account customer service. Credit card providers and bank account service site provide powerful tools for the individual to manage their credit card accounts, and allowing payment of monthly credit card. It is an essential for provider organizations to maximize the value delivered f itself and to their customers, to justify the expense involved in developing and maintaining these service sties and transactional banki links. This study builds upon previous work done on the relationship between personality type and internet usage. It expands this pervious world to focus specifically on user adoption of online credit card bil payment services. This work can provide online payment service designers insights into the personality traits of their users that mig be exploited in order to increase the value derived by the credit car payment service providers and banking organizations ads well as enabli wider adoption of online credit card payment services by addressing varying personality-based needs of users.
System Details:
System requirements: Adobe Reader.
Thesis:
Computer science and information systems
General Note:
Department of Computer Science and Engineering
Statement of Responsibility:
by Gary L. Borkan.

Record Information

Source Institution:
|University of Colorado Denver
Holding Location:
|Auraria Library
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
891144053 ( OCLC )
ocn891144053

Downloads

This item has the following downloads:


Full Text
ONLINE CREDIT CARD BILL PAYMENT AND PERSONALTIY TYPE
by
GARY L. BORKAN
B.S., National American University, 1995
M.S., University of Colorado Denver, 2000
A thesis submitted to the
Faculty of the Graduate School of the
University of Colorado in partial fulfillment
of the requirements for the degree of
Doctor of Philosophy
Computer Science and Information Systems
2013


ii
This thesis for the Doctor of Philosophy degree by
Gary L. Borkan
has been approved for the
Computer Science and Information Systems Program
by
Deborah Kellogg, Chair
Steven Walczak, Advisor
Michael Mannino
Madhavan Parthasarathy
Min-Hyung Choi
November 15, 2013


Borkan, Gary L. (Ph.D., Computer Science and Information Systems)
Online Credit Card Bill Payment and Personality Type
Thesis directed by Associate Professor Steven Walczak
ABSTRACT
The Internet has become a fully integrated part of the worlds business
community and social culture. An important part of this integration is the Internets
interaction with credit cards, both as a medium of exchange and transaction, and as an
interface for credit card account customer service. Credit card providers and bank
account service sites provide powerful tools for the individual to manage their credit card
accounts, and allowing payment of monthly credit card. It is essential for provider
organizations to maximize the value delivered for itself and to their customers, to justify
the expense involved in developing and maintaining these service sites and transactional
banking links. This study builds upon previous work done on the relationship between
personality type and Internet usage. It expands this previous work to focus specifically
on user adoption of online credit card bill payment services. This work can provide
online payment service designers insights into the personality traits of their users that
might be exploited in order to increase the value derived by the credit card payment
service providers and banking organizations as well as enabling wider adoption of online
credit card payment services by addressing varying personality-based needs of users.
The form and content of this abstract are approved. I recommend its publication.
Approved: Steven Walczak


IV
DEDICATION
For my parents, Lillian K. Borkan, B.S. History, & M.S. Education and Library
Science, and Jacob D. Borkan, B.S, Chemical Engineering.
First, to my Mother, educator and librarian. As a child she taught me stories of
her hero and her heroine, Dr. Jonas Salk who succeeded in ridding the world of polio, and
Hypatia of Alexandria who unsuccessfully fought to save the knowledge of five thousand
years of accumulated wisdom from destruction by a mob of religion blinded zealots.
Through her stories she instilled in me the deepest reverence for books, and a love for
science and the progress of knowledge. She viewed these as the underlying foundation of
civilization and as "the basis for the progress that has led us up from darkness to face
unafraid the light of the stars". These were her words and were her belief, and
throughout her lifetime what she taught her pupils. She achieved academic and career
goals far beyond those deemed appropriate for the average women to pursue in the late
1940's and early 1950's.
To my Father, who although disabled through service to our country, persevered
to complete a degree in the physical sciences, and entered upon a career of research. His
work led first to advances towards stealth technology in torpedo design, and later with his
research partner resulted in the development and patenting of a practical version of the
Silver Catalyst Reaction Control System. More commonly known as the hydrogen
peroxide reaction thruster (thrusters), his designs for the catalyst packs in these engines
were present on research vehicles such as ultra-high altitude models of the FI 04
Starfighter, and the FI5. Finally, working as a government subcontractor under FMC


Corporation and NASA, he built units that were installed in the escape and attitude
control systems aboard the Mercury, Gemini, and Apollo space capsules.
I grew up surrounded by blue prints, scale mockups, working test stand rockets
and surrounded by an entire library filled with encyclopedias, history books, and physics
manuals. The environment they provided for me yielded a wonder towards nature and
science, and a habit of lifelong learning even though for many years I did not purse
formal education beyond high school.
Their prematurely passing in the spring of 1989 caused me to take stock. At that
time I dedicated myself to accomplishing in some small way the attainment of the hopes
and dreams they had held for me and of which felt I had fallen short. This led to me
enrolling as a freshman in collage at the age of thirty five.
Now it is twenty three years later. Throughout those long years I have relied on
their examples. They are my models of courage, stamina, and determination in the face
of opposition.
They taught me through quite example a belief that you can accomplish what you
determine. That no matter what gender, culture, or minority you may come from,
whether you are strong or handicapped, that you must stand up after mis-steps,
impediments, or to those barring the way. Whether you have two good feet on which to
stand on or are in a wheel chair with nothing to stand upon other than your character that
you can and will succeed. That is what I learned witnessing my parents lives.
This work is dedicated with love to them.


VI
ACKNOWLEDGMENTS
I would like to extend my thanks and gratitude to all those many individuals without
whose support and encouragement this work could not have been completed.
Among those are a number of individuals that so stand out that I would be remiss
in not extending special thanks to them.
I want thank my Advisor, Professor Steven Walczak, whose support, guidance,
and patience endured through many years of work and effort and has never wavered. I
will always appreciate the time and attention he has extended to me, his optimistic
attitude, and his dry sense of humor.
Many thanks to Professor Michael Mannino for his participation as a committee
member and without whose vision and efforts this degree program would not exist.
I would also like to thank Professor Madavan Parthasarathy whose initial
encouragement gave me confidence to pursue this degree and has been an important
resource of information that has been applied to this improve this study.
To the other members of my committee, Chair Deborah Kellogg, and Min-Hyung
Choi, thank you very much for your time and many suggestions which have helped make
this a far better study.
My appreciation to the Faculty and Staff of the IS Department and the Business
Graduate School of the University of Colorado at Denver who have instructed me in the
many classes I have attended.


And finally I would like to thank my fellow students, many who have fallen by
the way, for their help through all the years as friends, and as project and study partners.
Thank you and good luck to all of you in your own studies and careers.


TABLE OF CONTENTS
viii
CHAPTER
I. INTRODUCTION........................................................ 1
II. BACKGOUND AND RELATED WORK ......................................... 6
TAM................................................................... 6
Diffusions of Innovations............................................. 9
Myers Briggs Type Indicator.......................................... 12
Personality Type and the Internet.................................... 14
Current Usage of Online Credit Card Bill Payment Services.............16
Scope and Contribution............................................... 18
III. APPROACH I: MODELING THE EFFECT OF PERSONALITY TYPE ON
ONLINE CREDIT CARD PAYMENT SERVICES.................................... 21
Conceptual Model......................................................21
Research Hypotheses.................................................. 22
Methodology.......................................................... 29
IV. ANALYSIS........................................................... 32
SEM Model Analysis................................................... 32
Assumptions of the Structural Model...................................32
Assumptions of the Reflective Measurement Model...................... 33
Reliability of the Constructs.........................................34
Normality of the Constructs...........................................35
Analysis of the Model Fit.............................................37
R and Squared Multiple Correlations...................................38
Convergent Validity...................................................39


ix
Discriminant Validity................................................... 40
Structural Measurement Analysis.........................................41
Mediation Testing....................................................... 43
Additional Variables Tested............................................. 44
Testing for Variance Due to Gender...................................... 44
PLS Model Analysis...................................................... 45
PLS Measurement Model................................................... 46
PLS Reliability......................................................... 47
PLS Validity............................................................ 48
PLS Structural Model.................................................... 49
Discussion.............................................................. 51
V. APPROACH II: DISCOVERING DESIRABLE AND UNWANTED ONLINE
CREDIT CARD PAYMENT SERVICE FEATURES FOR DIFFERENT
PERSONALITY TYPES......................................................... 55
Methodology............................................................. 56
Data.................................................................... 57
Analysis................................................................ 58
Demographics............................................................59
Cluster Analysis........................................................ 59
Results and Findings.................................................... 62
Aspects of Personality Types............................................ 62
MBTI and Perceptions of Online Payment Service Features................ 65
VI. LIMITATIONS AND IMPLICATIONS........................................... 72
Implications for Payment Service Systems Designers...................... 73


X
Implications for Research............................................. 75
VII. CONCLUSIONS....................................................... 79
Conclusions Approach I............................................... 79
Conclusions Approach II............................................... 80
REFERENCES............................................................. 82
APPENDIX
A. Survey Instruments................................................ 93
B. SEM Tests....................................................... 100
C. Survey II Variable Key.......................................... 102
D. Conceptual Model Construct Detail................................ 103


XI
LIST OF TABLES
Tables
1. Some antecedent variables added to TAM in previous research................. 9
2. Comparative constructs of DOI TAM & Credit Card Online Payments............. 12
3. Respondent demographics...................................................... 29
4. Myers Briggs Dichotomy Variable Conversion Values........................... 31
5. Reliability Indices.......................................................... 35
6. Assessment of normality utilizing algorithm (Logl0(8-X)*-1)................. 36
7. Measures of Fit.............................................................. 37
8. Squared Multiple Correlations (R2).......................................... 39
9. Convergent Validity Construct Factor Loadings............................. 40
10. Squared Correlations Estimate with AVE on diagonal......................... 41
11. Regression intercepts of MBTI............................................. 43
12. Variable to Variable Path Beta Coefficients/Regression Weights............. 43
13. Mediation Test Statistics................................................... 44
14. PLS Reliability Tests values................................................ 48
15. PLS Squared Correlations Estimate with AVE on Diagonal..................... 48
16. PLS Path Coefficients....................................................... 49
17. PLS MBTI Coefficients and Derived Pair Indicator.......................... 50
18. PLS Derived MBTI Types Per Constructs..................................... 50
19. Credit Card Website Perception Survey Respondent Demographics.............. 59
20. Cluster Analysis Agglomeration Schedule.................................... 60
21. Important credit card payment service features and personality aspects.... 63


xii
22. Important credit card payment service features for MBTI personality types........ 67
23. Feature perceptions by non-users of online credit card payment services........... 70


LIST OF FIGURES
xiii
Figure
1. Conceptual Model........................................................... 22
2. Structural Equation Model.................................................. 31
3. PLS Model diagram...........................................................47
4. SEM MBTI Dimension Indirect Effect on U................................... 53
5. SEM MBTI Dimension Direct Effect on PC, PU, & PEU..................... 54
6. Combined Linkage Dendrogram................................................ 61
7. Percentage of MBTI personality types in survey population..................66


1
CHAPTERI
INTRODUCTION
The Internet, having become an integral and ubiquitous part of society, culture, and
business is having a transformational effect on all aspects of the modern world (Schell,
2007). One of the most critical areas having been transformed is the area of Internet bill
payment. The very nature of the Internet makes the use of physical currency as an
exchange medium for online transactions untenable. Therefore business conducted
online relies extensively on the use of either credit card or electronic check transactions.
The Internet and the credit card have grown to have a symbiotic relationship unlike any
other market place and exchange medium previously seen. It seems only natural that
credit card providers and consumer banks would further enhance that relationship, as well
as exploiting the power and convenience of the Internet, by developing online credit card
bill payment service features. Credit card service websites parallel general banking
websites and allow credit card holders access to extensive account management services,
with both offering consumers a convenient method of paying monthly credit card bills.
This study considers payment sites hosted by credit card provider organizations, and
consumer banking services sites both as payment service sites for the purposes of
consumer payment of monthly major credit card bills and has not distinguished between
the two.
Examining electronic commerce utilization, specifically credit card payment services,
from an individual perspective is problematic due to the sheer volume of data that must
be analyzed to ensure accuracy of the research model. Therefore, a classification schema


2
for research populations that helps preserve individual differences is desired. Utilizing
personality type indicators may provide such a classification technique.
This research uses the accepted measurement tools of the Myers-Briggs Type Indicator
(MBTI) personality test, the Technology Acceptance Model (TAM), and the Theory of
Diffusion as lenses to analyze user personality type and determine what if any affect that
may have on the behavioral intention to use online credit card bill payment services as
their payment channel.
Studies of the effect of personality on the adoption of technologies have been
used to analyze various online user behaviors. Research has shown that usage of the
Internet, especially in the subcategories of social networking and interaction websites,
has been correlated to personality types (Amichai-Hamburger & Ben-Artzi, 2003).
However, as there has been little or no research done on the effects of personality on the
adoption of e-commerce generally and online credit card bill payment specifically, this
study extends research into this new area.
Credit card transactions play a pivotal role in online commercial transactions. Online
credit card service sites and their consumer bank partners provide valuable conveniences
to credit card holders meeting a growing demand for online account services, and
providing savings through the value chain to the provider through transaction cost
reduction (Langdon & Shaw, 2000). The initial start-up costs of an online service sites
and their accompanied development of sockets to consumer banks, coupled with ongoing
expenses incurred for maintenance and upgrade represent a significant capital investment
to the organization (Abramson & Morin, 2003). There are also legal considerations and
expenses involved with providing these online service sties and bank partner service


3
linkages. As transactions involve account access and management, any possible breach
of security on these sites leave the provider exposed to possible litigation if users incur
damages. With these costs and risks to the credit card provider and its partners involved,
it becomes paramount that the online payment services yield value to the organizations.
Coupled with these issues are data which indicate a falloff in the number and rate of Bank
Account Number Payments (BANP) which reflects a fundamental business problem of
increased cost per transaction (US Federal Reserve System, 2011). To increase value, it
is not only beneficial, but vital, to increase the customers use of online payment services
in order to increase the value to service providers.1 Therefore, there is benefit to an
organization in understanding those attributes that contribute to a potential to increase
any particular credit card holders likelihood to use the online payment services.
It is also important for service site providers and designers to understand the needs of
their predominant users. Website designers who learn the personality profile of their
clientele will be better able to focus on their needs (Albert, Goes, & Gupta, 2004;
Amichai-Hamburger, 2002; Guadagno, Okdie, & Eno, 2008). For many individuals,
online business transactions hold an elevated level of risk and uncertainty (Lee & Turban,
2001). Differences in personality type have an effect on a user's tolerance for uncertainty
1 It has been estimated that online payments may reduce cost per transaction to as low as $.01 each. Some
institutions calculate an over the counter payment as costing $1.07 per transaction (Akinci, Aksoy, &
Atilgan, 2004; Didio, 1998; Borzekowski, & Kiser, 2007; Nath, Parzinger, & Bank, 2001, Ozdemir, &
Trott, 2009). That would represent of savings of $1.06 per payment. This equals $1060.00 per thousand
online users per month. With 576.4 million credit card bills being paid in the US each month, even a small
increase in the percentage of users represents a substantial potential savings (Nilson Staff, 2010).


4
and acceptance of risk, as well as threshold levels for perception of the intrusion of
privacy (Myers, 1962). Once identified, user personality type and needs can by leveraged
by designers to attract specific target markets increasing overall usage, improving
retention, and fine tuning advertisement placements which all increase service site value
and return on investment.
This study is carried out through two approaches. The primary approach of this research,
titled Modeling the Effects of Personality Type on Online Credit Card Payment Services,
seeks to answer a number of research questions. This primary approach addresses
whether different personality types as measured by the MBTI personality test affect
user perceptions of the usefulness, ease of use of online credit card payment services, and
the compatibility of the features and services provided by online credit card service
websites. To accomplish this the research utilizes several elements adopted from TAM.
These TAM elements include Perceived Usefulness (PU), Perceived Ease of Use (PEU),
and Usage (U). In addition to these metrics borrowed from TAM, the construct of
Perceived Compatibility (PC) is employed from the Diffusion of Innovations theory. The
research then examines whether those perceptions, as varied by personality type, have
any effect on the individuals intention to use online payment services and features.
To broaden and fill out this study, in a secondary approach titled, Discovering Desirable
and Unwanted Online Credit Card Payment Service Features for Different Personality
Types, this research presents a descriptive study of data collected independently of the
primary approach that examines how personality type, as classified by the MBTI,
distinguishes features that promote utilization of e-commerce payment services by


5
individuals and also which website features are considered as unimportant by different
personality types.


6
CHAPTER II
BACKGROUND AND RELATED WORK
This research is cross disciplinary and draws upon research from psychologys
fields of behavior science and personality types; information systems (IS) TAM and
work from both marketing and IS in the area of Diffusion of Innovation.
TAM
In IS, TAM (and its many variants) stands out as a well-documented and
thoroughly tested methodology. It is used to understand and predict the behavior to adopt
or the intention to use and actual usage of a given technology in the work place. TAM
has been used to predict consumer intentions in a number of specific online environments
(Lai & Li, 2005; Lederer, Maupin, Sena, & Zhuang, 2000).
This study makes use of the main constructs of TAM in its simplified form in
order to ascertain the relationship between personality type and the usage of online credit
card payment services. TAM finds that Perceived Usefulness (PU) and Perceived Ease of
Use (PEU) of a technology are the major determinants of a users Intent to Use (IU) and
Usage (U) of a given technology (Davis, 1989; Davis, Bagozzi, & Warshaw, 1989).
There have been a number of foundational studies that have found inclusion of TAMs
central constructs of PEU and PU was essential to understanding what draws users to
online mass customization websites, adding further weight to their inclusion here (Ajzen
& Fishbein, 1977; Lee & Chang, 2011).
IU, a behavioral intent, has been shown to be one of the most powerful predictors
of future behavior (Davis, 1989; Venkatesh & Davis, 2000; Venkatesh & Morris, 2000).


7
Future intention to use impacts current behavior and is incorporated within the U
measurement construct. This research attempts to describe a model of utilization
behavior in order to benefit online credit card payment services designers efforts to
retain and attract additional users. Determining the predictors of U can be put to use by
designers to tailor their work to include attributes attractive to those personality types
determined to have a higher probability of utilization (Davis, 1989; Davis, Bagozzi &
Warshaw, 1989).
TAM and its constructs of PU and PEU have been tested repeatedly using a wide
array of differing applications and information technologies (Lederer, Maupin, Sena, &
Zhuang, 2000). They have been examined as intermediate factors between various
external variables and the outcome of usage of a given technology (Legris, Ingham, &
Collerette, 2003). King and He carried out a meta-analysis of 88 different studies
testing the validity of the TAM model and found it to be generally a valid and robust
model" (King & He 2006, p. 1). They concluded that both PEU and U were reliable
measures across a wide spectrum of adoptions. They found in most cases that PU's effect
on U was highly significant and included most of PEU's influence on U. But when
focused on Internet applications found that PEU's direct effect on U was significant in
itself and outside of its relationship to PU. This suggested that when used in measuring
Internet U, models not incorporating the standard directional PEU to PU relationship
might display higher indices of fit than those models in which it is included (King & He,
2006). This study tested both models and found that including PEU's mediating
influence in the model did result in better fit statistics. Therefore we include the both the
direct PEU to U and PEU to PU relationships. Studies have further identified four


8
categories of modifications to the standard TAM model two of which are important to
this study (King & He, 2006). They are the common inclusion of prior factors, (in this
study MBTI) and the inclusion of constructs from other theories selected to increase
TAM's efficiency as applied to specific technologies, (in this case Compatibility from the
theory of Diffusion of Innovations).
The Extended TAM model (TAM2) was developed to explore and better
understand the antecedent determinants of PEU and PU in the business environment
(Venkatesh & Davis, 2000). They modeled influences of those constructs by social
influence and cognitive instrument processes in the work place. The social processes they
isolated are subjective norm, image, and voluntariness. The cognitive processes are job
relevance, output quality, and result demonstrability. Their model is focused directly on
business technologies. Other researchers as discussed below have found that in other
areas of technological application, better models may be formulated utilizing various
other antecedent determinants of PEU and PU, as well as additional constructs to stand
beside them.
Research positively supporting TAM has been conducted on the use and adoption
of the World Wide Web with an exploration into those specialized antecedent external
variables which predict PEU, and PU in this area (Lederer, Maupin, Sena, & Zhuang,
2000, Venkatesh & Davis, 2000).
A number of studies further confirm that TAMs central tenets are substantively
affected by exogenous variables giving credence to the validity of extensions of the TAM
model (Agarwal & Prasad, 1997; Fishbein & Ajzen, 1975; King & He, 2006;


Vijayasarathy, 2004). Table 1 shows some of this prior research and the independent
variables or constructs added as extensions of TAM.
9
Table 1. Some antecedent variables added to TAM in previous research
Domain New variables/constructs Reference
e-commerce trust and perceived risk (Lee & Chang, 2011; Pavlou, 2003)
online games social influence, attitude, and flow experience (Hsu & Lu, 2004)
online shopping compatibility, privacy, security, normative beliefs, and computer self- efficacy (Vijayasarathy, 2004)
mobile commerce Perceived risk, perceived costs, and compatibility (Chau, 1996,; Wu & Wang, 2005)
system administration Task-Technology Fit model (Dishaw & Strong, 1999)
Internet banking perceived credibility and financial and security risks (Wang, Wang, Lin, & Tang, 2003)
organizational behavior social influence and cognition (Venkatesh & Davis, 2000)
Thus we see a clearly demonstrable practice of extending TAM by both inclusion
of constructs standing side by side with PEU and PU, as well as extension by addition of
antecedent determinants of PEU and PU paralleling TAM 2's approach.
Diffusion Of Innovations
Choice of what metric to use in any given case is important and must fit that
which is being measured. Since credit card payment service usage is both a technological
application and a consumer service item it is felt that the appropriate metric tool blends
the measurement capabilities of both TAM and the theory of Diffusion of Innovations.


10
TAM's main use is to gauge adoption or usage of an IT Artifact by workers to accomplish
a business task (Venkatesh, Morris, Davis, & Davis, 2003). Online payment technologies
are adopted by individuals to fulfill a consumer service need rather than complete a
business process. In that context, the individual is making a decision which is motivated
not only by the attributes of usefulness, and ease of use, but by additional social and
cognitive personal pressures outside those of business environments. This study attempts
to incorporate these additional needs into the model by the use of extending TAM with
the inclusion of an element from the Theory of Diffusion of Innovations (Gefen,
Karahanna, & Staub, 2003). The Diffusion of Innovations Model is a framework for how
an innovation (a new idea, object, or practice) is spread through channels of
communication over time, throughout a population. According to the model the four
required elements for diffusion are: an innovation, a channel of communication, a time
period, and a social system (Rogers, 2003). The theory relates adoption behavior to the
following constructs: Compatibility of Technology, Complexity of Technology, Relative
Advantage, Trialability, and Observability.
The theory of Diffusion of Innovation has been applied to model the diffusion of
technology with good effect (Fisher & Pry, 1971). It has been also employed directly to
research online usage. In a particularly relevant example Parthasarathy and Bhattacheijee
(1998) used the Diffusion of Innovation theory as a theoretical framework to study the
continuance behavior of subscribers of Internet service providers. For the purpose of this
research, the use of technology is, by users, to fulfill a consumer service need. Parallels
can be drawn between some of the attribute constructs of TAM and those of the theory of
Diffusion of Innovation (See table 2). Ease of use is similar to complexity of


11
technology. Relative advantage relates to usefulness. Technology adoption is the
equivalent of intent to use. These three parallel attributes have been found to be
consistently valid and applicable to both diffusion and adoption in a number of empirical
studies (Fliegel & Kivlin, 1966; Ostlund, 1974; Parthasarathy & Bhattacherijee, 1998;
Tomatzky and Klein, 1982).
The only elements of the theory of Diffusion of Innovation not directly paralleled
in TAM are Compatibility of Technology, Trialability, and Observability. Online credit
card payment services and banking bill payment features allow users to access their
accounts and make use of payment services free of charge, from any location day or
night. By this unrestricted access the user is unencumbered in their ability to experiment
with usage of payment services and its observable outcomes, whether positive or
negative, making Trialability and Observability unneeded within the framework of online
credit card payment. Therefore of the Diffusion of Innovations' attributes only the
construct of Perceived Compatibility (PC) need be added to enhance the predictive
capability of this studys model beyond the TAM constructs of PEU and PU.


12
Table 2. Comparative constructs of DOI TAM & Credit Card Online Payments
Diffusion of Innovations Technology Acceptance Model Online Credit Card Payment
Complexity of Technology Perceived Ease of Use Perceived Ease of Use
Relative Advantage Perceived Usefulness Perceived Usefulness
Compatibility of Technology NPA* Perceived Compatibility
Technology Adoption Intent to Use Intent to Use
Trialability NPA* NR**
Observability NPA* NR**
* No parallel attribute * * Not rel evant to subj ect
Myers Briggs Type Indicator
Personality tests are instruments designed to ascertain various attributes of the
psychological profile and character of individuals. They fall in to two main categories:
those which test higher level "traits" and those that test lower level primary "types".
Some research considers traits as classifying the qualitative differences of individuals,
while types classify them quantitatively (Bernstein, Clarke-Stewart, & Roy, 2008). Trait
based tests such as the Big Five, NEO PI, or Circumplex Seven consider broad upper
level comprehensive traits (Digman, 1990). The MBTI, Kirton Adaptors Innovators,
Eysenck's Personality Indicator, and other type tests like them are based on Jung's work
founded on cognition which acts at a more primary psychological level (Jung, 1921,
1971; Kirton, 1976; Mershon & Gorsuch, 1988).
Cognition deals with how an individual acquires information and makes
decisions, and has been found to be more determinant in predicting behavioral intent and
therefore utilization than those higher level traits dealt with by the Big Five (Paunonen &
Ashton, 2001; Digman, 1990). Higher level traits have been found at times to be
variable due to mood or situational condition independent of any change in core


13
personality (Gendlin, 1964; Roberts & Walton, 2006; Rogers, 1957). Lower level core
based types due to their foundation in deeper seated cognitive values are longitudinally
stable, independent of mood or situation, and therefore more reliable and accurate
indicators of a user's behavioral intent (Fuchs, 2001; Jung, 1921, 1933, 1971; Mershon &
Gorsuch, 1988). The MBTI is used across a spectrum of scientific disciplines as a
psychometric survey administered to subjects to ascertain their psychological attributes
based on Jungs Psychological Types (Myers, 1962; Myers & McCauley, 1985). Jung
(1921, 1971) extrapolated eight psychological types based on permutations of two basic
functions, extraversion and introversion, and four attributes, sensation, intuition, thinking,
and feeling.
The MBTI employs four diametric scales (Bayne, 1995). The pairs are
Extraversion (E) and Introversion (I), Sensing (S) and Intuition (N), Thinking (T) and
Feeling (F), and Judgment (J) and Perception (P) (Myers, & McCauley, 1985). These are
usually treated categorically; that is for any and each one of the four diametric scales,
depending on the value of an individuals survey answers for that scale, the individual is
assigned either one or the other of the diametric poles as a category, giving rise to 16
possible states, or Personality Types.
The MBTI was chosen for this research as it is based on cognition with its better
predictive capability, is generally seen as reliable and valid, and is the most widely
accepted and recognized of the type tests used to measure cognition based on Jung's
concepts (Carlson, 1985; Wheeler, Hunton, & Bryant, 2004). Furthermore, the MBTI
testing forms have undergone extensive peer review and been repeatedly confirmed as
statistically accurate and internally consistent in identifying its' deeply theoretically based


14
personality types. It is readily available as a turnkey testing instrument without the added
burden to researchers of independent development and validation of a questionnaire.
Finally, the researcher has firsthand working knowledge and prior experience in usage of
the MBTI system.
Personality Type And The Internet
Previous research examining interaction effects between personality and website
usage typically examine a single trait or two and do not try to examine the whole
personality. One early example of research on personality and Internet usage examined
the relationship between extraversion and neuroticism, and Internet usage (Brenner,
1997; Eysenck, & Eysenck, 1975; Hamburger & Ben-Artzi, 2000; Sybil, Eysenck,
Eysenck, & Barrett, 1985). Another research study implied that extroverted versus
introverted personality types account for much of the variance of Internet leisure and
social services usage (Hamburger & Ben-Artzi, 2000); showing that males having
extraversion increased their use of leisure services, whereas for females, extraversion
decreased use of social services (Hamburger & Ben-Artzi, 2000). These early studies
form the basis of the theoretical background linking user psychology, personality and
Internet user behaviors.
A later study addressed user behavior in light of a particular personality trait
rather than a categorical personality type, focusing on the usage patterns of commercial
sites by individuals with varying levels of a need for closure (Amichai-Hamburger, Fine,
& Goldstein, 2004). This touches directly on a number of underlying aspects of both
personality traits and types.


15
The Five Factor Model based on trait theory is one of the major frameworks
currently used in the field of psychology to describe human personality traits (McCrae &
John, 1992). Tuten and Bosnjak (2001) make use of the Five Factor Model of
personality, compared with cognition, to examine web usage preferences. Their findings
indicated cognition a better overall predictor of web usage than any of the Five Factor
Model traits. They also found that extraversion had no effect on the type of web usage,
but did negatively affect the duration of use compared to introversion. This yields
theoretical support for the preference of utilizing both personality typing over trait
analysis and the usage of Jungian based models when studying web usage behaviors.
Guadagno, Okdie, and Eno (2008) also used the Five Factor Model to discover
that Openness to New Experience and Neuroticism were positive predictors of blogging.
Amichai-Hamburger (2002) and others; have provided motivation for further
investigation into user personality type effects on Internet usage by finding that web
designers tended to ignore the personality types of users when making design decisions to
the determent of the end product (Amichai-Hamburger et al, 2004; Benet-Martinez &
John, 1998). Fuchs (2001) conducted a case study of personality of Real Estate agents
through the use of the MBTI and followed up with personal interviews as confirmatory
analysis of the indicator type findings. The results were incorporated into guidelines used
to design a real estate agency e-commerce website. Fuchs (2001) makes reference to an
earlier study by Norman (1999) which supported the case that when users are given
systems that are not only functional, but also serve to satisfies the internal needs arising
from the users core personality, the usage of that system becomes intuitive, even
satisfying needs of which the user is not even consciously aware. That study concluded


16
that the usage of personality type indicators in website design facilitate development of
the conceptual model, identifying essential elements of the graphical interface, and help
in defining usability, and functionality of the web design.
As seen then personality profiling is becoming a more accepted methodology in
understanding web usage, for developing site design, and generally in business to
facilitate teamwork, customer relationship management, and other business applications
(Torres, 2013; Emergenetics, 2013). Personality profiling has reportedly been used by a
wide variety of organizations including: Bank of America, Blue Cross Blue Shield,
Cisco, Cricket, Denver International Airport, Great West Life, Hilton and Marriott hotels,
Microsoft, National Semiconductor (Malaysia), Pratt & Whitney (Canada), Sembawang
Logistics (Singapore), and the United States Air Force among others (Emergenetics,
2013). Utilizing personality type in e-commerce research has been shown to be
applicable yet its use is still limited meriting further attention.
Current Usage Of Online Credit Card Bill Payment Services
The main aim of this research is to find determinants to usage that can be
leveraged by designers, marketing, and other management, to improve adoption and
retention as discussed above, however, other issues have also been found that relate to the
relevance of this research.
It is felt that a consumer's payment of their monthly credit card bill is distinct and
different from other bill payments they may make for one major reason. Their credit card
bill payment history is directly and immediately tied to their credit score. Consumer
credit scores have increased in importance to the point where they affect nearly all facets
of a consumers' life. The time when a credit score was merely used as a reference of


17
credit worthiness is long gone. Now, credit scores can effect who will allow you to rent
from them, whether you will be employed by certain employers, what you pay for
insurance premiums and much more. If a customer misses a water or electric payment
they may have their service interrupted. This is an isolated event, and once the balance is
brought up to date rarely has any further effect. If however they miss their credit card
bill payment this can lower their credit score effectively economically handicapping
them, and influencing wide areas of their lives. This makes payment of credit card bills a
unique activity and payment behavior of those bills an extremely important issue.
Available statistics show that 13.8 percent of consumers paid some credit card
bills online (Hayashi, & Klee, 2003). When compared to the 62 percent of US consumers
who had Internet access in 2003, it is clear there is considerable room for expansion of
the number of adopters of online Credit Card bill payment (US CESRC, 2010). This is
an important area of finance since US consumers carry $835.5 billion in credit card debt
(Federal Reserve Bank of Boston, 2010). That total was distributed across 576.4 million
individual credit cards holders in the US (Nilson Staff, 2010). On average each adult in
the US had 3.5 credit cards issued in their name as of the beginning of 2009 (Federal
Reserve Bank of Boston, 2010).
In 2001, 19 percent of all US households reported using some type of online
banking and this value rose to 34 percent in 2004 and to 53 percent in 2007 (US Census
Bureau, 2011). The focus of this study is online payment of monthly credit card bills
which are primarily transacted with online BANP instruments. In 2008, the mean
number of bills paid monthly using online BANP instruments was 4.1 per consumer, but
fell to 3.0 bills paid monthly in 2009, a decline of 26.1 percent over one year (US Federal


18
Reserve System, 2011). During this same period 73.4 percent in 2008, and 56.3 percent
in 2009, of all US consumers had used an online BANP instrument in the payment of
some type of monthly bill including credit cards, a 17.0 percent decrease, attributed to a
decrease in consumer confidence of the security of personal information maintained by
online banking websites (US Federal Reserve System, 2011).
The decline in online payment instruments for all bill types discussed above
shows that users who previously adopted online payments, are not continuing use of this
service. This infers an underutilization of this type of payment service suggesting that a
large percentage of consumers have yet to adopt or readopt online payment instruments
for cyclical credit card bills. Online credit card payment service's design and the menu
of features they offer have been linked to both adoption and continuance behavior. As a
result the set of features offered by these service sites become highly important to the
website designer (Albert, Goes, & Gupta, 2004; Amichai-Hamburger, 2002). Personality
type provides designers with a powerful lens with which to determine design elements
and features to retain current users as well as attracting new adopters.
Scope And Contribution
Credit card debt has been shown to represent a vast amount of money (Federal
Reserve, 2010). Any contribution to understanding the online payment behavior of users
is therefore important and can be employed to help design sites that will attract more
potential adopters. As discussed above, data indicates that there is a considerable
segment of the consumer population that could be persuaded to adopt with subsequent
benefits of efficiency and profit accruing to both financial institutions and users.


19
A body of research exists on TAMs relationship to personality types, and general
Internet usage, service usage, user perception, usage patterns and other related topics as
discussed in the literature review section of this paper. This study extends and builds
upon that previous work in the area of TAM, personality types, and website usage into
the unstudied area of online credit card bill payment. This area represents a gap in the
research literature as indicated by the lack of papers addressing this specific topic, and as
evidenced by the deficit of statistical data existent in the financial, and governmental
consumer payment behavior databases.
For the first time data will be collected specifically indicating personality types
relationship to adopter and usage behavior of online credit card bill payment. In addition
it will contribute a body of data dealing directly with the adoption of online non-cash
payment instruments for cyclic credit card bill payment, which is currently lacking.
Additional it will aid in the identification of payment site service and feature preferences
which might be utilized in attraction of new and retention of current users.
A further contribution is found in that any significant findings will be available to
be applied by website designers to enhance current and future sites. User Centered
Design uses Personas, Scenarios, and User Cases as analytical tools to design websites
better matched to the needs, and behaviors of the user (Cooper, 1999). Personas are
created by carrying out field research of either a qualitative or quantitative nature, and
then applying the findings to create as accurate of a user model as possible (Brickey,
Walczak, & Burgess 2010; Pruitt, & Grudin, 2003). Personas have been shown to be a
beneficial tool for systems and site development, yet often provide too course a screen
designating all users into a very small number of prototypes (typically 3 or 4) (Dharwada,


20
Greenstein, Gramopadhye, & Davis, 2007; Mulder & Yaar, 2007; Pruitt & Adlin, 2006).
To capture more individual nuances in e-commerce perceptions and utilization, a finer
grained sieve is needed. This study is not a design methodology in itself but rather
carries out field research specific to credit card payment service adopters. Usage of any
findings about users personality type from this study will contribute to the UCD
designers ability to develop personas that are more accurate, and scenarios and user
cases that are more effective. Findings from the secondary research conducted help to
identify specific design and feature preferences to be employed by those designers.


21
CHAPTER III
APPROACH I: MODELING THE EFFECT OF PERSONALITY TYPE ON
ONLINE CREDIT CARD PAYMENT SERVICES
Conceptual Model
This research attempts to methodically codify personalitys effect on a specific IS
technologys adoption and use. This model extends TAM first by including the construct
of PC from the theory of Diffusion of Innovation (Rogers, 2003; Wu & Wang, 2005).
Second, it extends TAM further by postulating that user personality can be used as an
antecedent determinant of and having a prior and direct effect on user perceptions of ease
of use, usefulness, and compatibility of online credit card payment services. These
extensions are employed in place of TAM 2 extensions, as this research postulates that
the MBTI and compatibility constructs better model the social and cognitive aspects for
personal credit card payment behavior than TAM 2's model for business function
applications use. Therefore this research attempts to systematically look at how
variations of personality type affect the credit card holder users perceptions of online
payment services, and their use of them.
A large body of confirmatory work in TAM indicates the validity and
applicability of PU and PEU both generally and specifically to the adoption of online
technologies (Moon & Kim, 2001). Figure 1 depicts the model following the classical
form of TAM using the constructs of PU and PEU, while complying with the simplified
version as per Adams, Nelson, and Todd (1992), eliminating the construct of Attitude
Toward Using, and Intent to Use, and employing Usage as the dependent variable (Lu &


22
Gustafson, 1994; Venkatesh & Davis, 2000; Venkatesh & Morris, 2000; Venkatesh,
Morris, Davis, & Davis, 2003).
Figure 1: Conceptual Model
Research Hypotheses
Twelve Hypotheses have been derived. Work in the field suggests that differing
personality profiles have some association to general Internet usage and that further study
of personality type and Internet service usage is critical to the ongoing development and
improvement of Internet-based services (Amichai-Hamburger & Ben-Artzi, 2003,
Amichai-Hamburger, 2004, 41, Hamburger & Ben-Artzi, 2000). Additional research
shows that specific personality profiles influence the perceptions of users and that
Internet usage patterns differ individually related to content and services (Hills & Argyle,
2003). Research into the theory of Diffusion of Innovation indicates that there is


23
relevance to the individuals personality being a factor in adopter behavior (Rogers,
2003).
The MBTI Judging dimension describes individuals preferring self-regulated,
managed life styles. Seeking orderly systems to plan their lives and make decisions, they
are most comfortable with situations that achieve closure. Their life styles are typically
well structured, organized, and scheduled (Myers, 1962; Myers & McCauley, 1985).
Judgementals prefer to be settled, having external factors under control and in a
completed state. Their defining characteristics are; "systematic, scheduled, organized,
methodical, make long and short term plans, avoidance of last minute stresses" (Myers,
1998). Judgmental types prefer to know the what and how of things, compared to
the perceptive types that are mainly concerned with why (Myers & Myers, 1995).
Online direct payments are almost real time. The action is complete, unlike a
check in the mail taking days to arrive and longer to receive confirmation of clearance. It
is highly compatible with the need for being settled which Judging personality type
prefers, and they are provided with confirmation at the completion of the transaction
providing closure. This supports a perception that online payments are useful to this
type, which leads to the hypothesis:
HI a: The Judgment type will have a positive effect on consumers PU of online
credit card payment services.
Hlb: The Judgment type will have an indirect positive effect on consumers' U
through its' interaction with PU of online credit card payment services.
Users can make use of online payment services according to their own schedules
day or night. It can be extrapolated that for this reason Judgmental users would be more


24
likely to find paying their credit card bills online both easier to use and more compatible
with their life styles.
The following two hypotheses follow:
H2a: The Judgment type will have a positive effect on consumers PEU of credit
card payment services.
H2b: The Judgment type will have an indirect positive effect on consumers' U
through its' interaction with PEU of online credit card payment services.
H3a: The Judgment type will have a positive effect on consumers PC of online
credit card payment services.
H3b: The Judgment type will have an indirect positive effect on consumers' U
through its' interaction with PC of online credit card payment services.
Those falling on the extraverted side of the MBTI Extraversion/Introversion
scale are seen to be actively engaged in the outer world. Involved with activities and
people, they are energized by both. They are action takers, and willing to take risks at an
acceptable level, with their attention placed squarely on the world around them (Myers,
1962; Myers & McCauley, 1985). They are characterized as; "attuned to the
environment, prefer to communicate by talking (rather than writing), work out ideas by
talking them through, learn best by doing, have broad interests, sociable and expressive,
ready to take initiative in work and relations (Myers, 1998, pg. 9).
As seen in the literature on Internet Banking usage, rational judgments about user
trust in the credibility of a website as to security and risk reduction relate to user


25
perceptions and intent (Friedman, Kahn, & Howe, 2000; Hoffman, Novak, & Peralta,
1999; Wang, Wang, Lin, & Tang, 2003). The use of direct bank account transactions, as
any online financial transactions, has an associated perception of risk to the user (Lee &
Turban, 2001). It is postulated that extraverted types would take this in stride easier than
those less amenable to risk and less accustomed to taking action. They easily accept risk
on both physical and emotional levels as a natural expression of their unreserved nature.
Outgoing types with many varied interests are often leaders and opinion makers, tending
to be early adopters of ideas and technologies (Fisher & Pry, 1971; Rogers, 2003). It is
expected that the stronger the Extraversion scores the more likely will be the likelihood to
use online payment services, and the greater usefulness perceived. This is based on the
findings that extroverted individuals are more inclined to assume risk, and we postulate
that they will perceive fewer impediments to PEU (Bayne, 1994; Myers & McCauley,
1985). It is felt that these attributes, much like the characteristics of Judgment types
contribute to the likelihood that an individual would find online payment systems easier
to use and highly useful leading to H4 and H5.
H4a: The Extraversion type will have a positive effect on consumers' PU of
online credit card payment services.
H4b: The Extraversion type will have an indirect positive effect on consumers' U
through its' interaction with PU of online credit card payment services.
H5a: The Extraversion type will have a positive effect on consumers' PEU of
online credit card payment services.
H5b: The Extraversion type will have an indirect positive effect on consumers' U
through its' interaction with PEU of online credit card payment services.


26
While introverted users prefer less direct contact and communicate by writing, mailing
checks to pay credit card debt satisfies these desires with much less perceived risk
[Myers, 1998],
It is thought that the Extraversion factor of personality will become more
important as we see the fall off of trust in the security of online financial sites (Federal
Reserve Bank of Boston, 2010) and so H6 is postulated:
H6a: The Extraversion type will have a positive effect on consumers' PC and U
of online credit card payment services.
H6b: The Extraversion type will have an indirect positive effect on consumers' U
through its' interaction with PC of online credit card payment services.
Sensing types tend to rely on actualities, and thrive on exactitudes. They depend
on "what is" rather than on the "what is possible" that intuitive types focus on (Myers, &
Myers, 1995). Persons of the sensing type depend on their own five senses rather than
relying on what others may say or feel about external factors (Bayne, 1994; Myers, 1998;
Quenk, 2009). Intuitives being grounded in inspiration are more likely to be innovators
than Sensing types, but that also means they are more likely to be frivolous and change
ideas and methods more easily and often than their Sensing counter parts (Myers &
Myers, 1995). Sensing types typically prefer custom and tradition. This plays out in
establishing processes in their lives that are straight forward and provide positive
feedback as to results (Myers, 1962; Myers, 1995; Myers & McCauley, 1985; Myers &
Myers, 1998). This meshes well with online payment services that provide real time
feedback by way of instant confirmation of payment. This type would feel this method
both easier and more compatible than one in which the action is separated from the


27
confirmation by a longer stretch of time and they would find the instant feedback highly
useful. This leads to the following three hypotheses:
H7a: The Sensing type will have a positive effect on consumers' PU of online
credit card payment services.
H7b: The Sensing type will have an indirect positive effect on consumers' U
through its' interaction with PU of online credit card payment services.
H8a: The Sensing type will have a positive effect on consumers PEU of online
credit card payment services.
H8b: The Sensing type will have an indirect positive effect on consumers' U
through its' interaction with PEU of online credit card payment services.
H9a: The Sensing type will have a positive effect on consumers' PC of online
credit card payment services.
H9b: The Sensing type will have an indirect positive effect on consumers' U
through its' interaction with PC of online credit card payment services.
The Thinking/Feeling dimension deals with the making of decisions. Thinking types
approach the decision making process from the point of view of what is true versus what
is false. For the Thinker type everything is about impersonal objective truth, while to the
Feeling type everything is personal. Thinking type individuals will associate online
systems with their logical, computer-based underpinnings, and therefore feel that using
online systems for paying their credit cards is more logical and precise and therefore
more useful. And thus:
HlOa: The Thinking type will have a positive effect on consumers' PU of online
credit card payment services.


28
HI Ob: The Thinking type will have an indirect positive effect on consumers' U
through its' interaction with PU of online credit card payment services.
The Feeling type person approaches the process from the view point of what is agreeable
versus disagreeable (Myers, 1962; Myers & Myers, 1995). What is perceived as
disagreeable is harder to find of use. The Feeling type may be more likely to prefer the
more personal process of filling out the bill, writing the check by hand, and licking to seal
envelopes, or paying their credit card in person face to face at local bank branch (Myers,
1998; Myers & McCauley, 1985; Myers & Myers, 1995). They should find the online
system much less compatible with their life style when compared to Thinking types who
would be very comfortable making payments in the online environment. By nature
therefore Thinking types will find online payments easier to use than their Feeling
counterparts. Hence, the final two hypotheses:
HI la: The Thinking type will have a positive effect on consumers' PEU of online
credit card payment services.
HI lb: The Thinking type will have an indirect positive effect on consumers' U
through its' interaction with PEU of online credit card payment services.
H12a: The Thinking type will have a positive effect on consumers' PC of online
credit card payment services.
HI2b: The Thinking type will have an indirect positive effect on consumers' U
through its' interaction with PC of online credit card payment services.


29
Methodology
Data was gathered utilizing a self-administered online survey hosted on the
Survey Monkey website. Data was actively collected starting in September of 2011 and
terminating in June of 2013. The survey was offered to perspective respondents as part of
the University of Colorado Denver Business School's "Business Engagement Assignment
Program". This is a program run by Professor Stefanie Johnson of the business school to
encourage student participation in academic and research activities outside of their
enrolled classes. Students enrolled in undergraduate programs are required to participate
in one activity over the course of each semester. They select their activity from a list of
prospective surveys and other research projects. Upon completion they are given credit
for an assignment by their respective instructors per the class's syllabus. Four hundred
thirty six students responded to the survey administered online. No screening of the
respondents took place and no identifiable information was recorded. It has been found
that TAM is invariant for differences of gender, age, and IT efficacy, and therefore has a
broader applicability than the sampling frame work would imply if TAM were not in use
(Lai & Li, 2005). Demographics for the respondents are shown in Table 3.
Table 3. Respondent demographics
Category (mean, range if applicable)
Sex Male 49.8% Female 50.2%
Age (18-61) (Mean age = 28) 18-29 70% 30-39 16.5% 40 or Older 13.5%
Education (Max Completed) HS 39.3% AS/AA- 10% BS/BA 35.2% Graduate degree 15.5%
Employment Unemployed 23% Part-time 36.5% Full-time 40.5%
Marital status Single 62.4 Married 29.7% Prev. married 7.9%
Income ($45,699, $0- $100,000+) < $30K 53.7% S31-59K 26.2% >$59K 20.1%


30
The survey is in two sections. Section A, composed of 19 questions focuses on
collecting data for the four variables; PU, PEU, and PC. These questions are based on a
summated rating Likert Scale, employing a seven point agreement type configuration.
This is adapted from the field of psychology where this type scale is commonly used to
measure attitudes and intentions which closely fits the measurement constructs our study
is testing. (Blumberg, Cooper & Shindler, 2008; Likert, 1932; Spector, 1992). Section
B ascertains the personality type of the subject. It is comprised of the standard MBTI
Form M self-scorable test. Structural Equation Modeling (SEM) is employed as the
primary analytical tool due to the exploratory nature of this research, and as it is
particularly well suited to analyze the type of data in use, as constructs are measured by
multiple indicator values (Hair et al, 2005). SEM allows evaluation of a flexible set of
variables including continuous and discrete variables, and allowing use of non-directly
measured variables (latent variables) (Hair et al, 2005). The SEM model for the research
is shown in Figure 2 with an explanation of the value ranges for the corresponding
MBTI variables displayed in Table 4. As seen in the SEM Model Diagram (Figure 2)
this research employs MBTI dimensional variables (treated as direct variables), the
latent variables PU, PEU, PC, and dependent variable IU.
As an additional line of analysis of the data a post hoc parallel analysis has been
carried out using the Partial Least Squares (PLS) methodology. That analysis is
discussed in a separate section below.


31
Figure 2: Structural Equation Model
Table 4: Myers Briggs Dichotomy Variable Conversion Values
Dichotomy Axis Bi-categorical Values
Extraversion (E) Introversion (I) E= -1 to <0 1= 0 to +1
Sensing (S) Intuition (N) S= -1 to <0 N=0 to +1
Thinking (T) Feeling (F) F= -1 to <0 1=0 to +1
Judgment (J) Perception (P) J= -1 to <0 P=0 to +1


32
CHAPTER IV
ANALYSIS
SEM Model Analysis
This study is an exploratory investigation of an area of user behavior in which no
known causal model currently exists. The model considered in this research is used to
test the hypothesized directional correlation relationships between the models' latent
variables, comprising the model's constructs and are all reflective causal structures
(Edwards & Bagozzi, 2000). As path analysis tests directional relationships only
between manifest variables, and confirmatory factor analysis tests non-directional
relationships between latent variables, neither of those sub-types of SEM analysis were
utilized (Shah & Goldstein, 2006). Both absolute and incremental fit indices were
considered in determining the fit of the models. The Maximum Likelihood ratio (ML)
was used as the estimation method for all models.
Assumptions of the Structural Model
The studies structural model specifies the following relationships: MBTI>PC;
MBTI>PU; and MBTIPEU. The four exogenous variables El, SN, FT, and JP
represent the MBTI direct measurements for the individual's personality type. This
personality type exists as the core psychological makeup of an individual prior to any PU,
PEU, or PC attitudes they may develop upon taking up the usage of technologies.
Therefore there is temporal precedence evident.
As discussed in the introduction and background sections above there is a body of work
establishing an association between personality and behavioral intent, and usage of both
technology in general, and specifically web based applications and services. This serves


33
in fact as the theoretical underpinnings of this study as addressed in the Personality Type
and the Internet section of Chapter II. Thus the assumption of association is met.
Normality of the data is generally discussed below in the Normality section below. After
the application of noted transformation algorithm acceptable normality was met.
The specification of the casual relationship between the exogenous variables MBTI
and the endogenous constructs was discussed above. As to the meeting correctness
assumptions of specification of directional causality of the endogenous variables to
endogenous variables PC>U, PEU>PU, PEU>U, and PU>U, this study relies on
the sound theoretical foundation of TAM and DOI.
Assumptions Of The Reflective Measurement Model
This study's model employees four reflective constructs, PC, PU, PEU, and U.
This is theoretically justified in that each of these are endogenous variables and represent
a hypothetical construct in the conceptual model with its respective error term
represented in the SEM. Further each of these constructs are latent variables that exist
independent of any measurement used. Variation in the latent constructs would cause the
indicator items to vary. Any variation in the measurement instrument would not result in
variation of the construct, and the measurement indicators are caused by and
manifestations of the construct. All the measurement indicators share a common theme
and are interchangeable. Any addition or deletion of any particular measurement
indicator would not change the concept of the construct.
The empirical justifications for reflectivity as indicated by the immediate following
sections are that all the measurement indicators are highly and positively correlated (see


34
Chronbach's alpha values in Table 5), exhibit reliability and internal consistency, and
show theoretical content validity through convergent and discriminant metrics.
Reliability Of The Constructs
To determine the fit of the measurement model, the data for the each of the four
constructs were tested for reliability using both an internal consistency methodology
Chronbach's alpha, and a split-half reliability methodology the Spearman-Brown
coefficient. Cronbach's alpha was evaluated using a value of .70 as the commonly
acceptable lower limit (Hair et al, 2005). All four constructs exceeded the minimum
value for the Cronbachs alpha reliability test, as displayed in Table 5. The Spearman-
Brown coefficient heuristically has a lower limit of .6 as a cut off for exploratory work, .8
and higher is considered to show adequate reliability for confirmatory research, and .9 or
higher is considered very good reliability (Garson, 2001). The constructs: PEU, PC, and
Usage all scored above .8. The lowest Spearman-Brown Coefficient was .767 for PU,
which is well above the minimum of .6 used in exploratory work, and close to an
adequate reliability of .8.


35
Table 5. Reliability Indices
Construct Chronbach's alpha Spearman-Brown Coefficient Equal Length/Unequal length Number of Observed variables in Construct
Perceived Usefulness .834 .767/773 5
Perceived Ease of Use .835 ,835/.835 4
Perceived Compatibility .900 .887/891 5
Usage .881 .883/ 886 5
Normality Of The Constructs
The constructs were tested for normality of distribution. All four constructs were
found to have a negative skew, and to exhibit some kurtosis. To deal with this a
commonly used data transformation was employed to correct for negative skewness of J-
shaped distributions. The transformation algorithm (Algorithm 1) used was reflective,
and used a compensation factor for reversion to original ranking.
New X = Logio(8-X)*-1 (1)
After transformation all constructs fell within the commonly accepted limits of -1
to +1 for skew and kurtosis values for normality as displayed in Table 6 in (Garson,
2012, Tabachnick & Fidell, 2007).


36
Table 6. Assessment of normality utilizing algorithm (Logl0(8-X)*-1)
Variable Skew Ku rtosis
EU1 -.70895 -.54547
EU2 .02774 -.75854
EU3 .06149 -.63650
EU4 .-08151 -.85419
PU1 -.85301 -.37135
PU2 -.49970 -.92798
PU3 -.58177 -.90982
PU4 .25647 -.98315
PU5 -.43470 -.98760
PCI -.51872 -.71473
PC2 -.34460 -1.0272
PC3 -.76498 -.58538
PC4 -.52964 -.72693
PC5 -.34276 -.84655
Ul -.54601 -.82954
U2 -.99615 -.14084
U3 -.76750 -.78295
U4 -.80998 -.45958
U5 -.69980 -.62146


37
Analysis Of Model Fit
Using SPSS's AMOS application, a series of SEM models were developed and
tested. The optimal model, which is reported in this article, achieved an absolute model
fit measured with a Chi-square of 1236.75 with 197 degrees of freedom (DF).
Comparing the Chi-square to the DF, a CMIN/DF ratio of 6.28 is derived as
shown in Table 7. This value exceeds the upper limits of the range but is not uncommon
for a model of this complexity (Carmines & Mclver, 1981). The Chi-square at a
probability level equal to .000 is uninformative and thus with the high CMIN/DF ratio
leads to the necessity for consideration of the relative fit indices for establishing the
models fit (Browne & Cudek, 1993).
Incremental fit measured by both the Relative Fit Index (RFI) of .758, and the
Comparative Fit Index (CFI) of .819 is acceptable (Bentler, 1990). The model's root
mean square error of approximation (RMSEA) of .110 at the upper limits of reasonable
fit, is a minimally acceptable value for exploratory research (Browne & Cudek, 1993).
The last indicators of fit employed are the Akaike information criterion (AIC), and the
Browne-Cudeck criterion (BCC), measured at 1392.75 and 1401.46 respectively (Akaike,
1973, Browne & Cudek, 1989). These two values are moderately high, (highest of any
model tested) and so viewed as respectable given the high complexity of the model
involved (Browne & Cudek, 1989).
Table 7. Measures of Fit
Model CMIN DF CMIN/DF RFI CFI RMSEA AIC
Default 1236.751 197 6.277 .758 .819 .110 1392.75


38
R2 And Squared Multiple Correlations
2
The estimates for R interpreted in SEM as Squared Multiple Correlation (SMC),
are displayed in Table 8. For a latent dependent variable in a recursive model R or SMC
represents the variance in the model explained by all predictor variables by which it is
directly affected (incoming arrows in the SEM diagram) (Byrne, 2010; Garson, 2012).
The value for El in this model indicates that 67.5% of its variance is explained by its
predictor variables PU, PC, and PEE1.
For indicator variables of an SEM model SMC provides an estimate of
communality. Communality is a measure of the percent of variance for an indicator
explained by the latent variable. Communality can be understood as an additional
measure of the reliability of an indicator (Garson, 2012). For instance PC2, one of the
indicators of PEE1 has an SMC of .760. Therefore, PEE12 is said to be 76.1% reliable as
an indicator of PEE1. This is the strongest SMC of any of the indicator variables. The
weakest indicators in our model are PU4 at .229 and U1 at .255. In the final round of
analysis the SMC of these indicators would be considered along with measures of
convergent validity, and importance of the variable to the theoretical model for possible
elimination from the SEM. Of these the closest scrutiny would be given to PU4 and U1
as they have little more than 20% reliability.


39
Table 8. Squared Multiple Correlations (R2)
Estimate
U .675
PEU1 .373
PEU2 .760
PEU3 .708
PEU4 .491
PU1 .755
PU2 .744
PU3 .535
PU4 .229
PU5 .359
PCI .716
PC2 .760
PC3 .450
PC4 .701
PC5 .655
III .255
U2 .719
U3 .429
U4 .748
U5 .644
Convergent Validity
To determine convergent validity each of the four constructs' factor loadings were
examined and the Average Variance Extracted (AVE) assessed. The commonly accepted
lower limit for factor loadings is considered to be .5 or higher and in the best case .7 or
higher, with AVE values expected to be .5 or higher to indicate sufficient convergent
validity (Byrne, 2010; Garson, 2012). All of the models constructs exceed the loading
minimum except PU4 with a loading of .48 with Ell at .51 the next lowest. (See Table 9).
These two indicators were already under scrutiny due to their low SCM values. U1 met
the minimum but with PE14 failing another measure it was decided to eliminate this
indicator from the model. It was determined to keep U1 even with its low SMC as it met
the minimum for convergence and yielded a better RMSEA value when left in the model.


40
Overall each of the constructs AVE values exceeded .5 further indicating adequate
convergent validity (See Table 9).
Table 9. Convergent Validity Construct Factor Loadings
Factor Construct
Loading AVE
PEU 1 .610
PEU 2 .872
PEU 3 .841
PEU 4 .701
PU1 .869
PU 2 .862
PU 3 .732
PU 4 .482
PU 5 .599
PCI .846
PC 2 .872
PC 3 .671
PC 4 .838
PC 5 .809
U1 .505
U 2 .848
U 3 .655
U 4 .865
U 5 .802
PEU .668
PU .687
PC .721
u .689
Discriminant Validity
To determine Discriminant Validity this study uses the common approach of a
Squared Correlations Estimate Matrix with AVE on the diagonal. The AVE value is
higher for each construct compared to the square of the inter-construct correlation
relationship values (See table 10) (Byrne, 1995; Garson, 2012). This indicates a positive
divergence for the constructs.


41
Table 10. Squared Correlations Estimate with AYE on diagonal
PEU PU PC U
PEU .668
PU .328 .687
PC .460 .635 .721
u .374 .538 .646 .689
Structural Measurement Analysis
The main relationships being evaluated in this research are first, those between
the individual MBTI manifest variables (El, SN, FT, and JP), and the latent variables
PEU, PU, and PC, and second, the latent variables PEU, PU, and PC's relationships to the
independent latent variable U, and third PEU's mediation effect on PU's relationship to U.
The values of the MBTI variables have been converted to a scale centralized on
a value of 0. Any negative values for regression weights indicate values for the left hand
member of the MBTI pairing, and positives indicate values for the right hand members.
The four manifest MBTI variables are standardized at .00 intercept, and when
unstandardized exhibit the intercepts displayed in Table 11 as they relate to the
intermediate constructs. The beta coefficients/regression weights for each path
(displayed in Table 12) measure the relationships of the SEM model. These SEM path
coefficients/regression weights (Standardized Total Effects) are used as approximations
for the relative connection strength between the MBTI variables and the TAM and
Diffusion of Innovation Theory constructs. For the PU construct the E, T, S, and J poles
have the strongest to weakest relative effect respectively on the PU construct, indicating
that the overall MBTI type which perceives the greatest usefulness from using a credit
card payment service is the ESTJ type.
The MBTI pole relations for PC listed by decreasing magnitude are then T, E, J,
S, indicating that the overall MBTI which perceives the greatest compatibility with


42
using a credit card payment service is again the ESTJ type. The MBTI pole relations
for PEU listed by decreasing magnitude are S, J, E, T, indicating that the overall MBTI
which perceives the greatest ease of use when using a credit card payment service is also
the ESTJ type.
The strongest of any of the observed MBTI manifest variables' relationships to
the latent TAM construct variables is the effect of S on PEU with a regression weight of -
.083, followed by T on PC with a weight of .078, E on PC with a weight of -.071, and E
on PEU again with a weight of -.071. The next strongest are J to PC with a weight of -
.0652, and E to PU and T to PEU with weights of -.065 and .052 respectively.
The next relationships evaluated are those between the latent TAM constructs PU,
and PEU and the Diffusion of Innovation Theory construct PC, to the dependent latent
variable U. These are the endogenous to endogenous relationships measured by the beta
coefficients and are all reported using standardized weights. The strongest of these is PC
to U with a value of .738, and is significant with a P value of less than .001. That is
followed by PU to IU at .285, significant at P less than .001, and the weakest PEU to U at
.098, significant with P value of .031. The relationship between PEU and PU is .536,
also with significance at P value of less than .001. The effect of PC on U is roughly 2.6
times greater than PU's effect on U, and 7.5 times greater than PEU's effect on U. The
PEU to PU relationship is strong with a value of .536.
The relationships analyzed to this point are all direct relationships. Table 12
shows the values for the indirect relationships between the observed MBTI variables
and the dependent latent U variable. The El to U value indicates the E pole factor is
dominant over the other MBTI type factors effect on U. The JP to U and SN to U


43
estimates both show the significant effect of the S and J types on U. Overall the greatest
indirect effect on the variance of U of any combined MBTI type is that of ESTJ.
Tablell. Regression intercepts of MBTI
MBTI pair Standardized Intercept Unstandardized Intercept
El .00 -.12
SN .00 -.05
FT .00 .15
JP .00 -.17
Table 12. Variable to Variable Path Beta Coefficients/Regression Weights
JP FT El SN PEU PU PC U
PEU -.023* .052* -.071* -.083*
PU -.047* .049* -.065* -.047* .536**
PC -.065* .078* -.071* -.035*
U -.064* .077* -.078* -.047* .098* .285** .738**
* significance = <.05 ** significance = <.001
Mediation Testing
The model is a recursive causational model, which includes one mediating
variable which is PEU's to PU's relationship. The model specifies and expects to see no
other mediation. To test PEU's mediation effect on U and to confirm that no other
variables are acting as mediators a series of tests were run. These included the Sobel,
Aroian, and Goodman tests. The results confirm PEU to PU mediation to a significance
of better than .05 for all three tests. All other relationships tested showed non-significant
P values confirming that no other variables are behaving as mediators (See table 13).


44
Table 13. Mediation Test Statistics
Relationship Sobel Test P Value Aroian Test P Value Goodman Test P Value
El>PC>IU -0.04900 0.08400 0.19800 0.00800 -0.58317 0.55945735
SN>PC>IU -0.11800 0.09100 0.19800 0.00800 -1.29493 0.19498146
FT>PC>IU 0.11900 0.09300 0.19700 0.00800 1.27785 0.20093297
JP->PC~>IU -0.11400 0.08600 0.19700 0.00800 -1.32366 0.18525256
El>PU>IU -0.05200 0.07400 0.16500 0.01200 -0.70179 0.48165367
SN>PU>IU -0.11000 0.09100 0.17900 0.00900 -1.20656 0.22701373
FT>PU>IU 0.05700 0.09200 0.17800 0.00900 0.61926 0.5352224
JP->PU~>IU -0.10700 0.08500 0.17800 0.00900 -1.25628 0.20843377
El>PEU>IU -0.05200 0.07400 0.16500 0.01200 -0.70179 0.48165367
SN>PEU>IU -0.13800 0.08000 0.16500 0.01200 -1.71158 0.08615156
FT>PEU>IU 0.12400 0.08200 0.16400 0.01200 1.50302 0.13180731
JP>PEU>IU -0.08100 0.07600 0.16400 0.01200 -1.06256 0.28669408
PEU>PU>IU 0.59400 0.04600 0.14100 0.01000 9.52292 0.00000
Additional Variables Tested
Several additional models were tested adding Age, Gender, and Income to the
SEM. In the best case the addition of these variables showed a very slight improvement
in model fit moving RMSEA from 110 to 105. However, in that case all beta
coefficients were low and non-significant in all path relationships. The strongest were PC
to Gender and PU to Gender at -.032 and -.033 respectively (See Appendix B). Therefore
Age and Income were dismissed from the model, while Gender exhibiting two of the
stronger relationships of the additional variables was subjected to a further test to finalize
determination of variance insensitivity to that variable.
Testing For Variance Due To Gender
As previously mentioned TAM has been found to be invariant for differences of
gender (Lai & Li, 2005). To confirm this for the sake of this study, and determine
whether Gender should be kept in the model or dismissed, an Independent Samples T-test


45
was run to determine if there is was any difference in the variance of the dependent
variable U due to gender. The Male group had a mean of 5.88 (see Group Statistics
Table Appendix II), with a Std. Error of .0787. The Female group had a mean 6.03, with
std. error of .0796. The significance for equal variances assumed was 125 (see
Independent Sample Table Appendix B), and being greater than the minimum of .05 it is
found that there is no significant difference in the variability of the two genders.
Additionally the sig (2-Tailed) value was .172, again greater than .05, showing no
significant difference in means for the two gender conditions (Tabachnick & Fidell,
2007). Those findings therefore imply that influences on U due to gender cannot be
shown to be significant.
PLS Model Analysis
Although we regard this study as exploratory it is felt the usage of the TAM and
DOI meet the criteria of SEM's requirement of being theoretically well founded, which it
is argued is the underlying reason SEM is considered applicable for confirmatory rather
than exploratory research (Vinzi, Chin, Henseler, & Wang, 2010). In addition the data
responded well to the normalizing algorithm applied. This study therefore has made use
of covariance based SEM as its primary analysis tool. However there has been
discussion of the possible merits of employing the Partial Least Squares (PLS) method in
place of or in addition to the SEM approach. In general PLS is considered applicable
when the study is of an exploratory nature, you have formative constructs rather than or
along with reflective constructs, the sample size is small (less than 10 to 20 times the
number of dependent variables), and when data does not exhibit normal distribution
(Gefen, Straub, & Boudreau,2000 ; Vinzi, Chin, Henseler, & Wang, 2010 ). This study


46
is considered to be exploratory, and prior to application of normalizing algorithms is non-
normally distributed, all the constructs are reflective, and the sample exceeds the minimal
SEM requirements. In a sense then this study falls somewhere between the two methods.
Since this is the case to be thorough a PLS analysis has been carried out.
The results of the PLS analysis is reported in two steps. These are the
measurement model dealing with reliability and validity of the measurement constructs,
and the structural model dealing with the manifest and latent variables relationships.
PLS Measurement Model
Smart PLS was used as the PLS application software to carry out the analysis
(Ringle, Wende, & Will, 2005). The primary model employed parallels closely the SEM
model and is show in figure 3. As indicated it is composed of the four MBTI type
indicators as directly measured variables, and the research models constructs PC, PU,
PEU and with the dependent construct U. It contains all the same relationship paths as
the SEM model.


47
Figure 3: PLS Model diagram
PLS Reliability
The primary test for reliability in PLS is considered to be composite reliability
with an acceptable lower limit of .70 (Fomell & Larker, 1981; Vinzi et al, 2010).
Composite reliability met or exceeded the .70 minimum for all the constructs. The
loading for each factor was examined as well and all were found to exceed its minimum
of .50 (Hulland, 1999). As a final confirmation of reliability Cronbach's alpha was
considered, and all constructs were found to exceed the minimum standard level of .7
(Fomell & Larker, 1981; Vinzi et al, 2010). Table 14 displays the composite reliability,
factor loading and Cronbach's alpha for the constructs.


48
Table 14. PLS Reliability Tests values
Construct or Factor Composite Reliability Factor Loading Chronbach's Alpha
U .897 .855
U1 .697
U2 .902
U3 .714
U4 .870
U5 .855
PC .909 .873
PCI .858
PC2 .887
PC3 .641
PC4 .848
PC5 .834
PU .883 .823
PU1 .871
PU2 .854
PU3 .824
PU5 .680
EU .869 .803
EU1 .767
EU2 .844
EU3 .839
EU4 .708
PLS Validity
Convergent validity is determined by the AVE for a given construct. It is
required to be greater than .50 (Vinzi et al, 2010). As seen in Tablel5 all constructs meet
this requirement ranging from the lowest U at .641 to the highest PC at .669.
Discriminant validity is shown on Table 15 where in the AVE for any given
construct is higher than the squared correlation estimate for the related constructs. The
constructs all meet this requirement (Vinzi et al, 2010).
Table 15. PLS Squared Correlations Estimate with AVE on Diagonal
PEU PU PC U
PEU .626
PU .311 .657
PC .424 .594 .669
u .342 .473 .634 .641


49
PLS Structural Model
The Structural Model analyzes the relative strength of the statistical relationships
between the variables. These are output as coefficients measuring the relationship as
indicated by a path connection in the model. The coefficients for the models inner
constructs are shown in Table 16.
Table 16. PLS Path Coefficients
Path Coefficient (T-Value) R2
PC ~> U .601 (7.84*) .654
PU ~> U .170 (2.54*)
PEU ~> U .089 (1.93*)
PEU ~> PU .558 (13.24*)
* = > significant at .05
The strongest relationship is PC to U at .601 showing PC's high relevance to
usage of online payments. Next highest is PEU to PU at .558 showing PEU's strong
influence on PEI. Looking at the R for U of .654 indicates that 65.4 percent of the
variance is explained by its predictor variables PU, PC, and PEU.
As seen earlier in the discussion of the SEM analysis the MBTI variables are
based on a centralized value of zero. Used in this way any coefficient with a negative
value indicates the relationship is effected by the left hand member of the MBTI
dimension pairing and a positive value indicates the right hand member of the pairing.
These coefficients are displayed in Table 17 along with the corresponding MBTI
dimension pair derived for each of the model constructs. Table 18 displays an overall
MBTI type as derived from the analysis.


50
Table 17. PLS MBTI Coefficients and Derived Pair Indicator
MBTI Dimension Coefficients (MBTI Derived Pair Member)
PC PU PEU U
El -.057 (E) -.011 (E) -.060 (E) -.048 (E)
SN -.035 (S) -.004 (S) -.072 (S) -.035 (S)
FT .047 (T) -.019(F) .062 (T) .037 (T)
JP -.050 (J) -.040 (J) -.013 (J) -.039 (J)
Table 18. PLS Derived MBTI Types Per Constructs
PC PU PEU U
MBTI Type ESTJ ESFJ ESTJ ESTJ
The PLS output related to the MBTI coefficients closely parallels the findings from
SEM. The SEM Analysis showed a homogeneous result ESTJ for all of the constructs.
The only difference being the results for the PU Construct with PLS indicating a derived
type of ESFJ. This still does not change the overall type outcome for the dependent U,
again ESTJ which coincides with the SEM analysis.
The PLS's analysis having found that the FT>PU relationship is an F value
rather than the T value arrived at by the SEM could be accounted for in several ways.
First its coefficient was one of the smallest values arrived at, indicating that on the
measurement survey its combined responses would have clustered around the zero mark.
The PLS model was run using non-transformed data while the SEM models data was
subjected to the smoothing effects of a transformational algorithm. With the original
metric hovering near zero the smoothing effect for this one variable may have been just
enough to push it over into the F dimension. As was suggested this may be noise from
the data; we may just be seeing this smoothing effect in action. Also, the MBTI
system is dynamic such that the combined type is considered greater than the sum of its'
parts. There are functions considered to be dominant and inferior. Due to the dynamic


51
nature an individual can fluctuate between their dominant and inferior characteristics. In
this case we see the SEM yielding an ESTJ type which is Extraverted Thinking with
Introverted Sensing. The PLS yields an ESFJ type which is Extraverted Feeling with
Introverted Sensing. In both cases the Extraverted dimension is the active dimension of
the dominant function, with the Sensing being the active function of the inferior function
(Myers, 1962; Myers & McCauley, 1985; Quenk, 2009). In other words the ESTJ and
ESFJ behaviors are mostly influenced by the E and S dimensions. As such it is thought
that with these affecting only the PU dimension and the limitations of the differences
there is little net difference in the overall model.
Discussion
Of the three intermediary latent constructs represented in the model, PC was
found to have the greatest positive effect on the U construct. PU has the next greatest
effect on U followed by PEU with the least relative effect on U under the constraints of
this model.
From the negative or positive value of the beta coefficients for the SEM paths
between the exogenous MBTI variables and the endogenous construct variables as seen
in Table 12 it is determined that tentatively all 12 hypotheses are confirmed to some
extent. Out of that twelve, eleven can be grouped into 3 divisions based on the absolute
values of their corresponding SEM path beta coefficients. The first five hypotheses had
corresponding absolute path values above .060. They are H3a & b, HI5a & b, H6a & b,
H8a & b, and H12a & b which are considered to be strongly confirmed hypotheses. The
next three hypotheses all have corresponding absolute path values between .030 and .059.
They are HI a & b, H9a & b, and HI la & b which are considered to be confirmed


52
hypotheses. The final three all have corresponding absolute path values of less than
.030. They are H2a & b, H4a & b, and H7a & b which are considered to be weakly
confirmed hypotheses.
The parallel values for SEM's path beta coefficients are the PLS model's
coefficients seen in Table 17. These values confirm the SEM findings for all paths
except PEI to FT. This value found that F was the operative MBTI pole for this path
compared to SEM's finding of the T pole. This path relates to HlOa & b. While PLS
shows that the MBTI relationship FT to PU as F, it also shows that the relationship to U
as T, as does SEM. Due to these disparities HlOa is considered only partially confirmed,
as only U is positively affected by the T type and so HI Ob is confirmed.
The analysis clearly shows one of the sixteen MBTI personality types emerging
as an antecedent determinant to the model. First the study has found that the MBTI
type ESTJ has the greatest direct effect on the model constructs PC, and PEU, as well as
indirectly on the dependent U of all the MBTI types as seen in Figure 4. ESTJ and
ESFJ had the greatest effects over all on PU as indicated by SEM and PLS respectively.
In the Myers-Briggs psychology literature ESTJ stands for an Extraverted Thinking
supported by Sensing type. The Extraverted Thinkers (the dominant function of the
ESTJ's), both ES and EN types, tend to be analytical and impersonal. They are logical,
decisive and organized. When Extraverted Thinkers are of the Sensing type they are also
very practical, down to earth, and matter of fact. These are individuals that make
decisions using logical thought processes based on their own five senses. They are
described using the following terms; logical; analytical; objectively critical; decisive;


53
clear; and assertive. Typically they are focused on the present, on what is real and actual
(Myers, 1998). ESTJ's are action and task oriented, preferring reliable procedures and
systems (Bayne, 1995; Myers, 1962; Myers, 1998; Myers & McCauley, 1985; Myers &
Myers, 1995). This would seem to fit well with the theoretical underpinnings discussed
previously and the qualities inherent in online payment services.
MBTI Dimension Indirect Effect on IU
Figure 4: SEM MBTI Dimension Indirect Effect on U


54
MBTI Dimension Direct Effect on
PC, PU, & PEU
-.065 PC/-.047 PU/ -.023 PEU
Figure 5: SEM MBTI Dimension Direct Effect on PC, PU, & PEU


55
CHAPTER V
APPROACH II: DISCOVERING DESIRABLE AND UNWANTED ONLINE
CREDIT CARD PAYMENT SERVICE FEATURES FOR DIFFERENT
PERSONALITY TYPES
The secondary approach conducted for this research is an exploratory descriptive
study which looks at features and services preferences among users and non-users of
online credit card payment services related to MBTI Type.
In addition to the primary research methodology a second line of research was
conducted. This is an initial study to focus in on cogent features and services that look
promising for future in depth research.
The primary work looked a Personality Types' effect on usage of online credit
card bill payment services. It is thought that if a determinative link is established website
designers and service managers might then leverage that relationship to attract more users
and increase retention percentages of those services through targeted design, promotions,
advertising methods, and other means.
This second approach extends that primary research in several ways. First, it
distinguishes features that promote utilization of e-commerce payment services by
individuals and also identifies which website features are considered as unimportant by
different personality types. Additionally, this second study adds confirmation of
personality type's effect on usage.
Which e-commerce website features and mechanisms promote user perceptions of
perceived ease-of-use and perceived benefit and consequently their intention to continue
utilizing these e-commerce websites has been previously studied (Yousafzai, Pallister, &


56
Foxall, 2005). As an example, Walczak and Gregg (2009; 2010) demonstrate that
website content features promote trust and consequently engender perceptions of
corporate capability and intention to transact. However these studies are all based on a
generic user-type representative of average consumer values across a particular region or
even across the world, similar to personas. Variance among individuals is mostly ignored.
As discussed earlier in the introduction to the primary approach above, the
examination of online credit card payment services from an individual perspective is
problematic due to the sheer volume of data to be analyzed in order to ensure accuracy of
the research model. Therefore to again make use of a classification schema to help
preserve individual differences, and complement and build upon the primary approach,
this approach also utilizes the MBTI type indicator to provide such a classification
technique.
Methodology
A self-administered survey instrument was developed and used to gather data
through an online site hosted by Survey Monkey. Data was actively collected starting in
January 2013 and terminating in July of 2013. The survey was offered to perspective
respondents as part of the University of Colorado Denver Business School's "Business
Engagement Assignment Program". This is a program run by Professor Stefanie Johnson
of the business school to encourage student participation in academic and research
activities outside of their enrolled classes. Students enrolled in undergraduate programs
are required to participate in one activity over the course of each semester. They select
their activity from a list of prospective surveys and other research projects. Upon
completion they are given credit for an assignment by their respective instructors per the


57
class's syllabus. The survey is made up of 6 major sections. Section 1 collects basic
demographic information. Section two and three make up the MBTI short form
comprised of twenty questions, (not shown) with responses used in calculating the
individuals MBTI. Sections 5 and 6 collect payment behavior. Section 5 utilizes 6
questions based on a summated rated seven point Likert scale ranging from Disagree
Totally to Agree Totally commonly utilized in the Psychology field (Blumberg, Cooper,
& Shindler, 2008; Likert, 1932; Spector, 1992). And Section 6 is made up of 5 open
response questions collecting information about user preferences as to online payment
site features, and why they pay or do not pay online. The survey was tested on a small
focus group to make sure that the questions were easily understandable. Potential
respondents were invited from both undergraduate and graduate courses in the Business
School, with the qualification that they own at least one credit card and that they pay their
own credit card bills. Being an urban university the respondents were a blend of older
and employed students taking courses alongside more traditional students. Most of the
respondents received course assignment credit for completion of a course requirement to
participate in external research activities. No other compensation was offered. No
identifiable information was recorded, and respondents were not screened prior to starting
the survey.
Data
The data consists of two main sets of variables. The first set is composed of 6
variables making up the demographic information, 8 variables storing the MBTI polar
sums, 4 variables containing the computed MBTI dimensions, and 20 variables
containing user preference values. These are listed in the variable Key in Appendix C.


58
The second set of Data contains responses from five open ended questions. Three
of the questions are addressed by those who often or always pay their credit card bills
online, and 2 are addressed by those respondents that seldom or never pay their credit
card bills online.
Analysis
A total of 107 respondents started the survey, but 27 of these indicated they did
not pay their own credit card bills and were disqualified from completing the survey. The
remaining 80 respondents were divided into two categories: regularly pays their credit
card online (n=74) and rarely or never paid their credit card online (n=6). The small
number of respondents causes some limitations in generalizing the results and further in
interpreting results for subgroups or personality types with very small response
populations (less than 5% of the total population). Future research will continue to
collect data and try to determine if the initial descriptive results reported in this article are
consistent with a larger population and generalizable to the population of credit card
users at large. Until that point it is stressed that caution should be taken in the
interpretation of the inferences and or conclusions drawn in from this approach.
Demographics for the respondents are listed in Table 19.


59
Demographics
Table 19. Credit Card Website Perception Survey Respondent Demographics
Sex 58.75% male 41.25% female
Age 6.25% 18-20 62.50% 21-29 23.75% 30- 39 6.25% 1.25% 40-49 50-59
Education 3.75% High School 41.25% Some college or Associate degree 45% Bachelor's degree 10% Graduate degree
Income in $ 17.5% <10,000 13.75% 10,000-19,999 17.50% 20,000-29,999 6.25% 30,000-39,999
7.5% 40,000-49,999 6.25% 50,000-59,999 13.75% 60,000-69,999 3.75% 70,000-79,999
2.5% 80,000-89,999 3.75% 90,000-99,999 7.5% > 100,000
The demographics indicate that the respondents, even though students, are more
mature than traditional college age students. Slightly more males than females
responded. Over half of the respondents had already finished their first Bachelors
degree or higher. Interestingly, the income question has a bubble below 30,000 which
may indicate full time students who can only work part-time and thus have more limited
income and a second, almost normal curve above the low income bubble, with a median
around the 60K income range. The low income bubble represented just slightly less than
50 percent of the respondents, indicating that at least half are earning income well above
the poverty guideline level for the USA and should therefore have disposable income for
making and paying credit card purchases (USDHHS, 2013).
Cluster Analysis
A Hierarchal Cluster analysis with dendrograms has been carried out for
respondents who pay online. This was not conducted for non-payers due to the small
sample size. The Complete-Linkage technique is being employed to leverage its strength
in creating tight clustering (Tabachnick & Fidel, 2007). This will also eliminate problems


60
that would be seen if Single-Linkage were used such as chaining (Hair et al. 2005).
The Furthest Neighbor cluster method was used. The cluster analysis was run employing
the Squared Euclidean method, clustering on the preference variables as shown below in
the Agglomeration Schedule in Table 2 with its associated dendrogram in Figure 6.
Table 20. Cluster Analysis Agglomeration Schedule
Agglomeration Schedule
Stage Cluster Combined Coefficients Stage Cluster First Appears Next Stage
Cluster 1 Cluster 2 Cluster 1 Cluster 2
1 1 2 24.098 0 0 4
2 4 9 46.005 0 0 7
3 5 8 56.649 0 0 5
4 1 6 57.980 1 0 6
5 3 5 88.417 0 3 8
6 1 7 89.622 4 0 7
7 1 4 109.030 6 2 8
8 1 3 169.328 7 5 0


61
Dendrogram using Complete Linkage
Rescaled Distance Cluster Combine
0 5 10 15 20 25
Latefees 1
OnTime 2
CredrtScore 6
Confirmation 7
Security 4
Saf ew ay 9
Tutorial 5
OLHelp 8
SocialMedia 3
Figure 6: Combined Linkage Dendrogram
Clustering analysis is employed to identify meaningful groups in the data. There
is no set heuristic for drawing the line. Drawing the line at this point creates the most
detail while maintaining the most homogenous of groups. Analyzing the dendrogram the
logical place to draw the cut offline is at approximately 14 on the distance scale. This
resolves three distinct and informative groups pointing to the individual respondent's
preferences. The first group clusters the Latefees, OnTime, CreditScore, and


62
Confirmation variables. These are all issues that relate to insuring payments are made on
time and their impact on late fees and maintenance of good credit standing.
The second cluster comprised of the Security and Safeway variables both address
online security issues. These users are mainly concerned with protection of information
assets.
The third cluster groups Tutorial tightly with OLHelp, and includes SocialMedia
slightly loosely. This group would appear to be primarily interested in ease of use
factors, and social perceptions.
In addition the homogenous natures of these clustering groups are a confirmation
of the accuracy of the survey questions in measuring the intended factors.
Results And Findings
There are two ways to examine the surveys results: looking at individual aspects
of each personality type, recalling that they are in diametric pairs; or looking at each of
the 16 complete personality types as defined by MBTI. First it must be noted that as
also seen in the SEM analysis the responses did not vary significantly by sex as measured
by a Kolmogorov-Smirnov test, withp = 0.97936, indicating that the two populations of
responses have nearly identical distributions.
Aspects Of Personality Types
Table 21 shows the demographic percentage of each personality type aspect along
with response percentages for each of the measured factors. The presence of security
features was deemed as very important by all personality type aspects and was seen as the
most important feature for all but the N and F personality aspects. This is not unexpected
with the prevalence of identity theft (Identity Theft Assistance Center, 2013).


63
The second most desired or beneficial feature was uniformly to make payments
on time, except for the N and F aspects, where it was the most important. This is
surprising, as perceptions of feeling safe conducting online payments (the last column) is
intuitively correlated with the presence of website security features. This is indicated
since the feeling of safety was third for most personality type aspects, except for N where
it was tied for most important feature. The two remaining features which both had higher
than 50 percent recognition by the respondents as important or beneficial features were
avoiding late payments and receiving confirmation of a payment being made.
Table 21. Important credit card payment service features and personality aspects
MBTI Pole & Pop.# Percent of total Avoid late fees Make Payments on time Tutorial available Live Chat available Confirm payment Improv e Credit Score Heard it was a good idea on Social Media Security features Perception of Safety
64.00% 74.00% 20.00% 14.00% 60.00% 48.00 6.00 82.00 74.00%
E 45 62.50 % % %
70.00% 83.33% 16.67% 13.33% 50.00% 50.00 0.00 93.33 80.00%
I 29 37.50 % % %
58.49% 73.58% 16.98% 11.32% 54.72% 47.17 1.89 88.68 71.70%
S 51 66.25 % % %
77.78% 85.19% 22.22% 18.52% 59.26% 51.85 3.70 81.48 85.19%
N 23 33.75 % % %
71.43% 90.48% 23.81% 19.05% 76.19% 42.86 4.76 85.71 76.19%
F 18 26.25 % % %
64.41% 72.88% 16.95% 11.86% 49.15% 50.85 1.69 86.44 76.27%
T 56 73.75 % % %
72.22% 83.33% 20.37% 14.81% 62.96% 55.56 0.00 90.74 87.04%
J 38 67.50 % % %
53.85% 65.38% 15.38% 11.54% 42.31% 34.62 7.69 76.92 53.85%
P 36 32.50 % % %
These results for personality type aspects indicate that it is very important for
online credit card payment services to maintain high security standards and promoting a
general feeling of safety of information and privacy of information as well as advertising
these features to consumers. Table 21 reveals that it is also very important to provide a


64
user-friendly confirmation of payments and attract potential users by educating them
about the efficiency of online credit card payment services with regard to making on time
payments.
An indication of the reliability of using MBTI as a determinant of e-commerce
payment service feature preferences is the question concerning the importance of external
influences on adoption of online credit card payment services. As may be seen from
Table 20, the perceptive (P) and extroverted (E) types were the personality type aspects
most affected by the external influence of social media. Perceiving types are commonly
seen as flexible and adaptable to the world and like to utilize external input for decision
making instead of imposing order (Myers, 1998; MyersBriggs.org, 2013). Extroverts are
also prone to seek recognition and approval from peer groups (Myers, 1998;
MyersBriggs.org, 2013). Both of these personality types would be more likely to utilize
social media influences over their MBTI counterparts of introversion and judgment.
The other aspect that should perceive external influences from online social networks as
important would be the feeling (F) aspect, which places importance on other people
affected or involved in a decision as opposed to factual data and this is the case with the F
type being the third most influenced by social media (Myers, 1998; MyersBriggs.org,
2013).


65
MBTI And Perceptions Of Online Payment Service Features
However, according to Jung (1921) and consequently the Myers-Briggs MBTI
individuals personality types are a composition of multiple aspects (Myers & McCauley,
1985). Next we examine the 16 personality types. Figure 7 shows the percentage of each
of the MBTI present in the survey response population. It is interesting to note that the
two largest populations are for ESTJ and ISTJ personality types with the only other
personality representing greater than 10% of the population being ESTP. This reveals
that current users of online credit card payment services tend to have the STJ aspects in
their personality and secondarily EST personality aspects. As mentioned in the
individual personality type aspect analysis, these two partial types both make sense. The
STJ type likes to take the world (or electronic world) as it is, dealing in facts, and finally
imposing order through their decisions, thus using an electronic resource enables the STJ
to get the facts of their account and deterministically decide how to affect their account
through an online payment and impose this reality immediately (Myers, 1998;
MyersBriggs.org, 2013). The EST type is similar, but substitutes interaction with the
online world for the imposition of will from the J personality aspect.


66
MBTI Sample Population and Percentage
ENFJ
ENFP
ENTJ
ENTP
ESFJ
ESFP
ESTJ
ESTP
INFJ
INFP
INTJ
INTP
Figure 7. Percentage of MBTI personality types in survey population
Table 21 reports for each of the 16 types results corresponding to those reported
for the individual aspects, but the percentages are with respect to all respondents with the
same MBTI (hence, ESTJ percentages represent the percentage of ESTJ respondents
that selected a value of 4 or 5 for the corresponding online credit card payment service
feature). The gray shading indicates MBTI personality types that did represent at least
5 percent of the overall population and thus pose an analysis limitation due to too few
responses of that category.


67
Table 22. Important credit card payment service features for MBTI personality
types
MB TI type Avoid late fees Make payments on time Tutorial availabl e Live Chat available Confirmation of payment Improv e Credit Score Heard it was a good idea on Social Media Security features Perception of Safety
ESFJ 2 66.67% 66.67% 33.33% 33.33% 66.67% 33.33% 0% 66.67% 66.67%
ESF PI 100% 100% 0 % 0 % 0 % 100% 0% 100% 100%
EST J20 66.67% 80.95% 19.05% 9.52% 66.67% 57.14% 0% 90.48% 90.48%
EST P9 33.33% 44.44% 11.11% 11.11% 44.44% 33.33% 11.11% 77.78% 33.33%
ENF J2 100% 100% 0 % 50.00% 100% 0% 0% 100% 50.00%
ENF P4 60.00% 80.00% 60.00% 0 % 80.00% 40.00% 20.00% 80.00% 60.00%
ENT .14 100% 83.33% 16.67% 33.33% 66.67% 66.67% 0% 66.67% 100%
ENT P3 33.33% 66.67% 0 % 0 % 0% 33.33% 0% 66.67% 66.67%
ISFJ 3 33.33% 100% 0 % 0 % 66.67% 33.33% 0% 100% 100%
ISFP 1 100% 100% 0 % 0 % 100% 0 % 0% 100% 0 %
ISTJ 12 66.67% 75.00% 25.00% 16.67% 50.00% 58.33% 0% 100% 75.00%
ISTP 3 66.67% 66.67% 0.00% 0 % 0 % 0% 0% 66.67% 33.33%
INFJ 3 75.00% 100% 25.00% 0 % 75.00% 75.00% 0% 75.00% 100%
INF P2 100% 100% 0.00% 100% 50.00% 50.00% 0% 50.00% 100%
INTJ 4 75.00% 75.00% 25.00% 0 % 50.00% 50.00% 0% 100% 100%
INT PI 100% 100% 0 % 0 % 0 % 100% 0% 100% 100%
The questions in the survey correspond to several distinct categories: security
(Security features and perception of safety), timeliness (make on time payments and


68
avoid late fees), personal benefit (avoid late fees and improve credit score), ease of use
(tutorial and online chat), confirmation of a completed transaction, and external social
influences (recommendation from friends or social media). From Table 22, looking at
only the non-shaded rows and thus those personality types that represent at least 5 percent
of the population of respondents, it may be seen that the average of the two security
influences is the most important (or tied for most important) for all personality types
except for ENTJ. This may not be a feature of personality type so much as it is a
necessity in todays online society with hackers and identity theft threats causing people
to desire the appearance of secure online transactions.
The second most influential feature category and for some personality types tied
for first, is timeliness followed closely by confirmation. Thus it appears that the ability to
make payments anytime and anywhere in an efficient manner is highly important to
online payment adopters with confirmation of the fact that the payment has occurred,
possibly for purposes of non-repudiation of the payment, the third most significant
influence. External influences appears to not have any influence on the desire to utilize
online credit card payment services by most personality types, with minor influence
extending to those personality types that have both E and P aspects, similar to the
findings for the individual aspects reported above.
Since security and efficiency of making payments with a subsequent confirmation
that the payment has in fact been made are all important features to online payment


69
service users, these are features that need to be promoted by credit card payment service
providers to continuation of their payment services by existing customers.
The open ended questions also supported the importance of the timeliness feature
of online credit card payment services. Samples of some of the open-ended question
responses for various respondents who pay regularly online are:
I pay my credit card bills online because it is convenient, fast, and hassle free.
Easy and convenient for a busy lifestyle.
Its easier and faster than paying by snail mail.
I pay online because of the increased access to information, instant payment, and
to save stamps and paper.
Convenience. Traditional mail, phone banking, or paying in person are all
unpleasant and time consuming.
The fourth item in the responses list shown above also indicates another theme in
the open ended question responses that occurred a little less often than timeliness, which
is personal benefit or in this case cost savings.
The survey has indicated the features perceived as beneficial or necessary by
online payment service users. However, a small population of credit card owners that did
not pay regularly online also answered the survey and now we look at the features that
this group perceived as being useful or needed as well as those features that were not
perceived as beneficial and as such would not lead to utilization of the online credit card
payment service. The MBTI personality type for the respondents that indicated they


70
did not pay their credit card debt online, were varied and included at least one case of
each personality type aspect. Table 22 represents the agreement or disagreement Likert
responses for questions similar to what was asked of the online payment service users,
but this time with a focus on what features would promote utilization by the current non-
users. If a row does not add up to 100 percent, this represents non-users who were
neutral on a specific category.
Table 23 indicates some similarities between users of online credit card payment
services and the non-users. Feeling safe, making timely payments with corresponding
benefits, and receiving confirmation of the transaction were all important features and
external influence from social media was not considered an important or attractive
feature.
Table 23. Feature perceptions by non-users of online credit card payment services
Feature Feature not desired Feature desired
Security 100% 0%
Feel Safe 0% 100%
Make payments on time 33% 50%
Avoid late fees 17% 83%
Improve my credit score 0% 50%
Receive confirmation 17% 83%
Tutorial 17% 50%
Online chat available 50% 17%
Hear about it on social media 83% 17%
An important difference between non-users and users of online credit card
payment services is the lack of recognition of security as an important feature. All
respondents who did not pay regularly online indicated that securing their personal and
financial information was not important. Is this due to a lack of awareness of online
financial transaction security risks, or perhaps a feeling that security will always be
present and as such is not an issue. Additionally, although the non-users indicated that


71
security was not important, they did universally indicate that they wanted to feel the
transaction was safe, indicating a disparity in their responses.
Further research is needed to examine and help determine why non-regular users
of online payment services and perhaps e-commerce transactions in general do not see a
correlation between the security features of a web service and the consequent privacy or
safety of their information.
Another interesting finding from the responses shown in Table 4, is that half of
the non-users for online credit card payment services indicated that the presence of an
online tutorial would increase the likelihood of their utilizing such a service, but a similar
percentage indicated that presence of an online chat feature would not increase their
willingness to utilize such online payment services. One might think that the online
payment service non-users were primarily introverted in nature and hence did not desire
to interact with another person, but in fact, less than 17 percent had the I aspect on their
MBTI type. A possible explanation is that these non-users feel their chat questions
would be considered naive and that the person(s) on the other side of the chat service
would be judgmental and think less of them for not knowing, thus decreasing their self-
stature. As such, online credit card payment providers should provide easy to understand
and yet thorough tutorials of their service and also promote the friendly, helpful, and
customer service focused nature of their live chat features, to lessen fears preventing
potential new users from utilizing these features to become regular users.


72
CHAPTER VI
LIMITATIONS AND IMPLICATIONS
One of the primary limitations of this research is its' exploratory nature. There is
no current known causational model for personality type's effect on online credit card
payment service usage. This study is testing a hypothesized model using the sampled
data. This research has a firm theoretical underpinning derived from the utilization of
MBTI variables, and TAM, and DOI constructs yet the true causal pathways underlying
relationships between personality type and the models constructs are only now being first
tested. SEM is best used when casual paths are well known (Kline, 2012). If the use of
these well founded variables and constructs had not been employed this study's
confirmation of hypotheses could only be viewed as an indication that the hypothesized
model cannot be found invalid. While it is felt that the employment of MBTI, TAM,
and DOI variables and constructs puts this study on somewhat firm ground than that, the
lack of confirmed pathways between the exogenous MBTI and the endogenous TAM
and DOI variables in the study points to the need for additional work focused to further
confirm the model.
There is concern as to the malleability of the personality as a possible limitation to
any findings related to types. If an individual's personality changes under day to day
variable situations, social influences, or other normal pressures then any linkages to
preferences based on a measurement of personality type might change as well. In general
the psychology literature finds that an individual's personality is primarily developed
early on in life with some researchers considering it fixed prior to adolescence, while
others holding that development continues into early adulthood out to perhaps 30 years of


73
age (Caspi & Roberts, 2001; Fenichel, 1945; Rothbart, Ahadi, & Evans, 2000; Srivastava,
Oliver, Gosling, & Potter, 2003). The current general consensus is that once it is formed,
the individual's core personality is primarily stable and while it can change this will
normally be a slow process over the course of one's life (Haan, Millsap, & Hartka, 1986;
Heatherton & Nichols, 1994; Jones & Meredith, 1996; Mroczek & Spiro, 2003). An
individual's behavior may differ in response to situational variables which is normal and
expected (Gendlin, 1964, Heatherton & Nichols, 1994). These normal changes in
responses are found to be within a behavioral range tied to higher level personality traits
and not indicative of any fluctuation to the core personality type (Gendlin, 1964; Roberts
& Walton, 2006; Rogers, 1957). In cases where abrupt personality change is observed it
is invariably related to extreme trauma with related stress, and/or symptomatic of a deep
rooted mental, emotional, or pathologically based personality disorder (A.P.A, 1980;
Rabkin & Struening, 1976; Sanislow & McGlashan, 1998; Vitaliano, DeWolfe, Maiuro,
& Katon, 1990). Due to the slow nature of any normal longitudinal changes to an
individual's core personality any effect they might have on the outcome of an MBTI
test would not be relevant to a study such as this, while rapid changes as noted are
abnormal exceptions, and therefore outside the scope of this research.
Implications For Payment Service System Designers
The goal of managers of credit card payment service systems is to reduce
transaction costs while satisfying customer needs. Increasing usage is of paramount
importance to maximize savings and gain return on the service site development
investment.


74
As indicated in the SEM analysis section, the MBTI type ESTJ individuals are the
most likely to utilize credit card service systems for managing their accounts. This
personality type indicates that being action-oriented with an outer-world focus (E) and a
desire for information that is tangible (S) with known logical consequences (T) naturally
will accept and use credit card payment service systems (Myers, 1998). Since these
systems are under used, then the question becomes how to attract this natural personality
type to the payment systems.
The extraverted nature implies that social networking would be an ideal way to
engage this type of individual, yet findings from the second approach in this study
indicate that most users and non-users alike are not largely influenced by social media
recommendations. This warrants additional research to determine if there is some
cognitive disconnect at play here.
Additional research could focus on addressing implementation of design features
calculated to attract personality types not currently making use of online payment
services. This might take the form of using "Green" strategies in order to make online
payment service have the perception of greater compatibility and usefulness to Feeling
types. The idea of "Going Paperless" can appeal to this type as they are guided by
personal values, compassion, and have a need for harmony and positive interactions
(Myers, 1998; Myers & Myers, 1995).
More introverted users that feel more comfortable writing a check, may want to
utilize an online system as long as they do not require any interaction with or requests for
help from another person. Augmenting payment sites to have self-paced tutorials and
intelligent help systems that automatically supply information about what is needed or the


75
next step to be performed, might succeed in making I personality types feel more
comfortable using the system on their own.
Intuitive personality types want to see the big picture (Myers, 1998). Therefore
advertising or messages on the payment website should enable N personality types to get
a grasp of how not only the payment system but also the credit card payment process
works as a whole and consequently how it benefits the user (e.g., through reduction in
late fees and improvement in their credit rating
The recommendations suggested here target the personality types that underutilize online
credit card payment services. The suggestions seek to both improve utilization of
personality types that are positive towards online credit card bill payments, but just
underutilize this service and to try to supply personality congruent features or advertising
strategies for those personality type components that typically would not intend to use a
credit card payment site. Future research is needed to verify if these personality-targeted
recommendations will bear fruit and to identify other personality trait effective
recruitment strategies.
The small response size of Approach 2 limits the generalizability of the results of
that section of the study to other areas of e-commerce. Future research is needed to
examine if personality may explain differential utilization of other types of e-commerce
sites or even websites in general.
Implications For Research
The fact that perceived compatibility was more significant in the current study
than the traditional TAM constructs may reflect growing computer self-efficacy. With
computers now being used in elementary schools and a generation that has grown up with


76
cell phones and laptops, it is not uncommon for new college graduates entering the
workforce to be comfortable with, fluent in the usage of, and to some extent reliant upon
information technology (El Nasser, 2012). Thus the introduction of compatibility into the
more traditional TAM model reflects the necessary adjustment of the model to account
for ever growing computer competency in the work force. Therefore, although the TAM
model is still thought to be a reliable indicator of acceptance of new applications and
technology, it is imperative that new models of information technology utilization include
some component like perceived compatibility to better model the technology savvy
population. The utilization of compatibility in improved TAM-like models is similar to
prior research which has shown in some professions the compatibility of a new
application with the existing work-flow is of paramount importance (Richards, 2008;
Sibona, Walczak, Brickey, & Parthasarathy, 2011; Walczak, 2003; Walter & Lopez,
2008).
The second major finding for research is the indication of the possible impact of
personality on usage of web services, specifically credit card payment sites. A wide
variety of future research can stem from the findings of the presented research. Can this
model be further confirmed? How does personality affect other information technology
adoption decisions?
Also, is it possible to develop a non-intrusive way to automatically assess personality
traits or will users have to take a mini-personality test for websites to be able to
adaptively present the best interface for each user?
This study has been limited to the use of one personality indicator test. Future research
should examine if other alternative personality tests such as The Big Five, or Eysencks


77
Personality Questionnaire provide similar ratings and whether these profiles might be
easier to implement. Alternatively, would it be possible to infer aspects of personality
type based on the cognitive selection of website customization features, which could in
turn promote compatibility through suggestions of other, as yet undiscovered, website
features for the user.
Individuals with the ESTJ personality traits were identified as being the most
likely to utilize online credit card payment services. However this leaves out a large
percentage of the population. Previously, recommendations have been presented for how
to attract other personality types, but future research is needed to explicitly examine how
different personality types and cultures interact with and utilize different types of web-
based services, especially e-commerce sites and applications. Prior research has already
found cultural differences in specific types of web utilization between collectivist
cultures versus individualistic cultures and this research should be continued but now
emphasizing individual personality traits in addition to cultural effects (Benet-Martinez &
John, 1998; Kalwar, 2010; Kim, Sohn, & Choi, 2011).
It is also thought that additional research should be carried out exploring the ESTJ
type from a reverse perspective. Following back from the finding of ESTJ as the
dominant type of adopters to their underlying motivations, preferences, and values may
yield valuable insights for exploitation by designers, site marketing managers, and
researchers.
Although a small sample size, there was some indication that non online payers
were found to be concerned with safety but did not cognitively link that to online security
issues. It is felt that additional inquiry into this finding would add increased granularity


78
into understanding the role of risk acceptance. And as touched on in the discussion to the
second approach this has implications for the need of further research in order to
determine why non-regular users of online payment services do not see a correlation
between the security features of a web service and the consequent privacy or safety of
their information and whether this extends even to e-commerce transactions in general.


79
CHAPTER VII
CONCLUSIONS
Conclusions Approach I
This study proposed that personality type has a strong influence on the decision to
use online credit card bill payment services and tested 12 subsequent hypotheses. The
analysis shows that MBTI's types including the Extraverted and Sensing factors had the
highest single direct effect on any of the latent unobserved variables of the MBTI
factors. They are present in the combined indicator types of ESTJ, and ESFJ which was
found to account for variances in the constructs of PEEi, PC, and PU and the independent
variable U. The study therefore provides evidence for confirmation of hypotheses HI a
through H9b, and HI la to H12b. Due to the disparate findings of the SEM and the PLS
analysis's related to T and F dimension path to the PU construct full confirmation of
HlOa & b are not confirmed.
The Judgment factor was found to have a strong effect on PEU both individually
and as a member of the combined ESTJ and ESFJ types on U, and having a contributory
effect on PC. These findings yield additional evidence to the confirmation of hypotheses
Hla & b, H2a & b and H3a & b.
The support for the hypotheses confirms that personality traits have a direct
effect on PU, and PC and indirectly on usage online credit card bill payment services.
Additional research is needed to see if these findings may be generalized to e-commerce
services in general.


80
Finally, the presented research has verified compatibility or PC as a reliable
predictor of usage of online payment services technology. This finding indicates that
future adoption and technology utilization models need to utilize PC as a factor, which
should improve their reliability. Future research may investigate if specific domains are
more prone to compatibility requirements, such as healthcare systems (Sibona, et al.,
2011; Walczak, 2003; Walter and Lopez, 2008).
Conclusions Approach II
The research presented in Approach 2 is an exploratory descriptive research. It
should be understood that this study is based on a small sample size. Any conclusions
drawn by the study should be taken with caution due to this limitation.
This approach attempts to better understand how individual personality may affect
which features of online credit card payment systems are perceived as useful and also
those perceived as unnecessary by both users and non-users of these services. For online
payment services, security, followed closely by timeliness or efficiency in making
payments were the two most valuable features of online payment services. Additionally
receiving confirmation of payment was also viewed as a valuable service.
For some reason, non-users of online credit card payment services uniformly disagreed
that security was a desirable feature of such websites. However, these non-users also
indicated they wanted to feel safe in making an online payment should they ever use such
a service, which raises issues and the need for future research about the cause of this
disparity.


81
The results in Table 3 may be used by payment service providers and other e-commerce
transaction providers to determine how to increase utilization of their services. As an
example, the ESTJ personality was indicated to be the personality that most frequently
utilizes online payment services. Noticing differences between this personality types
perceptions and other personality type perceptions may enable online payment service
providers to attract individuals with other personality types. For example, the row
indicating the perceptions of ENTJ (more intuitive thinkers than the ESTJ type [26])
indicates that this personality type places more emphasis on both avoidance of late fees
and also on a general feeling of safety than their ESTJ counterparts. Likewise, ISTJ
personality types indicated that with regard to security and timeliness of payments, they
were very similar to ESTJ users, but they placed slightly more emphasis on the presence
of tutorials and live chat help. The ISTJ increased emphasis on tutorials and help over
their ESTJ counterparts indicates that even though they prefer to solve problems on their
own, because of this they may require some additional learning or other help to feel
confident in their decisions.


82
REFFERENCES
Abramson, M. A., & Morin, T. C. (2003). E-government 2003. Lanham, MD: Rowan and
Littlefield.
Adams, D. A., Nelson, R. R., & Todd, P. A. (1992). Perceived usefulness, ease of use, and usage
of information technology: A replication. MIS Quarterly, 16(2), 227-247.
Agarwal, R., & Prasad, J. (1997). The role of innovation characteristics and perceived
voluntariness in the acceptance of information technologies. Decision Sciences, 28(3),
557-582.
Agarwal, R., & Prasad, J. (1999). Are individual differences germane to the acceptance of new
information technologies? Decision Sciences, 30(2), 361-391.
Ajzen, I. (1985). From Intentions to actions: A theory of planned behavior. In J. Kuhl and J.
Beckmann (Eds.), Action control: From cognition to behavior. Berlin, Heidelberg, New
York: Springer-Verlag.
Ajzen, I. (1991). The theory of planned behavior. Organizational Behavior Human and Decision
Processes, 50(2), 179-211.
Ajzen, I., & Fishbein, M. (1977). Attitude-behavior relations: A theoretical analysis and review
of empirical research. Psychological Bulletin, 84(5), 888-918.
Ajzen, T, & Fishbein, M. (1980). Understanding attitudes and predicting social behavior.
Englewood Cliffs, NJ: Prentice-Hall.
Akaike, H. (1973). Information theory and an extension of the maximum likelihood principle. In
Pertrov, B. N. & Csaki, F. [Eds.], Proceedings of the 2nd International Symposium on
Information Theory, 267-281.
Akinci, S., Aksoy, S., & Atilgan, E. (2004). Adoption of Internet banking among sophisticated
consumer segments in an advanced developing country. International Journal of Bank
Marketing, 22(3), 212-232.
Albert, T. C., Goes, P. B., & Gupta, A. (2004). GIST: A model for design and management of
content and interactivity of customer-concentric web sites. MIS Quarterly, 28(2), 161-
182.
Amichai-Hamburger, Y. (2002). Internet and personality. Computers in Human Behavior, 18(1),
1-10.
Amichai-Hamburger, Y., & Ben-Artzi, E. (2003). Loneliness and internet use. Computers in
Human Behavior, 19(1), 71.


83
Amichai-Hamburger, Y., Fine, A., & Goldstein, A. (2004). The impact of Internet interactivity
and need for closure on consumer preference. Computers in Human Behavior, 20(1),
103-117.
American Psychiatric Association (1980). Diagnostic and statistical manual of mental disorders
(3rd ed.), Washington D. C.: American Psychological Association Press.
Bayne, R. (1995). The Myers-Briggs type indicator: A critical review and practical guide.
London: Chapman and Hall.
Belleau, B. D., Summers, T. A., Xu, Y., & Pinel, R. (2007). Theory of reasoned action: Purchase
intention of young consumers. Clothing and Textiles Research Journal, 25(3), 244-257.
Benet-Martinez, V., & John, O. P. (1998). Los cinco grandes across cultures and ethnic groups:
Multitrait multi-method analyses of the Big Five in Spanish and English. Journal of
Personality and Social Psychology, 75(3), 729-750.
Bentler, P. M. (1990). Comparative fit indexes in structural models. Psychological Bidletin,
107(2), 238-246.
Bernstein, D. A., Clarke-Stewart, A., & Roy, E. J. (2008). Psychology (8th ed.), Boston MA:
Houghton Mifflin Company.
Blumberg, B., Cooper, D. R., & Schindler, P. S. (2008). Business research methods (Vol. 2).
New York: McGraw-Hill Higher Education.
Borzekowski, R., E.K. Kiser, "The Choice at the Checkout: Quantifying Demand Across
Payment Instruments", International Journal of Industrial Organization, Vol. 26:889-
902, 2008.
Brenner, V. (1997). Psychology of computer use: XLVII. Parameters of internet use, abuse and
addiction: The first 90 days of the Internet Usage Survey. Psychological Reports, 80(3),
879-882.
Brickey, J., Walczak, S., & Burgess, T. (2010). A comparative analysis of persona clustering
methods. AMCIS 2010 Proceedings, Paper 215.
Browne, M. W., & Cudek, R. (1989). Single Sample cross-validation indices for covariance
structures. Midtivariate Behavioral Research, 24(4), 445-455.
Browne, M. W., & Cudeck, R. (1993). Alternative ways of assessing model fit. In Bollan K. A.,
and Long J. S. EDs. Testing Structural Equation Models, Newbury Park: CA, Sage.
Byrne, B. M. (2010). Structural equation modeling with AMOS. (2nd ed.). Ottawa, OT:
Routledge.


84
Carlson, J. (1985). Recent assessments of the Myers-Briggs type indicator. Journal of
Personality Assessment, 49(4), 356.
Carmines, E. G., & Mclver, J. P. (1981). Analyzing models with unobserved variables. In
Bohrnstedt, G. W. and Brogatta, E. F. [Eds.] Social Measurement: Current Issues.
Beverly Hills: CA, Sage.
Caspi, A., Roberts, B. W. (2001). Personality development across the life course: The argument
for change and continuity. Psychological Inquiry 12(2), 49-66.
Chau, P. Y. K. (1996). An empirical assessment of a modified technology acceptance model.
Journal of Management Information Systems, 13(2), 185-204.
Cooper, A. (1999). The Inmates are Running the Asylum: Why high-tech products drive us crazy
and how to restore the sanity, Indianapolis: SAMS, a division of Macmillan Computer
Publishing.
Davis, F. D. (1989). Perceived usefulness, perceived ease of use, and user acceptance of
information technology. MIS Quarterly, 13(3), 319-340.
Davis, F. D., Bagozzi, R. P., & Warshaw, P. R. (1989). User acceptance of computer technology:
A comparison of two theoretical models. Management Science, 35(8), 982-1003.
Dharwada, P., Greenstein, J., Gramopadhye, A., & Davis, S. (2007). A case study on use of
personas in design and development of an audit management system. Proc. Human Factors
and Ergonomics Soc. 51st Ann. Meeting, 469-473.
Didio, L. (1998). Beta testers endorse e-checks. Computerworld, 32, 57.
Digman, J. M. (1990). Personality structure: Emergence of the five factor model, Annual Review
of Psychology, 41(1), 417-440.
Dishaw, M. T., & Strong, D. M. (1999). Extending the technology acceptance model with task-
technology fit constructs, hiformation & Management, 36(1), 9-21.
Edwards, J. R., & Bagozzi, R. P. (2000). On the nature and direction of relationships between
constructs and measures. Psychological Methods, 5(2), 155-174.
Emergenetics (2013). Case Studies. [Online] Retrieved from
http://www.emergenetics.com/results/case-studies.
Emergenetics (2013). Clients. [Online] Retrieved from http://www.emergenetics.com/
results/clients.


85
El Nasser, H. (2012). The office is shrinking as tech creates workplace everywhere. USA Today
[Online] Retrieved from http://www.usatoday.com/money/workplace/story/2012-06-
05/tech-creates-workplace-everywhere/55405518/l.
Eysenck, H. J., & Eysenck, S. B. G. (1975). Manual of the Eysenck personality questionnaire.
London: Hodder and Stoughton.
Federal Reserve Bank of Boston (2010). The survey of consumer payment choice [Online],
Retrieved from http://www.bostonfed.org/economic/cprc/SCPC/index.htm.
Fenichel, O. (1945). Father absence and the personality development of the male child.
Fishbein, M., & Ajzen, I. (1975). Belief, attitude, intention, and behavior: An introduction to
theory and research. Reading, MA: Addison-Wesley.
Fisher, J. C., & Pry, R. H. (1971). A simple substitution model of technological change.
Technological Forecasting and Social Change, 3(1), 75-88.
Fliegel, F. C., & Kivlin, J. E. (1966). Attributes of innovations as factors in diffusion. American
Journal of Sociology, 72(3), 235-248.
Friedman, B., Kahn, P. H., & Howe, D. C. (2000). Trust online. Communications of the ACM,
43(12), 34-40.
Fomell, C., Larcker, D. F. (1981) Structural equation models with unobservable variables and
measurement error: Algebra and statistics. Journal of Marketing Research, 18(3), 382-
388.
Fuchs, R. (2001). Personality traits and their impact on graphical user interface design: Lessons
learned from the design of a real estate website. In 2nd Workshop on Attitude,
Personality and Emotions in User Adapted Interaction.
Garson, G. D. (2012). Structural equation modeling (2012 ed.). Asheboro, NC: Statistical
Publishing Associates.
Garson, G. D. (2001). Reliability Analysis Reliability analysis faculty and staff North Carolina
State University [Online], Retrieved from
http://faculty.chass.ncsu.edu/garson/PA765/reliab.htm.
Garson, G. D. (2012). Structural equation modeling (2012 ed.). Asheboro, NC: Statistical
Publishing Associates.
Gefen, D., Karahanna, E., & Staub, D. W. (2003). Trust and TAM in online shopping: An
integrated model. MIS Quarterly, 27(1), 51-90.


86
Gefen, D., Straub, D. W., & Boudreau, M. C. (2000). Structural equation modeling and
regression: Guidelines for research practice. Communications of the Association for
Information Systems, 4(7), 2-77.
Gendlin, E. T. (1964). A theory of personality change. In P. Worchel & D. Byrne (EDs.),
Personality Change (pp. 100-148). New York, NY: John Wiley & Sons.
Gregg, D. G. & Walczak, S. (2010). The relationship between website quality, trust, and price
premiums at online auctions. Electronic Commerce Research, 10(1), 1-25.
Guadagno, R. E., Okdie, B. M., & Eno, C. A. (2008). Who blogs? Personality predictors of
blogging. Computers in Human Behavior, 24(5), 1993-2004.
Hair, J. F, Black, B., Babin, B., Anderson, R. E., & Tatham, R. L. (2005). Structural equation
modeling: An introduction, multivariate data analysis (Sixth ed. pp. 729-793). Boston:
Pearson.
Hamburger, Y. A., & Ben-Artzi, E. (2000). The relationship between extraversion and
neuroticism and the different uses of the internet, Computers in Human Behavior, 16(4),
441-449.
Haan, N., Millsap, R., & Hartka, E. (1986). As time goes by: Change and stability in personality
over fifty years. Psychology and Aging, 1(3), 220-232.
Hayashi, F., & Klee, E. (2003). Technology Adoption and Consumer Payment Choice.
http://www.bostonfed.org/economic/cprc/SCPC/index.htm, Accessed on May 10, 2012.
Heatherton, T. F., & Nichols, P. A. (1994). Conceptual issues in assessing whether personality
can change. In J. L. Weinberger (ED.) Can Personality Change? (pp. 3-18). Washington
D.C: American Psychological Association Press.
Hills, P., & Argyle, M. (2003). Uses of the internet and their relationships with individual
differences in personality. Computers in Human Behavior, 19(1), 59-70.
Hoffman, D. L., Novak, T. P., & Peralta, M. (1999). Building consumer trust online,
Communications of the ACM, 42(4), 80-85.
Hsu, C., & Lu, H. (2004). Why do people play on-line games? An extended TAM with social
influences and flow experience. Information Management, 41(7), 853-868.
Hulland, J. (1999). Use of partial least squares (PLS) in strategic management research: A
review of four recent studies. Strategic Management Journal, 20(2), 195-204.


87
Identity Theft Assistance Center. (2013). More than 12 million identity fraud victims in 2012
according to latest javelin strategy & research report [Online], Retrieved from
http://www.identi tytheftassistance.org/pageview.php?cateid=47.
Jones, C. J., Meredith, W. (1996). Patterns of personality change across the life span. Psychology
and Aging 11(1), 57-65.
Jung, C. G. (1921, 1971). Collected Works: Psychological types (Vol. 6). Princeton, N.J.:
Princeton University Press.
Jung, C. G. (1933). Modern man in search of a soul. New York: Harvest book.
Kalwar, S. K. (2010). Comparison of human anxiety based on different cultural backgrounds.
Cyberpsychology, Behavior, and Social Networking, 13(4), 443-446.
Kim, Y., Sohn, D., & Choi, S. M. (2011). Cultural difference in motivations for using social
network sites: A comparative study of American and Korean college students. Computers
in Human Behavior, 27(1), 365-372.
King, W. R., & He, J. (2006). A Meta -analysis of the technology acceptance model.
Information and Management, 4(6), 740-755.
Kirton, M. J. (1976). Adaptors and Innovators: A description and measure. Journal of Applied
Psychology, 61(5), 622-629.
Kline, R. B. (2012). Assumptions in structural equation modeling. In R. Hoyle (Ed.), Handbook
of structural equation modeling (pp. 111-125), New York: Guildford.
Lai, V. S., & Li, H. (2005). Technology acceptance model for internet banking: an invariance
analysis. Information and Management, 42(2), 373-386.
Langdon, C. S., & Shaw, M. J. (2000, January). The online retailing challenge: Forward
integration and e-backend development, In ECIS (pp. 1025-1028).
Lederer, A. L., Maupin, D. J., Sena, M. P., & Zhuang, Y. (2000). The technology acceptance
model and the world wide web, Decision Support Systems, 29(3), 269-282.
Lee, H. H., & Chang, E. (2011). Consumer attitudes toward online mass customization: An
application of extended technology acceptance model. Journal of Computer-Mediated
Communication, 16(2), 171-200.
Lee, M. K., & Turban, E. (2001). A trust model for consumer internet shopping. International
Journal of Electronic Commerce, 6(1), 75-91.


Full Text

PAGE 1

ONLINE CREDIT CARD BILL PAYMENT AND PERSONALTIY TYPE by GARY L. BORKAN B.S. National American University, 1995 M.S., University of Colorado Denver, 2000 A thesis submitted to the Faculty of the Graduate School of the University of Colorado in partial fulfillment of the requirements for the degree of Doctor of Philosophy Computer Science and Information Systems 2013

PAGE 2

ii This thesis for the Doctor of Philosophy degree by Gary L. Borkan has been approved for the Computer Science and Information Systems Program by Deborah Kellogg, Chair Steven Walczak, Advisor Michael Mannino Madhavan Parthasarathy Min Hyung Choi November 15, 2013

PAGE 3

iii Borkan, Gary L. (Ph.D., Computer Science and Information Systems) Online Credit Card Bill Payment and Personality Type Thesis directed by Associate Professor Steven Walczak ABSTRACT The Internet has become a fully integrated part of the worlds business community and social culture. An important part of this integration is the Internets interaction with credit cards, both as a medium of exchange and transaction, and as an interface for credit card account customer service. Credit card providers and bank account service sites provide powerful tools for the individual to manage their credit card accounts, and allowing payment of monthly credit card. It is essential for provider organizations to maximize the value delivered for itself and to their customers, to justify the expense involved in developing and maintaining these service sites and transactional banking links. This study builds upon previous work done on the relationship be tween personality type and Internet usage. It expands this previous work to focus specifically on user adoption of online credit card bill payment services. This work can provide online payment service designers insights into the personality traits of the ir users that might be exploited in order to increase the value derived by the credit card payment service providers and banking organizations as well as enabling wider adoption of online credit card payment services by addressing varying personality based needs of users. The form and content of this abstract are approved. I recommend its publication. Approved: Steven Walczak

PAGE 4

iv DEDICATION For my parents, Lillian K. Borkan, B.S. History, & M.S. Education and Library Science, and Jacob D. Borkan, B.S, Chemi cal Engineering. First, to my Mother, educator and librarian. As a child she taught me stories of her hero and her heroine, Dr. Jonas Salk who succeeded in ridding the world of polio, and Hypatia of Alexandria who unsuccessfully fought to save the knowle dge of five thousand years of accumulated wisdom from destruction by a mob of religion blinded zealots. Through her stories she instilled in me the deepest reverence for books, and a love for science and the progress of knowledge. She viewed these as the underlying foundation of civilization and as "the basis for the progress that has led us up from darkness to face unafraid the light of the stars". These were her words and were her belief, and throughout her lifetime what she taught her pupils. She achieved academic and career goals far beyond those deemed appropriate for the average women to pursue in the late 1940's and early 1950's. To my Father, who although disabled through service to our country, persevered to complete a degree in the physical sciences, and entered upon a career of research. His work led first to advances towards stealth technology in torpedo design, and later with his research partner resulted in the development and patenting of a practical version of the Silver Catalyst Reactio n Control System. More commonly known as the hydrogen peroxide reaction thruster (thrusters), his designs for the catalyst packs in these engines were present on research vehicles such as ultrahigh altitude models of the F104 Starfighter, and the F15. F inally, working as a government subcontractor under FMC

PAGE 5

v Corporation and NASA, he built units that were installed in the escape and attitude control systems aboard the Mercury, Gemini, and Apollo space capsules. I grew up surrounded by blue prints, scale m ockups, working test stand rockets and surrounded by an entire library filled with encyclopedias, history books, and physics manuals. The environment they provided for me yielded a wonder towards nature and science, and a habit of lifelong learning even t hough for many years I did not purse formal education beyond high school. Their prematurely passing in the spring of 1989 caused me to take stock. At that time I dedicated myself to accomplishing in some small way the attainment of the hopes and dreams t hey had held for me and of which felt I had fallen short. This led to me enrolling as a freshman in collage at the age of thirty five. Now it is twenty three years later. Throughout those long years I have relied on their examples. They are my models of courage, stamina, and determination in the face of opposition. They taught me through quite example a belief that you can accomplish what you determine. That no matter what gender, culture, or minority you may come from, whether you are strong or handicapped, that you must stand up after mis steps, impediments, or to those barring the way. Whether you have two good feet on which to stand on or are in a wheel chair with nothing to stand upon other than your character that you can and will succeed. That is what I learned witnessing my parents lives. This work is dedicated with love to them.

PAGE 6

vi ACKNOWLEDGMENTS I would like to extend my thanks and gratitude to all those many individuals without whose support and encouragement this work could not have bee n completed. Among those are a number of individuals that so stand out that I would be remiss in not extending special thanks to them. I want thank my Advisor, Professor Steven Walczak, whose support, guidance, and patience endured through many years of work and effort and has never wavered. I will always appreciate the time and attention he has extended to me, his optimistic attitude, and his dry sense of humor. Many thanks to Professor Michael Mannino for his participation as a committee member and w ithout whose vision and efforts this degree program would not exist. I would also like to thank Professor Madavan Parthasarathy whose initial encouragement gave me confidence to pursue this degree and has been an important resource of information that has been applied to this improve this study. To the other members of my committee, Chair Deborah Kellogg, and MinHyung Choi, thank you very much for your time and many suggestions which have helped make this a far better study. My appreciation to the Facul ty and Staff of the IS Department and the Business Graduate School of the University of Colorado at Denver who have instructed me in the many classes I have attended.

PAGE 7

vii And finally I would like to thank my fellow students, many who have fallen by the way, f or their help through all the years as friends, and as project and study partners. Thank you and good luck to all of you in your own studies and careers.

PAGE 8

viii TABLE OF CONTENTS CHAPTER I. INTRODUCTION ....................................................................................................... 1 II.BACKGOUND AND RELATED WORK ... 6 TAM 6 Diffusions of Innovations 9 Myers Briggs Type Indicator ... 12 Personality Type and the Internet 14 Current Usage of Online Credit Card Bill Payment Services... 16 Scope and Contribution .... 18 III.APPROACH I: MODELING THE EFFECT OF PERSONALITY TYPE ON ONLINE CREDIT CARD PAYMENT SERVICES .. 21 Conceptual Model 21 Research Hypotheses .. 22 Methodology ... 29 IV. ANALYSIS .... 32 SEM Model Analysis .. 32 Assumptions of the Structural Model .. 32 Assumptions of the Reflective Measurement Model .. 33 Reliability of the Constructs 34 Normality of the Constructs 35 Analysis of the Model Fit 37 R2 and Squared Multiple Correlations 38 Convergent Validity 39

PAGE 9

ix Discriminant Validity .. 40 Structural Measureme nt Analysis 41 Mediation Testing ... 43 Additional Variables Tested ... 44 Testing for Variance Due to Gender .. 44 PLS Model Analysis ... 45 PLS Measurement Model ... 46 PLS Reliability 47 PLS Validity ... 48 PLS Structural Model 49 Discussion ... 51 V. APPROACH II: DISCOVERING DESIRABLE AND UNWANTED ONLINE CREDIT CARD PAYMENT SERVICE FEATURES FOR DIFFERENT PERSONALITY TYPES 55 Methodology ... 56 Data 57 Analysis 58 Demographics .. 59 Cluster Analysis .. 59 Results and Findings ... 62 Aspects of Personal ity Types .. 62 MBTI and Perceptions of Online Payment Service Features .......... 65 VI. LIMITATIONS AND IMPLICATIONS 72 Implications for Payment Service Systems Designers .. 73

PAGE 10

x Implications for Research 75 VII. CONCLUSIONS 79 Conclusions Approach I .. 79 Conclusions Approach II 80 REFERENCES 82 APPENDIX A. Survey Instruments 93 B. SEM Tests 100 C. Survey II Variable Key 102 D. Conceptual Model Construct Detail 103

PAGE 11

xi LIST OF TABLES Tables 1. Some antecedent variables added to TAM in previous research 9 2. Comparative constructs of DOI TAM & Credit Card Online Payments ................... 12 3. Respondent demographics.......................... 29 4. Myers Briggs Dichotomy Variable Conversion Values 31 5. Reliability Indices ... 35 6. Assessment of normality utilizing algorithm (Log10(8 X)* 1) .. 36 7. Measures of Fit... 37 8. Squared Multiple Correlations (R2) .... 39 9. Convergent Validity Construct Factor Loadings ...... 40 10. Squared Correlations Estimate with AVE on diagonal ..... 41 11. Regression intercepts of MBTI .. 43 12. Variable to Variable Path Beta Coefficients/Regression Weights .... 43 13. Mediation Test Statistics ....... 44 14. PLS Reliability Tests values ..... 48 15. PLS Squared Corre lations Estimate with AVE on Diagonal 48 16. PLS Path Coefficients ....... 49 17. PLS MBTI Coefficients and Derived Pair Indicator ..... 50 18. PLS Derived MBTI Types Per Constructs ... 50 19. Credit Card Website Perception Survey Respondent Demographics ... 59 20. Cluster Analysis Agglomeration Schedule ...... 60 21. Important credit card payment service features and personality aspects ..... 63

PAGE 12

xii 22. Important credit card payment service features for MBTI personality types 67 23. Feature perceptions by nonusers of online credit card payment services. ... 70

PAGE 13

xiii LIST OF FIGURES Figure 1. Conceptual Model... 22 2. Str uctural Equation Model...... 31 3. PLS Model diagram..... 47 4. SEM MBTI Dimension Indirect Effect on U... 53 5. SEM MBTI Dimension Direct Effect on PC, PU, & PEU... ... 54 6. Combined Linkage Dendrogram.. .. 61 7. Percentage of MBTI personality types in survey population....... 66

PAGE 14

1 CHAPTER I INTRODUCTION The Internet, having become an integral and ubiquitous part of society, culture, and business is having a transformational effect on all aspects of the modern world (Schell, 2007). One of the most critical areas having been transformed is the area of Inte rnet bill payment. The very nature of the Internet makes the use of physical currency as an exchange medium for online transactions untenable. Therefore business conducted online relies extensively on the use of either credit card or electronic check tr ansactions. The Internet and the credit card have grown to have a symbiotic relationship unlike any other market place and exchange medium previously seen. It seems only natural that credit card providers and consumer banks would further enhance that r elationship, as well as exploiting the power and convenience of the Internet, by developing online credit card bill payment service features. Credit card service websites parallel general banking websites and allow credit card holders access to extensive account management services, with both offering consumers a convenient method of paying monthly credit card bills. This study considers payment sites hosted by credit card provider organizations, and consumer banking services sites both as payment servic e sites for the purposes of consumer payment of monthly major credit card bills and has not distinguished between the two. Examining electronic commerce utilization, specifically credit card payment services, from an individual perspective is problematic due to the sheer volume of data that must be analyzed to ensure accuracy of the research model. Therefore, a classification schema

PAGE 15

2 for research populations that helps preserve individual differences is desired. Utilizing personality type indicators may provide such a classification technique. This research uses the accepted measurement tools of the Myers Briggs Type Indicator (MBTI) personality test, the Technology Acceptance Model (TAM), and the Theory of Diffusion as lenses to analyze user personalit y type and determine what if any affect that may have on the behavioral intention to use online credit card bill payment services as their payment channel. Studies of the effect of personality on the adoption of technologies have been used to analyze var ious online user behaviors. Research has shown that usage of the Internet, especially in the subcategories of social networking and interaction websites, has been correlated to personality types (Amichai Hamburger & Ben Artzi, 2003). However, as there ha s been little or no research done on the effects of personality on the adoption of e commerce generally and online credit card bill payment specifically, this study extends research into this new area. Credit card transactions play a pivotal role in online commercial transactions. Online credit card service sites and their consumer bank partners provide valuable conveniences to credit card holders meeting a growing demand for online account services, and providing savings through the value chain to the provider through transaction cost reduction (Langdon & Shaw, 2000). The initial start up costs of an online service sites and their accompanied development of sockets to consumer banks, coupled with ongoing expenses incurred for maintenance and upgrade repre sent a significant capital investment to the organization (Abramson & Morin, 2003). There are also legal considerations and expenses involved with providing these online service sties and bank partner service

PAGE 16

3 linkages. As transactions involve account acc ess and management, any possible breach of security on these sites leave the provider exposed to possible litigation if users incur damages. With these costs and risks to the credit card provider and its partners involved, it becomes paramount that the online payment services yield value to the organizations. Coupled with these issues are data which indicate a falloff in the number and rate of Bank Account Number Payments (BANP) which reflects a fundamental business problem of increased cost per transacti on (US Federal Reserve System, 2011). To increase value, it is not only beneficial, but vital, to increase the customers use of online payment services in order to increase the value to service providers.1 Therefore, there is benefit to an organization i n understanding those attributes that contribute to a potential to increase any particular credit card holders likelihood to use the online payment services. It is also important for service site providers and designers to understand the needs of their predominant users. Website designers who learn the personality profile of their clientele will be better able to focus on their needs (Albert, Goes, & Gupta, 2004; Amichai Hamburger, 2002; Guadagno, Okdie, & Eno, 2008). For many individuals, online busi ness transactions hold an elevated level of risk and uncertainty (Lee & Turban, 2001). Differences in personality type have an effect on a user's tolerance for uncertainty 1 It has been estimated that online payments may reduce cost per transaction to as low as $.01 each. Some institutions calculate an over the counter payment as costing $1.07 per transaction (Akinci, Aksoy, & Atilgan, 2004; Didio, 1998; Borzekowski, & Kiser 2007; Nath, Parzinger, & Bank, 2001, Ozdemir, & Trott, 2009). That would represent of savings of $1.06 per payment. This equals $1060.00 per thousand online users per month. With 576.4 million credit card bills being paid in the US each month, even a small increase in the percentage of users represents a substantial potential savings (Nilson Staff, 2010).

PAGE 17

4 and acceptance of risk, as well as threshold levels for perception of the intrusion of privacy (Myers, 1962). Once identified, user personality type and needs can by leveraged by designers to attract specific target markets increasing overall usage, improving retention, and fine tuning advertisement placements which all increase service site value and return on investment. This study is carried out through two approaches. The primary approach of this research, titled Modeling the Effects of Personality Type on Online Credit Card Payment Services, seeks to answer a number of research que stions. This primary approach addresses whether different personality types as measured by the MBTI personality test affect user perceptions of the usefulness, ease of use of online credit card payment services, and the compatibility of the features and services provided by online credit card service websites. To accomplish this the research utilizes several elements adopted from TAM. These TAM elements include Perceived Usefulness (PU), Perceived Ease of Use (PEU), and Usage (U). In addition to these metrics borrowed from TAM, the construct of Perceived Compatibility (PC) is employed from the Diffusion of Innovations theory. The research then examines whether those perceptions, as varied by personality type, have any effect on the individuals intent ion to use online payment services and features. To broaden and fill out this study, in a secondary approach titled, Discovering Desirable and Unwanted Online Credit Card Payment Service Features for Different Personality Types, this research presents a descriptive study of data collected independently of the primary approach that examines how personality type, as classified by the MBTI, distinguishes features that promote utilization of e commerce payment services by

PAGE 18

5 individuals and also which website f eatures are considered as unimportant by different personality types.

PAGE 19

6 CHAPTER II BACKGROUND AND RELATED WORK This research is cross disciplinary and draws upon research from psychologys fields of behavior science and personality types; information system s (IS) TAM and work from both marketing and IS in the area of Diffusion of Innovation. TAM In IS, TAM (and its many variants) stands out as a well documented and thoroughly tested methodology. It is used to understand and predict the behavior to adopt or the intention to use and actual usage of a given technology in the work place. TAM has been used to predict consumer intentions in a number of specific online environments (Lai & Li, 2005; Lederer, Maupin, Sena, & Zhuang, 2000). This study makes use o f the main constructs of TAM in its simplified form in order to ascertain the relationship between personality type and the usage of online credit card payment services. TAM finds that Perceived Usefulness (PU) and Perceived Ease of Use (PEU) of a technol ogy are the major determinants of a users Intent to Use (IU) and Usage (U) of a given technology (Davis, 1989; Davis, Bagozzi, & Warshaw, 1989). There have been a number of foundational studies that have found inclusion of TAMs central constructs of PEU and PU was essential to understanding what draws users to online mass customization websites, adding further weight to their inclusion here ( Ajzen & Fishbein, 1977; Lee & Chang, 2011). IU, a behavioral intent, has been shown to be one of the most powerful predictors of future behavior (Davis, 1989; Venkatesh & Davis, 2000; Venkatesh & Morris, 2000).

PAGE 20

7 Future intention to use impacts current behavior and is incorporated within the U measurement construct. This research attempts to describe a model of utilization behavior in order to benefit online credit card payment services designers efforts to retain and attract additional users. Determining the predictors of U can be put to use by designers to tailor their work to include attributes attractive to those personality types determined to have a higher probability of utilization (Davis, 1989; Davis, Bagozzi & Warshaw, 1989). TAM and its constructs of PU and PEU have been tested repeatedly using a wide array of differing applications and information technologies (Lederer, Maupin, Sena, & Zhuang, 2000). They have been examined as intermediate factors between various external variables and the outcome of usage of a given technology (Legris, Ingham, & Collerette, 2003). King and He carried out a met a analysis of 88 different studies testing the validity of the TAM model and found it to be generally a valid and robust model" (King & He 2006, p. 1). They concluded that both PEU and U were reliable measures across a wide spectrum of adoptions. They f ound in most cases that PU's effect on U was highly significant and included most of PEU's influence on U. But when focused on Internet applications found that PEU's direct effect on U was significant in itself and outside of its relationship to PU. This suggested that when used in measuring Internet U, models not incorporating the standard directional PEU to PU relationship might display higher indices of fit than those models in which it is included (King & He, 2006). This study tested both models and found that including PEU's mediating influence in the model did result in better fit statistics. Therefore we include the both the direct PEU to U and PEU to PU relationships. Studies have further identified four

PAGE 21

8 categories of modifications to the standa rd TAM model two of which are important to this study ( King & He, 2006). They are the common inclusion of prior factors, (in this study MBTI) and the inclusion of constructs from other theories selected to increase TAM's efficiency as applied to specific technologies, (in this case Compatibility from the theory of Diffusion of Innovations). The Extended TAM model (TAM2) was developed to explore and better understand the antecedent determinants of PEU and PU in the business environment (Venkatesh & Davis, 2000). They modeled influences of those constructs by socia l influence and cognitive instrument processes in the work place. The social processes they isolated are subjective norm, image, and voluntariness. The cognitive processes are job relevance, output quality, and result demonstrability. Their model is focused directly on business technologies. Other researchers as discussed below have found that in other areas of technological application, better models may be formulated utilizing various other antecedent determinants of PEU and PU, as well as additional c onstructs to stand beside them. Research positively supporting TAM has been conducted on the use and adoption of the World Wide Web with an exploration into those specialized antecedent external variables which predict PEU, and PU in this area (Lederer, Maupin, Sena, & Zhuang, 2000, Venkatesh & Davis, 2000). A number of studies further confirm that TAMs central tenets are substantively affected by exogenous variables giving credence to the validity of extensions of the TAM model (Agarwal & Prasad, 1997; Fishbein & Ajzen, 1975; King & He, 2006;

PAGE 22

9 Vijayasarathy, 2004). Table 1 shows some of this prior research and the independent variables or constructs added as extensions of TAM. Table 1. Some antecedent variables added to TAM in previous r esearch Domain New variables/constructs Reference e commerce trust and perceived risk (Lee & Chang, 2011; Pavlou, 2003) online games social influence, attitude, and flow experience (Hsu & Lu, 2004) online shopping compatibility, privacy, security, normative beliefs, and computer self efficacy (Vijayasarathy, 2004) mobile commerce Perceived risk, perceived costs, and compatibility (Chau, 1996, ; Wu & Wang, 2005) system administration Task Technology Fit model (Dishaw & Strong, 1999) Internet banking perceived credibility and financial and security risks (Wang, Wang, Lin, & Tang, 2003) organizational behavior social influence and cognition (Venkatesh & Davis, 2000) Thus we see a clearly demonstrable practice of extending TAM by both inclusion of constructs standing side by side with PEU and PU, as well as extension by addition of antecedent determinants of PEU and PU paralleling TAM 2's approach. Diffusion Of Innovations Choice of what metric to use in any given case is important and must fit that which is being measured. Since credit card payment service usage is both a technological application and a consumer service item it is felt that the appropriate metric tool blends the measurement capabilities of both TAM and the theory of Di ffusion of Innovations.

PAGE 23

10 TAM's main use is to gauge adoption or usage of an IT Artifact by workers to accomplish a business task (Venkatesh, Morris, Davis, & Davis, 2003). Online payment technologies are adopted by individuals to fulfill a consumer servic e need rather than complete a business process. In that context, the individual is making a decision which is motivated not only by the attributes of usefulness, and ease of use, but by additional social and cognitive personal pressures outside those of business environments. This study attempts to incorporate these additional needs into the model by the use of extending TAM with the inclusion of an element from the Theory of Diffusion of Innovations (Gefen, Karahanna, & Staub, 2003). The Diffusion of Innovations Model is a framework for how an innovation (a new idea, object, or practice) is spread through channels of communication over time, throughout a population. According to the model the four required elements for diffusion are: an innovation, a channel of communication, a time period, and a social system (Rogers, 2003). The theory relates adoption behavior to the following constructs: Compatibility of Technology, Complexity of Technology, Relative Advantage, Trialability, and Observability. The theory of Diffusion of Innovation has been applied to model the diffusion of technology with good effect (Fisher & Pry, 1971). It has been also employed directly to research online usage. In a particularly relevant example Parthasarathy and Bhattacherjee (1998) used the Diffusion of Innovation theory as a theoretical framework to study the continuance behavior of subscribers of Internet service providers. For the purpose of this research, the use of technology is, by users, to fulfill a consumer service need. Parallels can be drawn between some of the attribute constructs of TAM and those of the theory of Diffusion of Innovation (See table 2). Ease of use is similar to complexity of

PAGE 24

11 technology. Relative advantage relates to usefulness. Technology adoption is the equivalent of intent to use. These three parallel attributes have been found to be consistently valid and applicable to both diffusion and adoption in a number of empirical studies (Fliegel & Kivlin, 1966; Ostlund, 1974; Parthasarathy & Bhattacherijee, 1998; Tornatzky and Klein, 1982). The only elements of the theory of Diffusion of Innovation not directly paralleled in TAM are Compatibility of Technology, Trialability, and Observability. Online credit card payment services and banking bi ll payment features allow users to access their accounts and make use of payment services free of charge, from any location day or night. By this unrestricted access the user is unencumbered in their ability to experiment with usage of payment services and its observable outcomes, whether positive or negative, making Trialability and Observability unneeded within the framework of online credit card payment. Therefore of the Diffusion of Innovations' attributes only the construct of Perceived Compatibility (PC) need be added to enhance the predictive capability of this studys model beyond the TAM constructs of PEU and PU.

PAGE 25

12 Table 2. Comparative constructs of DOI TAM & Credit Card Online Payments Diffusion of Innovations Technology Acceptance Model Online Credit Card Payment Complexity of Technology Perceived Ease of Use Perceived Ease of Use Relative Advantage Perceived Usefulness Perceived Usefulness Compatibility of Technology NPA* Perceived Compatibility Technology Adoption Intent to Use Intent to Use Trialability NPA* NR** Observability NPA* NR** No parallel attribute ** Not relevant to subject Myers Briggs Type Indicator Personality tests are instruments designed to ascertain various attributes of the psychological profile and character of individuals. They fall in to two main categories: those which test higher level "traits" and those that test lower level primary "types". Some research considers traits as classifying the qualitative differences of individuals, while typ es classify them quantitatively (Bernstein, Clarke Stewart, & Roy, 2008). Trait based tests such as the Big Five, NEO PI, or Circumplex Seven consider broad upper level comprehensive traits (Digman, 1990). The MBTI, Kirton Adaptors Innovators, Eysenck's Personality Indicator, and other type tests like them are based on Jung's work founded on cognition which acts at a more primary psychological level (Jung, 1921, 1971; Kirton, 1976; Mershon & Gorsuch, 1988). Cognition deals with how an individual acquires information and makes decisions, and has been found to be more determinant in predicting behavioral intent and therefore utilization than those higher level traits dealt with by the Big Five (Paunonen & Ashton, 2001; Digman, 1990). Higher level trai ts have been found at times to be variable due to mood or situational condition independent of any change in core

PAGE 26

13 personality (Gendlin, 1964; Roberts & Walton, 2006; Rogers, 1957) Lower level core based types due to their foundation in deeper seated cognitive values are longitudinally stable, independent of mood or situation, and therefore more reliable and accurate indicators of a user's behavioral intent (Fuchs, 2001; Jung, 1921, 1933, 1971; Mershon & Gorsuch, 1988). The MBTI is used across a spectrum of scientific disciplines as a psychometric survey administered to subjects to ascertain their psychological attributes based on Jungs Psychological Types (Myers, 1962; Myers & McCauley, 1985). Jung (1921, 1971) extrapolated eight psychological types ba sed on permutations of two basic functions, extraversion and introversion, and four attributes, sensation, intuition, thinking, and feeling. The MBTI employs four diametric scales (Bayne, 1995). The pairs are Extraversion (E) and Introversion (I), Sensin g (S) and Intuition (N), Thinking (T) and Feeling (F), and Judgment (J) and Perception (P) (Myers, & McCauley, 1985). These are usually treated categorically; that is for any and each one of the four diametric scales, depending on the value of an individuals survey answers for that scale, the individual is assigned either one or the other of the diametric poles as a category, giving rise to 16 possible states, or Personality Types. The MBTI was chosen for this research as it is based on cognition with its better predictive capability, is generally seen as reliable and valid, and is the most widely accepted and recognized of the type tests used to measure cognition based on Jung's concepts (Carlson, 1985; Wheeler, Hunton, & Bryant, 2004). Furthermore, t he MBTI testing forms have undergone extensive peer review and been repeatedly confirmed as statistically accurate and internally consistent in identifying its' deeply theoretically based

PAGE 27

14 personality types. It is readily available as a turnkey testing in strument without the added burden to researchers of independent development and validation of a questionnaire. Finally, the researcher has firsthand working knowledge and prior experience in usage of the MBTI system. Personality Type And T he Internet Previous research examining interaction effects between personality and website usage typically examine a single trait or two and do not try to examine the whole personality. One e arly example of research on personality and Internet usage examined the rel ationship between extraversion and neuroticism, and Internet usage (Brenner, 1997; Eysenck, & Eysenck, 1975; Hamburger & BenArtzi, 2000; Sybil, Eysenck, Eysenck, & Barrett, 1985). Another research study implied that extroverted versus introverted person ality types account for much of the variance of Internet leisure and social services usage (Hamburger & Ben Artzi, 2000); showing that males having extraversion increased their use of leisure services, whereas for females, extraversion decreased use of social services (Hamburger & Ben Artzi, 2000). These early studies form the basis of the theoretical background linking user psychology, personality and Internet user behaviors. A later study addressed user behavior in light of a particular personality trait rather than a categorical personality type, focusing on the usage patterns of commercial sites by individuals with varying levels of a need for closure (Amichai Hamburger, Fine, & Go ldstein, 2004). This touches directly on a number of underlying aspects of both personality traits and types.

PAGE 28

15 The Five Factor Model based on trait theory is one of the major frameworks currently used in the field of psychology to describe human personali ty traits (McCrae & John, 1992). Tuten and Bosnjak (2001) make use of the Five Factor Model of personality, compared with cognition, to examine web usage preferences. Their findings indicated cognition a better overall predictor of web usage than any of the Five Factor Model traits. They also found that extraversion had no effect on the type of web usage, but did negatively affect the duration of use compared to introversion. This yields theoretical support for the preference of utilizing both personali ty typing over trait analysis and the usage of Jungian based models when studying web usage behaviors. Guadagno, Okdie, and Eno (2008) also used the Five Factor Model to discover that Openness to New Experience and Neuroticism were positive predictors of blogging. Amichai Hamburger (2002) and others; have provided motivation for further investigation into user personality type effe cts on Internet usage by finding that web designers tended to ignore the personality types of users when making design decisi ons to the determent of the end product ( Amichai Hamburger et al, 2004; Benet Martinez & John, 1998) Fuchs (2001) conducted a case study of personality of Real Estate agents through the use of the MBTI and followed up with personal interviews as confirm atory analysis of the indicator type findings. The results were incorporated into guidelines used to design a real estate agency ecommerce website. Fuchs (2001) makes reference to an earlier study by Norman (1999) which supported the case that when user s are given systems that are not only functional, but also serve to satisfies the internal needs arising from the users core personality, the usage of that system becomes intuitive, even satisfying needs of which the user is not even consciously aware. That stud y conclude d

PAGE 29

16 that the usage of personality type indicators in website design facilitate development of the conceptual model, identifying essential elements of the graphical interface, and help in defining usability, and f unctionality of the web desig n. As seen then p ersonality profiling is becoming a more accepted methodology in understanding web usage, for developing site design, and generally in busi ness to facilitate teamwork, customer relationship management and other business applications (To rres, 2013; Emergenetics, 2013). Personality profiling has reportedly been used by a wide variety of organizations including: Bank of America, Blue Cross Blue Shiel d, Cisco, Cricket, Denver International Airport, Great West Life, Hilton and Marriott hotel s, Microsoft, National Semiconductor (Malaysia), Pratt & Whitney (Canada), Sembawang Logistics (Singapo re), and the United States Air Force among others (Emergenetics, 2013). Utilizing personality type in e commerce research has been shown to be applicable yet its use is still limited meriting further attention. Current Usag e O f Online Credit Card Bill Payment Services The main aim of this research is to find determinants to usage that can be leveraged by designers marketing, and other management, to improve adoption and retention as discussed above, however, other issues have also been found that relate to the relevance of this research. It is felt that a consumer's payment of their monthly credit card bill is distinct and different from other bill payments they may make for one major reason. Their credit card bill payment history is directly and immediately tied to their credit score. Consumer credit scores have increased in importance to the point where they affect nearly all facets of a consumers' life. The time when a credit score was merely used as a reference of

PAGE 30

17 credit worthiness is long gone. Now, credit scores can effect who will allow you to rent from them, whether you will be employed by certain employers, what you pay for insurance premiums and much more. If a customer misses a water or electric payment they may have their service interrupted. This is an isolated event, and once the balance is brought up to date rarely has any further effect. If however they miss thei r credit card bill payment this can lower their credit score effectively economically handicapping them, and influencing wide areas of their lives. This makes payment of credit card bills a unique activity and payment behavior of those bills an extremely important issue. Available statistics show that 13.8 percent of consumers paid some credit card bills online (Hayashi, & Klee, 2003). When compared to the 62 percent of US consumers who had Internet access in 2003, it is clear there is considerable room for expansion of the number of adopters of online Credit Card bill payment (US CESRC, 2010). This is an important area of finance since US consumers carry $835.5 billion in credit card debt (Federal Reserve Bank of Boston, 2010). That total was distributed across 576.4 million individual credit cards holders in the US (Nilson Staff, 2010). On average each adult in the US had 3.5 credit cards issued in their name as of the beginning of 2009 (Federal Reserve Bank of Boston, 2010). In 2001, 19 percent of all US households reported using some type of online banking and this value rose to 34 percent in 2004 and to 53 percent in 2007 (US Census Bureau, 2011). The focus of this study is online payment of monthly credit card bills which are primarily transact ed with online BANP instruments. In 2008, the mean number of bills paid monthly using online BANP instruments was 4.1 per consumer, but fell to 3.0 bills paid monthly in 2009, a decline of 26.1 percent over one year (US Federal

PAGE 31

18 Reserve System, 2011). Dur ing this same period 73.4 percent in 2008, and 56.3 percent in 2009, of all US consumers had used an online BANP instrument in the payment of some type of monthly bill including credit cards, a 17.0 percent decrease, attributed to a decrease in consumer co nfidence of the security of personal information maintained by online banking websites (US Federal Reserve System, 2011). The decline in online payment instruments for all bill types discussed above shows that users who previously adopted online payments, are not continuing use of this service. This infers an underutilization of this type of payment service suggesting that a large percentage of consumers have yet to adopt or readopt online payment instruments for cyclical credit card bi lls. Online credit card payment service's design and the menu of features they offer have been linked to both adoption and continuance behavior. As a result the set of features offered by these service sites become highly important to the website designer ( Albert, Goes, & Gupta, 2004; Amichai Hamburger, 2002) Personality type provides designers with a powerful lens with which to determine design elements and features to retain current users as well as attracting new adopters. Scope A nd Contribution Credit c ard debt has be en shown to represent a vast amount of money (Federal Reserve, 2010). Any contribution to understanding the online payment behavior of users is therefore important and can be employed to help design sites that will attract more potential adopters. As dis cussed above, data indicates that there is a considerable segment of the consumer population that could be persuaded to adopt with subsequent benefits of efficiency and profit accruing to both financial institutions and users.

PAGE 32

19 A body of research exists on TAMs relationship to personality types, and general I nternet usage, service usage, user perception, usage patterns and other related topics as discussed in the literature review section of this paper. This study extends and builds upon that previous work in the area of TAM, personality types, and website usage into the unstudied area of online credit card bill payment. This area represents a gap in the research literature as indicated by the lack of papers addressing this specific topic, and as evidenced by the deficit of statistical data existent in the financial, and governmental consumer payment behavior databases. For the first time data will be collected specifically indicating personality types relationship to adopter and usage behavior of online credit card bill payment. In addition it will contribute a body of data dealing directly with the adoption of online noncash payment instruments for cyclic credit card bill payment, which is currently lacking. Additional it will aid in the identification of payment site service and feature preferences which might be utilized in attraction of new and retention of current users. A further contribution is found in that any significant findings will be available to be applied by website designers to enhance current and future sites. User Centered Design uses Personas, Scenarios, and User Cases as analytical tools to design websites better matched to the needs, and behaviors of the user (Cooper, 1999). Personas are created by carrying out field research of either a qualitative or quantitative nature, and then applying the findings to create as accurate of a user model as possible (Brickey, Walczak, & Burgess 2010; Pruitt, & Grudin, 2003). P ersonas have been shown to be a beneficial tool for systems and si te development yet often provide too course a screen designa ting all users into a very small number of prototypes (typically 3 or 4) (Dharwada,

PAGE 33

20 Greenstein, Gramopadhye, & Davis, 2007; Mulder & Yaar, 2007; Pruitt & Adlin, 2006). To capture more individual nuances in e commerce perceptions and utilization, a finer grained sieve is needed. This study is not a design methodology in itself but rather carries out field research specif ic to credit card payment service a dopters. Usage of any findings about users personality type from this study will contribute to the UCD designers ability to develop personas that are more accurate, and scenarios and user cases that are more effective. Findings from the secondary resea rch conducted help to identify specific design and feature preferences to be employed by those designers.

PAGE 34

21 CHAPTER III APPROACH I: MODELING THE EFFECT OF PERSONALITY TYPE ON ONLINE CREDIT CARD PAYMENT SERVICES Conceptual Model This research attempts to methodically codify personalitys effect on a specific IS technologys adoption and use. This model extends TAM first by including the construct of PC from the theory of Diffusion of Innovation (Rogers, 2003; Wu & Wang, 2005). Second, it extends TAM fur ther by postulating that user personality can be used as an antecedent determinant of and having a prior and direct effect on user perceptions of ease of use, usefulness, and compatibility of online credit card paym ent services These extensions are em ployed in place of TAM 2 extensions, as this research postulates that the MBTI and compatibility constructs better model the social and cognitive aspects for personal credit card payment behavior than TAM 2's model for business function applications use. Therefore this research attempts to systematically look at how variations of personality type affect the credit card holder users percep tions of online payment services, and their use of them. A large body of confirmatory work in TAM indicates the val idity and applicability of PU and PEU both generally and specifically to the adoption of online technologies (Moon & Kim, 2001). Figure 1 depicts the model following the classical form of TAM using the constructs of PU and PEU, while complying with the si mplified version as per Adams, Nelson, and Todd (1992), eliminating the construct of Attitude Toward Using, and Intent to Use, and employing Usage as the dependent variable (Lu &

PAGE 35

22 Gustafson, 1994; Venkatesh & Davis, 2000; Venkatesh & Morris, 2000; Venkatesh, Morris, Davis, & Davis, 2003). Figure 1: Conceptual Model Research Hypotheses Twelve Hypotheses have been derived. Work in the field suggests that differing personality profiles have some association to general Internet usage and that further study of personality type and Internet service usage is critical to the ongoing development and improvement of Internet based services (Amichai Hamburger & Ben Artzi, 2003, Amichai Hamburger, 2004, 41, Hamburger & BenArtzi, 2000) Additional research shows that specific personality profiles influence the perceptions of users and that Internet usage patterns differ individually related to content and services (Hills & Argyle, 2003). Research into the theory of Diffusion of Innovati on indicates that there is

PAGE 36

23 relevance to the individuals personality being a factor in adopter behavior (Rogers, 2003). The MBTI Judging dimension describes individuals preferring self regulated managed life styles. Seeking orderly systems to plan thei r lives and make decisions, they are most comfortable with situations that achieve closure. Their life styles are typically well structured, organized, and scheduled (Myers, 1962; Myers & McCauley, 1985). Judgementals prefer to be settled, having external factors under control and in a completed state. Their defining characteristics are; "systematic, scheduled, organized, methodical, make long and short term plans, avoidance of last minute stresses" (Myers, 1998). Judgmental types prefer to know the w hat and how of things, compared to the perceptive types that are mainly concerned with why (Myers & Myers, 1995). Online direct payments are almost real time. The action is complete, unlike a check in the mail taking days to arrive and longer to re ceive confirmation of clearance. It is highly compatible with the need for being settled which Judging personality type prefers, and they are provided with confirmation at the completion of the transaction providing closure. This supports a perception that online payments are useful to this type, which leads to the hypothesis: H1 a : The Judgment type will have a positive effect on consumers PU of online credit card payment services. H1b: The Judgment type will have an indirect positive effect on consumer s' U through its' interaction with PU of online credit card payment services. Users can make use of online payment services according to their own schedules day or night. It can be extrapolated that for this reason Judgmental users would be more

PAGE 37

24 likely to find paying their credit card bills online both easier to use and more compatible with their life styles. The following two hypotheses follow: H2 a : The Judgment type will have a positive effect on consumers PEU of credit card payment services H2b: The Judgment type will have an indirect positive effect on consumers' U through its' interaction with PEU of online credit card payment services. H3 a : The Judgment type will have a positive effect on consumers PC of online credit card payment services H3b: The Judgment type will have an indirect positive effect on consumers' U through its' interaction with PC of online credit card payment services. Those falling on the extraverted side of the MBTI Extraversion/Introversion scale are seen to be actively engaged in the outer world. Involved with activities and people, they are energized by both. They are action takers, and willing to take risks at an acceptable level, with their attention placed squarely on the world around them (Myers, 1962; Myers & McCa uley, 1985). They are characterized as; "attuned to the environment, prefer to communicate by talking (rather than writing), work out ideas by talking them through, learn best by doing, have broad interests, sociable and expressive, ready to take initiative in work and relations (Myers, 1998, pg. 9). As seen in the literature on Internet Banking usage, rational judgments about user trust in the credibility of a website as to security and risk reduction relate to user

PAGE 38

25 perceptions and intent (Friedman, Ka hn, & Howe, 2000; Hoffman, Novak, & Peralta, 1999; Wang, Wang, Lin, & Tang, 2003). The use of direct bank account transactions, as any online financial transactions, has an associated perception of risk to the user (Lee & Turban, 2001). It is postulated that extraverted types would take this in stride easier than those less amenable to risk and less accustomed to taking action. They easily accept risk on both physical and emotional levels as a natural expression of their unreserved nature. Outgoing type s with many varied interests are often leaders and opinion makers, tending to be early adopters of ideas and technologies (Fisher & Pry, 1971; Rogers, 2003). It is expected that the stronger the Extraversion scores th e more likely will be the likelihood to use online payment services and the greater usefulness perceived. This is based on the findings that extroverted individuals are more inclined to assume risk, and we postulate that they will perceive fewer impediments to PEU (Bayne, 1994; Myers & McCau ley, 1985). It is felt that these attributes, much like the characteristics of Judgment types contribute to the likelihood that an individual would find online payment systems easier to use and highly useful leading to H4 and H5. H4 a : The Extraversion t ype will have a positive effect on consumers PU of online credit card payment services H4b: The Extraversion type will have an indirect positive effect on consumers' U through its' interaction with PU of online credit card payment services. H5 a : The Ex traversion type will have a positive effect on consumers PEU of online credit card payment service s. H5b: The Extraversion type will have an indirect positive effect on consumers' U through its' interaction with PEU of online credit card payment services

PAGE 39

26 While introverted users prefer less direct contact and communicate by writing, mailing checks to pay credit card debt satisfies these desires with much less perceived risk [Myers, 1998]. It is thought that the Extraversion factor of personality will b ecome more important as we see the fall off of trust in the security of online financial sites (Federal Reserve Bank of Boston, 2010) and so H6 is postulated: H6 a : The Extraversion type will have a positive effect on consumers PC and U of online credit card payment services H6b: The Extraversion type will have an indirect positive effect on consumers' U through its' interaction with PC of online credit card payment services. Sensing types tend to rely on actualities, and thrive on exactitudes. They depend on "what is" rather than on the "what is possible" that intuitive types focus on (Myers, & Myers, 1995). Persons of the sensing type depend on their own five senses rather than relying on what others may say or feel about external factors (Bay ne, 1994; Myers, 1998; Quenk, 2009). Intuitives being grounded in inspiration are more likely to be innovators than Sensing types, but that also means they are more likely to be frivolous and change ideas and methods more easily and often than their Sensi ng counter parts (Myers & Myers, 1995). Sensing types typically prefer custom and tradition. This plays out in establishing processes in their lives that are straight forward and provide positive feedback as to results (Myers, 1962; Myers, 1995; Myers & McCauley, 1985; My ers & Myers, 1998). T his meshes well with online payment services that provid e real time feedback by way of instant confirmation of payment This type would feel this method both easier and more compatible than one in which the action i s separated from the

PAGE 40

27 confirmation by a longer stretch of time a nd they would find the instant feedback highly useful. This leads to the following three hypotheses: H7 a : The Sensing type will have a positive effect on consumers PU of online credit card payment services H7b: The Sensing type will have an indirect positive effect on consumers' U through its' interaction with PU of online credit card payment services. H8 a : The Sensing type will have a positive effect on consumers PEU of online credit car d payment services H8b: The Sensing type will have an indirect positive effect on consumers' U through its' interaction with PEU of online credit card payment services. H9 a : The Sensing type will have a positive effect on consumers PC of online credit card payment services H9b: The Sensing type will have an indirect positive effect on consumers' U through its' interaction with PC of online credit card payment services. The Thinking/Feeling dimension deals with the making of decisions. Thinking types ap proach the decision making process from the point of view of what is true versus what is false. For the Thinker type everything is about impersonal objective truth, while to the Feeling type everything is personal. Thinking type individuals will associa te online systems with their logical, computer based underpinnings, and therefore feel that using online systems for paying their credit cards is more logical and precise and therefore more useful. And thus: H10 a : The Thinking type will have a positive ef fect on consumers PU of online credit card payment services

PAGE 41

28 H10b: The Thinking type will have an indirect positive effect on consumers' U through its' interaction with PU of online credit card payment services. The Feeling type person approaches the pro cess from the view point of what is agreeable versus disagreeable (Myers, 1962; Myers & Myers, 1995). What is perceived as disagreeable is harder to find of use. The Feeling type may be more likely to prefer the more personal process of filling out the bill, writing the check by hand, and licking to seal envelopes, or paying their credit card in person face to face at local bank branch (Myers, 1998; Myers & McCauley, 1985; Myers & Myers, 1995). They should find the online system much less compatible with their life style when compared to Thinking types who would be very comfortable making payments in the online environment. By nature therefore Thinking types will find online payments easier to use than their Feeling counterparts. Hence, the final two hypotheses: H11 a : The Thinking type will have a positive effect on consumers PEU of online credit card payment services H11b: The Thinking type will have an indirect positive effect on consumers' U through its' interaction with PEU of online credit card payment services. H12 a : The Thinking type will have a positive effect on consumers PC of online credit card payment services. H12b: The Thinking type will have an indirect positive effect on consumers' U through its' interaction with PC of online credit card payment services.

PAGE 42

29 M ethodology Data was gathered utilizing a self administered online survey hosted on the Survey Monkey website Data was actively collected starting in September of 2011 and terminating in June of 2013. The survey was offere d to perspective respondents as part of the University of Colorado Denver Business School's "Business Engagement Assignment Program". This is a program run by Professor Stefanie Johnson of the business school to encourage student participation in academic and research activities outside of their enrolled classes. Students enrolled in undergraduate programs are required to participate in one activity over the course of each semester. They select their activity from a list of prospective surveys and other research projects. Upon completion they are given credit for an assignment by their respective instructors per the class's syllabus. Four hundred thirty six students responded to the survey administered online. No screening of the respondents took place and no identifiable information was recorded. It has been found that TAM is invariant for differences of gender, age, and IT efficacy, and therefore has a broader applicability than the sampling frame work would imply if TAM were not in use (Lai & Li, 20 05). Demographics for the respondents are shown in Table 3. Table 3. Respondent de mographics Category (mean, range if applicable) Sex Male 49.8% Female 50.2% Age (18 61) (Mean age = 28) 18 29 70% 30 39 16.5% 40 or Older 13.5% Education (Max Completed) HS 39.3% AS/AA 10% BS/BA 35.2% Graduate degree 15.5% Employment Unemployed 23% Part time 36.5 % Full time 40.5% Marital status Single 62.4 Married 29.7% Prev. married 7.9% Income ($45,699, $0 $100,000+) < $30K 53.7% $31 59K 26.2% >$59K 20.1%

PAGE 43

30 The survey is in two sections. Section A, composed of 19 questions focuses on collecting data for the four variables; PU, PEU, and PC. These questions are based on a summated rating Likert Scale, employing a seven point agreement type configuration. This is adapted from the field of psychology where this type scale is commonly used to measure attitudes and intentions which closely fits the measurement constructs our study is testing. (Blumberg, Cooper & Shindler, 2008; Likert, 1932; Spector, 1992). Section B ascertains the personality type of the subject. It is comprised of the standard MBTI Form M self scorable test. Structural Equation Modeling (SEM) is employed as the primary analytical tool due to the exploratory nature of this research, and as it is particularly well suited to analyze the type of data in use, as constructs are measured by multiple indicator values (Hair et al, 2005). SEM allows evaluation of a flexible set of variables including continuous and discrete variables, and allowing use of nondirectly measured variables (latent variables) (Hair et al, 2005). The SEM model for the research is shown in Figure 2 with an explanation of the value ranges for the corresponding MBTI variables displayed in Table 4. As seen in the SEM Model Diagram (Figure 2) this research employs MBTI dimensional variables (treated as direct variables), the latent variables PU, PEU, PC, and dependent variable IU. As an additional line of analysi s of the data a post hoc parallel analysis has been carried out using the Partial Least Squares (PLS) methodology. That analysis is discussed in a separate section below.

PAGE 44

31 Figure 2: Structural Equation Model Table 4: Myers Briggs Dichotomy Variable Conversion Values Dichotomy Axis Bi categorical Values Extraversion (E) Introversion (I) E= 1 to < 0 I= 0 to +1 Sensing (S) Intuition (N) S= 1 to < 0 N=0 to +1 Thinking (T) Feeling (F) F= 1 to < 0 T =0 to +1 Judgment (J) Perception (P) J= 1 to < 0 P=0 to +1

PAGE 45

32 CHAPTER IV ANALYSIS SEM Model Analysis This study is an exploratory investigation of an area of user behavior in which no known causal model currently exists. The model considered in this research is used to test the hypothesized directional correlation relationships between the models' latent variables, comprising the model's constructs and are all reflective causal structures (Edwards & Bagozzi, 2000). As path analysis tests directional relationships only between manifest variables, and confirmatory factor analysis tests nondirectional relationships between latent variables, neither of those subtypes of SEM analysis were utilized (Shah & Goldstein, 2006). Both absolute and incremental fit indices were considered i n determining the fit of the models. The Maximum Likelihood ratio (ML) was used as the estimation method for all models. Assumptions of the Structural Model The studies structural model specifies the following relationships: MBTI ->PC; MBTI -> PU; and MBTI --PEU. The four exogenous variables EI, SN, FT, and JP represent the MBTI direct measurements for the individual's personality type. This personality type exists as the core psychological makeup of an individual prior to any PU, PEU, or PC attitudes they may develop upon taking up the usage of technologies. Therefore there is temporal precedence evident. As discussed in the introduction and background sections above there is a body of work establishing an association between personality and behavioral intent, and usage of both technology in general, and specifically web based applications and services. This serves

PAGE 46

33 in fact as the theoretical underpinnings of this study as addressed in the Personality Type and the Internet section of Chapter II. Thus the assumption of association is met. Normality of the data is generally discussed below in the Normality section below. After the application of noted transformation algorithm acceptable normality was met. The specification of the casual relat ionship between the exogenous variables MBTI and the endogenous constructs was discussed above. As to the meeting correctness assumptions of specification of directional causality of the endogenous variables to endogenous variables PC -->U, PEU ->PU, PEU >U, and PU ->U, this study relies on the sound theoretical foundation of TAM and DOI. Assumptions O f T he Reflective Measurement Model This study's model employees four reflective constructs, PC, PU, PEU, and U. This is theoretically justified in that eac h of these are endogenous variables and represent a hypothetical construct in the conceptual model with its respective error term represented in the SEM. Further each of these constructs are latent variables that exist independent of any measurement used. Variation in the latent constructs would cause the indicator items to vary. Any variation in the measurement instrument would not result in variation of the construct, and the measurement indicators are caused by and manifestations of the construct. Al l the measurement indicators share a common theme and are interchangeable. Any addition or deletion of any particular measurement indicator would not change the concept of the construct. The empirical justifications for reflectivity as indicated by the immediate following sections are that all the measurement indicators are highly and positively correlated (see

PAGE 47

34 Chronbach's alpha values in Table 5), exhibit reliability and internal consistency, and show theoretical content validity through convergent and discriminant metrics. Reliability Of T he Constructs To determine the fit of the measurement model, the data for the each of the four constructs were tested for reliability using both an internal consistency methodology Chronbach's alpha, and a split half re liability methodology the Spearman Brown coefficient. Cronbach's alpha was evaluated using a value of .70 as the commonly acceptable lower limit (Hair et al, 2005). All four constructs exceeded the minimum value for the Cronbachs alpha reliability test as displayed in Table 5. The SpearmanBrown coefficient heuristically has a lower limit of .6 as a cut off for exploratory work, .8 and higher is considered to show adequate reliability for confirmatory research, and .9 or higher is considered very good reliability (Garson, 2001). The constructs: PEU, PC, and Usage all scored above .8. The lowest Spearman Brown Coefficient was .767 for PU, which is well above the minimum of .6 used in exploratory work, and close to an adequate reliability of .8.

PAGE 48

35 Table 5. Reliability Indices Construct Chronbach's alpha Spearman Brown Coefficient Equal Length/Unequal length Number of Observed variables in Construct Perceived Usefulness .834 .767/.773 5 Perceived Ease of Use .835 .835/.835 4 Perceived Compatibility .900 .887/.891 5 Usage .881 .883/.886 5 Normality O f T he Constructs The constructs were tested for normality of distribution. All four constructs were found to have a negative skew, and to exhibit some kurtosis. To deal with this a commonly used data transformation was employed to correct for negative skewness of J shaped distributions. The transformation algorithm (Algorithm 1) used was reflective, and used a compensation factor for reversion to original ranking. New X = Log10(8 X)* 1 (1) After transformation all constructs fell within the commonly accepted limits of 1 to +1 for skew and kurtosis values for normality as displayed in Table 6 in (Garson, 2012, Tabachnick & Fidell, 2007).

PAGE 49

36 Table 6. Assessment of normality utilizing algorithm (Log10(8X)* 1) Variable Skew Kurtosis EU1 .70895 .54547 EU2 .02774 .75854 EU3 .06149 .63650 EU4 08151 .85419 PU1 .85301 .37135 PU2 .49970 .92798 PU3 .58177 .90982 PU4 .25647 .98315 PU5 .43470 .98760 PC1 .51872 .71473 PC2 .34460 1.0272 PC3 .76498 .58538 PC4 .52964 .72693 PC5 .34276 .84655 U1 .54601 .82954 U2 .99615 .14084 U3 .76750 .78295 U4 .80998 .45958 U5 .69980 .62146

PAGE 50

37 Analysis O f Model Fit Using SPSS's AMOS application, a series of SEM models were developed and tested. The optimal model, which is reported in this article, achieved an absolute model fit measured with a Chisquare of 1236.75 with 197 degrees of freedom (DF). Comparing the Chi square to the DF, a CMIN/DF ratio of 6.28 is derived as shown in Table 7. This value exceeds the upper limits of the range but is not uncommon for a model of this complexity (Carmines & McIver, 1981). The Chi square at a probability level equal to .000 is uninformative and thus with the high CMIN/DF ratio leads to the necessity for consideration of the relative fit indices for establishing the models fit (Browne & Cudek, 1993). Incremental fit measured by both the Relative Fit Index (RFI) of .758, a nd the Comparative Fit Index (CFI) of .819 is acceptable (Bentler, 1990). The model's root mean square error of approximation (RMSEA) of .110 at the upper limits of reasonable fit, is a minimally acceptable value for exploratory research (Browne & Cudek, 1993). The last indicators of fit employed are the Akaike information criterion (AIC), and the Browne Cudeck criterion (BCC), measured at 1392.75 and 1401.46 respectively (Akaike, 1973, Browne & Cudek, 1989). These two values are moderately high, (highe st of any model tested) and so viewed as respectable given the high complexity of the model involved (Browne & Cudek, 1989). Table 7. Measures of Fit Model CMIN DF CMIN/DF RFI CFI RMSEA AIC Default 1236.751 197 6.277 .758 .819 .110 1392.75

PAGE 51

38 R2 A nd Squared Multiple Correlations The estimates for R2, interpreted in SEM as Squared Multiple Correlation (SMC), are displayed in Table 8. For a latent dependent variable in a recursive model R2 or SMC represents the variance in the model explained by all predictor variables by which it is directly affected (incoming arrows in the SEM diagram) (Byrne, 2010; Garson, 2012). The value for U in this model indicates that 67.5% of its variance is explained by its predictor variables PU, PC, and PEU. For indicator variables of an SEM model SMC provides an estimate of communality. Communality is a measure of the percent of variance for an indicator explained by the latent variable. Communality can be understood as an additional measure of the reliability of an indicator (Garson, 2012). For instance PC2, one of the indicators of PEU has an SMC of .760. Therefore, PEU2 is said to be 76.1% reliable as an indicator of PEU. This is the strongest SMC of any of the indicator variables. The weakest indicators in our model are PU4 at .229 and U1 at .255. In the final round of analysis the SMC of these indicators would be considered along with measures of convergent validity, and importance of the variable to the theoretical model for possible elimination from the SEM. Of these the closest scrutiny would be given to PU4 and U1 as they have little more than 20% reliability.

PAGE 52

39 Table 8. Squared Multiple Correlations (R2) Estimate U .675 PEU1 .373 PEU2 .760 PEU3 .708 PEU4 .491 PU1 .755 PU2 .744 PU3 .535 PU4 .229 PU5 .359 PC1 .716 PC2 .760 PC3 .450 PC4 .701 PC5 .655 U1 .255 U2 .719 U3 .429 U4 .748 U5 .644 Convergent Validity To determine convergent validity each of the four constructs' factor loadings were examined and the Average Variance Extracted (AVE) assessed. The commonly accepted lower limit for factor loadings is considered to be .5 or higher and in the best case .7 or higher, with AVE values expected to be .5 or higher to indicate sufficient convergent validity (Byrne, 2010; Garson, 2012). All of the models constructs exceed the loading minimum except PU4 with a loading of .48 with U1 at .51 the next lowest. (See Ta ble 9). These two indicators were already under scrutiny due to their low SCM values. U1 met the minimum but with PU4 failing another measure it was decided to eliminate this indicator from the model. It was determined to keep U1 even with its low SMC a s it met the minimum for convergence and yielded a better RMSEA value when left in the model.

PAGE 53

40 Overall each of the constructs AVE values exceeded .5 further indicating adequate convergent validity (See Table 9). Table 9. Convergent Validity Construct F actor Loadings Factor Construct Loading AVE PEU 1 .610 PEU 2 .872 PEU 3 .841 PEU 4 .701 PU 1 .869 PU 2 .862 PU 3 .732 PU 4 .482 PU 5 .599 PC 1 .846 PC 2 .872 PC 3 .671 PC 4 .838 PC 5 .809 U 1 .505 U 2 .848 U 3 .655 U 4 .865 U 5 .802 PEU .668 PU .687 PC .721 U .689 Discriminant Validity To determine Discriminant Validity this study uses the common approach of a Squared Correlations Estimate Matrix with AVE on the diagonal. The AVE value is higher for each construct compared to the square of the inter construct correlation relationship values (See table 10) (Byrne, 1995; Garson, 2012). This indicates a positive divergence for the constructs.

PAGE 54

41 Table 10. Squared Correlations Estimate with AVE on diagonal PEU PU PC U PEU .668 PU .328 .687 PC .460 .635 .721 U .374 .538 .646 .689 Structural Measurement Analysis The main relationships being evaluated in this research are first, those between the individual MBTI manifest variables (EI, SN, FT, and JP), and the latent variables PEU, PU, and PC, and second, the latent variables PEU, PU, and PC's relationships to the independent latent variable U, and third PEU's mediation effect on PU's relationship to U. The values of the MBTI variables have been converted to a scale centralized on a value of 0. Any negative values for regression weights indicate values for the left hand member of the MBTI pairing, and positives indicate values for the right hand members. The four manifest MBTI variables are st andardized at .00 intercept, and when unstandardized exhibit the intercepts displayed in Table 11 as they relate to the intermediate constructs. The beta coefficients/regression weights for each path (displayed in Table 12) measure the relationships of the SEM model. These SEM path coefficients/regression weights (Standardized Total Effects) are used as approximations for the relative connection strength between the MBTI variables and the TAM and Diffusion of Innovation Theory constructs. For the PU construct the E, T, S, and J poles have the strongest to weakest relative effect respectively on the PU construct, indicating that the overall MBTI type which perceives the greatest usefulness from using a credit card payment service is the ESTJ type. The M BTI pole relations for PC listed by decreasing magnitude are then T, E, J, S, indicating that the overall MBTI which perceives the greatest compatibility with

PAGE 55

42 using a credit card payment service is again the ESTJ type. The MBTI pole relations for PEU l isted by decreasing magnitude are S, J, E, T, indicating that the overall MBTI which perceives the greatest ease of use when using a credit card payment service is also the ESTJ type. The strongest of any of the observed MBTI manifest variables' relatio nships to the latent TAM construct variables is the effect of S on PEU with a regression weight of .083, followed by T on PC with a weight of .078, E on PC with a weight of .071, and E on PEU again with a weight of .071. The next strongest are J to PC with a weight of .0652, and E to PU and T to PEU with weights of .065 and .052 respectively. The next relationships evaluated are those between the latent TAM constructs PU, and PEU and the Diffusion of Innovation Theory construct PC, to the dependent latent variable U. These are the endogenous to endogenous relationships measured by the beta coefficients and are all reported using standardized weights. The strongest of these is PC to U with a value of .738, and is significant with a P value of less than .001. That is followed by PU to IU at .285, significant at P less than .001, and the weakest PEU to U at .098, significant with P value of .031. The relationship between PEU and PU is .536, also with significance at P value of less than .001. The effect of PC on U is roughly 2.6 times greater than PU's effect on U, and 7.5 times greater than PEU's effect on U. The PEU to PU relationship is strong with a value of .536. The relationships analyzed to this point are all direct relationships. Table 1 2 shows the values for the indirect relationships between the observed MBTI variables and the dependent latent U variable. The EI to U value indicates the E pole factor is dominant over the other MBTI type factors effect on U. The JP to U and SN to U

PAGE 56

43 e stimates both show the significant effect of the S and J types on U. Overall the greatest indirect effect on the variance of U of any combined MBTI type is that of ESTJ. Table11. Regression intercepts of MBTI MBTI pair Standardized Intercept Unstanda rdized Intercept EI .00 .12 SN .00 .05 FT .00 .15 JP .00 .17 Table 12. Variable to Variable Path Beta Coefficients/Regression Weights JP FT EI SN PEU PU PC U PEU .023* .052* .071* .083* PU .047* .049* .065* .047* .536** PC .065* .078* .071* .035* U .064* .077* .078* .047* .098* .285** .738** significance = <.05 ** significance = <.001 Mediation Testing The model is a recursive causational model, which includes one mediating variable which is PEU's to PU's relationship. The model specifies and expects to see no other mediation. To test PEU's mediation effect on U and to confirm that no other variables are acting as mediators a series of tests were run. These included the Sobel, Aroian, and Goodman tests. The results confirm PEU to PU mediation to a significance of better than .05 for all three tests. All other relationships tested showed nonsignificant P values confirming that no other variables are behaving as mediators (See table 13).

PAGE 57

44 Table 13. Mediation Test Statistics Relationship Sobel Test P Value Aroian Test P Value Goodman Test P Value EI ->PC ->IU 0.04900 0.08400 0.19800 0.00800 0.58317 0.55945735 SN ->PC ->IU 0.11800 0.09100 0.19800 0.00800 1.29493 0.19498146 FT ->PC ->IU 0.11900 0.09300 0.19700 0.00800 1.27785 0.20093297 JP ->PC ->IU 0.11400 0.08600 0.19700 0.00800 1.32366 0.18525256 EI ->PU ->IU 0.05200 0.07400 0.16500 0.01200 0.70179 0.48165367 SN ->PU ->IU 0.11000 0.09100 0.17900 0.00900 1.20656 0.22701373 FT ->PU -> IU 0.05700 0.09200 0.17800 0.00900 0.61926 0.5352224 JP ->PU ->IU 0.10700 0.08500 0.17800 0.00900 1.25628 0.20843377 EI ->PEU ->IU 0.05200 0.07400 0.16500 0.01200 0.70179 0.48165367 SN ->PEU ->IU 0.13800 0.08000 0.16500 0.01200 1.71158 0.08615156 FT ->PEU ->IU 0.12400 0.08200 0.16400 0.01200 1.50302 0.13180731 JP ->PEU ->IU 0.08100 0.07600 0.16400 0.01200 1.06256 0.28669408 PEU ->PU ->IU 0.59400 0.04600 0.14100 0.01000 9.52292 0.00000 Additional Variables Tested Several additional models were tested adding Age, Gender, and Income to the SEM. In the best case the addition of these variables showed a very slight improvement in model fit moving RMSEA from .110 to .105. However, in that case all beta coefficients were low and non significant in all path relationships. The strongest were PC to Gender and PU to Gender at .032 and .033 respectively (See Appendix B ). Therefore Age and Income were dismissed from the model, while Gender exhibiting two of the stronger relationships of the additional variables was subjected to a further test to finalize determination of variance insensitivity to that variable. Testing F or Variance D ue T o Gender As previously mentioned TAM has been found to be invariant for differences of gender (Lai & Li, 2005). To confirm this for the sake of this study, and determine whether Gender should be kept in the model or dismissed, a n Independent Samples T test

PAGE 58

45 was run to determine if there is was any difference in the variance of the dependent variable U due to gender The Male group had a mean of 5.88 (see Group Statistics Table Appendix II), with a Std. Error of .0787. The Female group had a mean 6.03, with std. error of .0796. The significance for equal variances assumed was .125 (see Inde pendent Sample Table Appendix B ), and being greater than the minimum of .05 it is found that there is no significant difference in the variability of the two genders. Additionally the sig (2 Tailed) value was .172, again greater than .05, showing no significant difference in means for the two gender conditions (Tabachnick & Fidell, 2007). Those findings therefore imply that influences on U due to gender cannot be shown to be significant. PLS M odel A nalysis Although we regard this study as exploratory it is felt the usage of the TAM and DOI meet the criteria of SEM's requirement of being theoretically well founded, which it is argued is the underlying reason SEM is considered applicable for confirmatory rather than exploratory research (Vinzi, Chin, Henseler, & Wang, 2010). I n addition the data responded well to the normalizing algorithm applied. This study therefore has made use of covariance based SEM as its primary analysis tool. However there has been discussion of the possible merits of employing the Partial Least Squar es (PLS) method in place of or in addition to the SEM approach. In general PLS is considered applicable when the study is of an exploratory nature, you have formative constructs rather than or along with reflective constructs, the sample size is small (le ss than 10 to 20 times the number of dependent variables), and when data does not exhibit normal distribution (Gefen, Straub, & Boudreau,2000 ; Vinzi, Chin, Henseler, & Wang, 2010 ). This study

PAGE 59

46 is considered to be exploratory, and prior to application of normalizing algorithms is nonnormally distributed, all the constructs are reflective, and the sample exceeds the minimal SEM requirements. In a sense then this study falls somewhere between the two methods. Since this is the case to be thorough a PLS analysis has been carried out. The results of the PLS analysis is reported in two steps. These are the measurement model dealing with reliability and validity of the measurement constructs, and the structural model dealing with the manifest and latent variables relationships. PLS Measurement Model Smart PLS was used as the PLS application software to carry out the analysis (Ringle, Wende, & Will, 2005). The primary model employed parallels closely the SEM model and is show in figure 3. As indicated it is composed of the four MBTI type indicators as directly measured variables, and the research models constructs PC, PU, PEU and with the de pendent construct U. It contains all the same relationship paths as the SEM model.

PAGE 60

47 Figure 3 : PLS Model diagram PLS Reliability The primary test for reliability in PLS is considered to be composite reliability with an acceptable lower limit of .70 (F ornell & Larker, 1981; Vinzi et al, 2010). Composite reliability met or exceeded the .70 minimum for all the constructs. The loading for each factor was examined as well and all were found to exceed its minimum of .50 (Hulland, 1999). As a final conf irmation of reliability Cronbach's alpha was considered, and all constructs were found to exceed the minimum standard level of .7 (Fornell & Larker, 1981; Vinzi et al, 2010). Table 14 displays the composite reliability, factor loading and Cronbach's alpha for the constructs.

PAGE 61

48 Table 14. PLS Reliability Tests values Construct or Factor Composite Reliability Factor Loading Chronbach's Alpha U .897 .855 U1 .697 U2 .902 U3 .714 U4 .870 U5 .855 PC .909 .873 PC1 .858 PC2 .887 PC3 .641 PC4 .848 PC5 .834 PU .883 .823 PU1 .871 PU2 .854 PU3 .824 PU5 .680 EU .869 .803 EU1 .767 EU2 .844 EU3 .839 EU4 .708 PLS Validity Convergent validity is determined by the AVE for a given construct. It is required to be greater than .50 (Vinzi et al, 2010). As seen in Table15 all constructs meet this requirement ranging from the lowest U at .641 to the highest PC at .669. Discriminant validity is shown on Table 15 where in the AVE for any given construct is higher than the squared correlation estimate for the related constructs. The constructs all meet this requirement (Vinzi et al, 2010) Table 15. PLS Squared Correlations Estimate with AVE on Diagonal PEU PU PC U PEU .626 PU .311 .657 PC .424 .594 .669 U .342 .473 .634 .641

PAGE 62

49 PLS Structural Model The Structural Model analyzes the relative strength of the statistical relationships between the variables. These are output as coefficients measuring the relationship as indicated by a path connection in the model. The coefficients for the models inner constructs are shown in Table 16. Table 16. PLS Path Coefficients Path Coefficient (T Value) R 2 PC -> U .601 (7.84*) .654 PU -> U .170 (2.54*) PEU -> U .089 (1.93*) PEU -> PU .558 (13.24*) = > significant at .05 The strongest relationship is PC to U at .601 showing PC's high relevance to usage of online payments. Next highest is PEU to PU at .558 showing PEU's strong influence on PU. Looking at the R2 for U of .654 indicates that 65.4 perc ent of the variance is explained by its predictor variables PU, PC, and PEU. As seen earlier in the discussion of the SEM analysis the MBTI variables are based on a centralized value of zero. Used in this way any coefficient with a negative value indica tes the relationship is effected by the left hand member of the MBTI dimension pairing and a positive value indicates the right hand member of the pairing. These coefficients are displayed in Table 17 along with the corresponding MBTI dimension pair de rived for each of the model constructs. Table 18 displays an overall MBTI type as derived from the analysis.

PAGE 63

50 Table 17. PLS MBTI Coefficients and Derived Pair Indicator MBTI Dimension Coefficients ( MBTI Derived Pair Member) PC PU PEU U EI .057 (E) .011 (E) .060 (E) .048 (E) SN .035 (S) .004 (S) .072 (S) .035 (S) FT .047 (T) .019(F) .062 (T) .037 (T) JP .050 (J) .040 (J) .013 (J) .039 (J) Table 18. PLS Derived MBTI Types Per Constructs PC PU PEU U MBTI Type ESTJ ESFJ ESTJ ESTJ The PLS output related to the MBTI coefficients closely parallels the findings from SEM. The SEM Analysis showed a homogeneous result ESTJ for all of the constructs. The only difference being the results for the PU Construct with PLS indicat ing a derived type of ESFJ. This still does not change the overall type outcome for the dependent U, again ESTJ which coincides with the SEM analysis. The PLS's analysis having found that the FT ->PU relationship is an F value rather than the T value arr ived at by the SEM could be accounted for in several ways. First its coefficient was one of the smallest values arrived at, indicating that on the measurement survey its combined responses would have clustered around the zero mark. The PLS model was run using nontransformed data while the SEM models data was subjected to the smoothing effects of a transformational algorithm. With the original metric hovering near zero the smoothing effect for this one variable may have been just enough to push it over i nto the F dimension. As was suggested this may be noise from the data; we may just be seeing this smoothing effect in action. Also, the MBTI system is dynamic such that the combined type is considered greater than the sum of its' parts. There are functions considered to be dominant and inferior. Due to the dynamic

PAGE 64

51 nature an individual can fluctuate between their dominant and inferior characteristics. In this case we see the SEM yielding an ESTJ type which is Extraverted Thinking with Introverted Sensi ng. The PLS yields an ESFJ type which is Extraverted Feeling with Introverted Sensing. In both cases the Extraverted dimension is the active dimension of the dominant function, with the Sensing being the active function of the inferior function ( Myers, 1 962; Myers & McCauley, 1985; Quenk, 2009). In other words the ESTJ and ESFJ behaviors are mostly influenced by the E and S dimensions. As such it is thought that with these affecting only the PU dimension and the limitations of the differences there is l ittle net difference in the overall model. Discussion Of the three intermediary latent constructs represented in the model, PC was found to have the greatest positive effect on the U construct. PU has the next greatest effect on U followed by PEU with t he least relative effect on U under the constraints of this model. From the negative or positive value of the beta coefficients for the SEM paths between the exogenous MBTI variables and the endogenous construct variables as seen in Table 12 it is determined that tentatively all 12 hypotheses are confirmed to some extent. Out of that twelve, eleven can be grouped into 3 divisions based on the absolute values of their corresponding SEM path beta coefficients. The first five hypotheses had corresponding a bsolute path values above .060. They are H3a & b, H15a & b, H6a & b, H8a & b, and H12a & b which are considered to be strongly confirmed hypotheses. The next three hypotheses all have corresponding absolute path values between .030 and .059. They are H 1a & b, H9a & b, and H11a & b which are considered to be confirmed

PAGE 65

52 hypotheses. The final three all have corresponding absolute path values of less than .030. They are H2a & b, H4a & b, and H7a & b which are considered to be weakly confirmed hypotheses. The parallel values for SEM's path beta coefficients are the PLS model's coefficients seen in Table 17. These values confirm the SEM findings for all paths except PU to FT. This value found that F was the operative MBTI pole for this path compared to S EM's finding of the T pole. This path relates to H10a & b. While PLS shows that the MBTI relationship FT to PU as F, it also shows that the relationship to U as T, as does SEM. Due to these disparities H10a is considered only partially confirmed, as onl y U is positively affected by the T type and so H10b is confirmed. The analysis clearly shows one of the sixteen MBTI personality types emerging as an antecedent determinant to the model. First the study has found that the MBTI type ESTJ has the greatest direct effect on the model constructs PC, and PEU, as well as indirectly on the dependent U of all the MB TI types as seen in Figure 4 ESTJ and ESFJ had the greatest effects over all on PU as indicated by SEM and PLS respectively. In the Myers Briggs psychology literature ESTJ stands for an Extraverted Thinking supported by Sensing type. The Extraverted Thinkers (the dominant function of the ESTJ's), both ES and EN types, tend to be analytical and impersonal. They are logical, decisive and organized. When Extraverted Thinkers are of the Sensing type they are also very practical, down to earth, and matter of fact. These are individuals that make decisions using logical thought processes based on their own five senses. They are described using the following terms; logical; analytical; objectively critical; decisive;

PAGE 66

53 clear; and assertive. Typically they are focused on the present, o n what is real and actual (Myers, 1998). ESTJ's are action and task oriented, preferring reliable procedures and systems (Bayne, 1995; Myers, 1962; Myers, 1998; Myers & McCauley, 1985; Myers & Myers, 1995). This would seem to fit well with the theoretica l underpinnings discussed previously and the qualities inherent in online payment services. Figure 4 : SEM MBTI Dimension Indirect Effect on U

PAGE 67

54 Figure 5 : SEM MBTI Dimension Direct Effect on PC, PU, & PEU

PAGE 68

55 CHAPTER V APPROACH II: DISCOVERING DESIRABLE AND UNWANTED ONLINE CREDIT CARD PAYMENT SERVICE FEATURES FOR DIFFERENT PERSONALITY TYPES The secondary approach conducted for this research is an exploratory descriptive study which looks at features and services preferences among users and nonusers of online credit card payment services related to MBTI Type. In addition to the primary research methodology a second line of research was conducted. This is an initial study to focus in on cogent features and services that look promising for future in dept h research. The primary work looked a Personality Types' effect on usage of online credit card bill payment services. It is thought that if a determinative link is established website designers and service managers might then leverage that relationship t o attract more users and increase retention percentages of those services through targeted design, promotions, advertising methods, and other means. This second approach extends that primary research in several ways. First, it distinguishes features th at promote utilization of e commerce payment services by individuals and also identifies which website features are considered as unimportant by different personality types. Additionally, this second study adds confirmation of personality type's effect on usage. Which e commerce website features and mechanisms promote user perceptions of perceived easeof use and perceived benefit and consequently their intention to continue utilizing these e commerce websites has been previously studied (Yousafzai, Palli ster, &

PAGE 69

56 Foxall, 2005). As an example, Walczak and Gregg (2009; 2010) demonstrate that website content features promote trust and consequently engender perceptions of corporate capability and intention to transact. However these studies are all based on a generic user type representative of average consumer values across a particular region or even across the world, similar to personas. Variance among individuals is mostly ignored. As discussed earlier in the introduction to the primary approach above, the examination of online credit card payment services from an individual perspective is problematic due to the sheer volume of data to be analyzed in order to ensure accuracy of the research model. Therefore to again make use of a classification schema to help preserve individual differences, and complement and build upon the primary approach, this approach also utilizes the MBTI type indicator to provide such a classification technique. Methodology A self administered survey instrument was developed and used to gather data through an online site hosted by Survey Monkey. Data was actively collected starting in January 2013 and terminating in July of 2013. The survey was offered to perspective respondents as part of the University of Colorado Denver Busines s School's "Business Engagement Assignment Program". This is a program run by Professor Stefanie Johnson of the business school to encourage student participation in academic and research activities outside of their enrolled classes. Students enrolled in undergraduate programs are required to participate in one activity over the course of each semester. They select their activity from a list of prospective surveys and other research projects. Upon completion they are given credit for an assignment by their respective instructors per the

PAGE 70

57 class's syllabus. The survey is made up of 6 major sections. Section 1 collects basic demographic information. Section two and three make up the MBTI short form comprised of twenty questions, (not shown) with response s used in calculating the individuals MBTI. Sections 5 and 6 collect payment behavior. Section 5 utilizes 6 questions based on a summated rated seven point Likert scale ranging from Disagree Totally to Agree Totally commonly utilized in the Psychology field (Blumberg, Cooper, & Shindler, 2008; Likert, 1932; Spector, 1992). And Section 6 is made up of 5 open response questions collecting information about user preferences as to online payment site features, and why they pay or do not pay online. The sur vey was tested on a small focus group to make sure that the questions were easily understandable. Potential respondents were invited from both undergraduate and graduate courses in the Business School, with the qualification that they own at least one cre dit card and that they pay their own credit card bills. Being an urban university the respondents were a blend of older and employed students taking courses alongside more traditional students. Most of the respondents received course assignment credit fo r completion of a course requirement to participate in external research activities. No other compensation was offered. No identifiable information was recorded, and respondents were not screened prior to starting the survey. Data The data consists of two main sets of variables. The first set is composed of 6 variables making up the demographic information, 8 variables storing the MBTI polar sums, 4 variables containing the computed MBTI dimensions, and 20 variables containing user preference values These are listed in the variable Key in A ppendix C

PAGE 71

58 The second set of Data contains responses from five open ended questions. Three of the questions are addressed by those who often or always pay their credit card bills online, and 2 are addressed by those respondents that seldom or never pay their credit card bills online. Analysis A total of 107 respondents started the survey, but 27 of these indicated they did not pay their own credit card bills and were disqualified from completing the survey. T he remaining 80 respondents were divided into two categories: regularly pays their credit card online (n=74) and rarely or never paid their credit card online (n=6). The small number of respondents causes some limitations in generalizing the results and f urther in interpreting results for subgroups or personality types with very small response populations (less than 5% of the total population). Future research will continue to collect data and try to determine if the initial descriptive results reported in this article are consistent with a larger population and generalizable to the population of credit card users at large. Until that point it is stressed that caution should be taken in the interpretation of the inferences and or conclusions drawn in from this approach. Demographics for the respondents are listed in Table 19.

PAGE 72

59 Demographics Table 19. Credit Card Website Perception Survey Respondent Demographics Sex 58.75% male 41.25% female Age 6.25% 1820 62.50% 2129 23.75% 30 39 6.25% 40 49 1.25% 50 59 Education 3.75% High School 41.25% Some college or Associate degree 45% Bachelors degree 10% Graduate degree Income in $ 17.5% < 10,000 13.75% 10,000 19,999 17.50% 20,000 29,999 6.25% 30,000 39,999 7.5% 40,000 49,999 6.25% 50,000 59,999 13.75% 60,000 69,999 3.75% 70,000 79,999 2.5% 80,000 89,999 3.75% 90,000 99,999 7.5% The demographics indicate that the respondents, even though students, are more mature than traditional college age students. Slightly more males than females responded. Over half of the respondents had already finished their first Bachelors degree or higher. Interestingly, the income question has a bubble below 30,000 which may indicate full time students who can on ly work part time and thus have more limited income and a second, almost normal curve above the low income bubble, with a median around the 60K income range. The low income bubble represented just slightly less than 50 percent of the respondents, indicati ng that at least half are earning income well above the poverty guideline level for the USA and should therefore have disposable income for making and paying credit card purchases (USDHHS, 2013). Cluster Analysis A Hierarchal Cluster analysis with dendro grams has been carried out for respondents who pay online. This was not conducted for nonpayers due to the small sample size. The Complete Linkage technique is being employed to leverage its strength in creating tight clustering (Tabachnick & Fidel, 2007 ). This will also eliminate problems

PAGE 73

60 that would be seen if Single Linkage were used such as chaining (Hair et al. 2005). The Furthest Neighbor cluster method was used. The cluster analysis was run employing the Squared Euclidean method, clustering on t he preference variables as shown below in the Agglomeration Schedule in Table 2 with its a ssociated dendrogram in Figure 6. Table 20. Cluster Analysis Agglomeration Schedule Agglomeration Schedule Stage Cluster Combined Coefficients Stage Cluster First Appears Next Stage Cluster 1 Cluster 2 Cluster 1 Cluster 2 1 1 2 24.098 0 0 4 2 4 9 46.005 0 0 7 3 5 8 56.649 0 0 5 4 1 6 57.980 1 0 6 5 3 5 88.417 0 3 8 6 1 7 89.622 4 0 7 7 1 4 109.030 6 2 8 8 1 3 169.328 7 5 0

PAGE 74

61 Figure 6 : Combined Linkage Dendrogram Clustering analysis is employed to identify meaningful groups in the data. There is no set heuristic for drawing the line. Drawing the line at this point creates the most detail while maintaining the most homogenous of groups. Analyzing the d endrogram the logical place to draw the cut off line is at approximately 14 on the distance scale. This resolves three distinct and informative groups pointing to the individual respondent's preferences. The first group clusters the Latefees, OnTime, Cr editScore, and

PAGE 75

62 Confirmation variables. These are all issues that relate to insuring payments are made on time and their impact on late fees and maintenance of good credit standing. The second cluster comprised of the Security and Safeway variables both a ddress online security issues. These users are mainly concerned with protection of information assets. The third cluster groups Tutorial tightly with OLHelp, and includes SocialMedia slightly loosely. This group would appear to be primarily interested in ease of use factors, and social perceptions. In addition the homogenous natures of these clustering groups are a confirmation of the accuracy of the survey questions in measuring the intended factors. Results A nd Findings There are two ways to examine the surveys results: looking at individual aspects of each personality type, recalling that they are i n diametric pairs; or looking at each of the 16 complete personality types as defined by MBTI. First it must be noted that as also seen in the SEM analysis the responses did not vary significantly by sex as measured by a KolmogorovSmirnov test, with p = 0.97936, indicating that the two populations of responses have nearly identical distributions. Aspects Of P ersonality T ypes Table 21 shows the demographic percentage of each personality type aspect along with response percentages for each of the measur ed factors. The presence of security features was deemed as very important by all personality type aspects and was seen as the most important feature for all but the N and F personality aspects. This is not unexpected with the prevalence of identity thef t (Identity Theft Assistance Center, 2013).

PAGE 76

63 The second most desired or beneficial feature was uniformly to make payments on time, except for the N and F aspects, where it was the most important. This is surprising, as perceptions of feeling safe conducti ng online payments (the last column) is intuitively correlated with the presence of website security features. This is indicated since the feeling of safety was third for most personality type aspects, except for N where it was tied for most important fea ture. The two remaining features which both had higher than 50 percent recognition by the respondents as important or beneficial features were avoiding late payments and receiving confirmation of a payment being made. Table 21. Important credit card payme nt service features and personality aspects MBTI Pole & Pop.# Percent of total Avoid late fees Make Payments on time Tutorial available Live Chat available Confirm payment Improv e Credit Score Heard it was a good idea on Social Media Security features Perception of Safety E 45 62.50 64.00% 74.00% 20.00% 14.00% 60.00% 48.00 % 6.00 % 82.00 % 74.00% I 29 37.50 70.00% 83.33% 16.67% 13.33% 50.00% 50.00 % 0.00 % 93.33 % 80.00% S 51 66.25 58.49% 73.58% 16.98% 11.32% 54.72% 47.17 % 1.89 % 88.68 % 71.70% N 23 33.75 77.78% 85.19% 22.22% 18.52% 59.26% 51.85 % 3.70 % 81.48 % 85.19% F 18 26.25 71.43% 90.48% 23.81% 19.05% 76.19% 42.86 % 4.76 % 85.71 % 76.19% T 56 73.75 64.41% 72.88% 16.95% 11.86% 49.15% 50.85 % 1.69 % 86.44 % 76.27% J 38 67.50 72.22% 83.33% 20.37% 14.81% 62.96% 55.56 % 0.00 % 90.74 % 87.04% P 36 32.50 53.85% 65.38% 15.38% 11.54% 42.31% 34.62 % 7.69 % 76.92 % 53.85% These results for personality type aspects indicate that it is very important for online credit card payment services to maintain high security standards and promoting a general feeling of safety of information and privacy of information as well as adverti sing these features to consumers. Table 21 reveals that it is also very important to provide a

PAGE 77

64 user friendly confirmation of payments and attract potential users by educating them about the efficiency of online credit card payment services with regard to making on time payments. An indication of the reliability of using MBTI as a determinant of e commerce payment service feature preferences is the question concerning the importance of external influences on adoption of online credit card payment services As may be seen from Table 20, the perceptive (P) and extroverted (E) types were the personality type aspects most affected by the external influence of social media. Perceiving types are commonly seen as flexible and adaptable to the world and like to utilize external input for decision making instead of imposing order (Myers, 1998; MyersBriggs.org, 2013). Extroverts are also prone to seek recognition and approval from peer groups (Myers, 1998; MyersBriggs.org, 2013). Both of these personality types w ould be more likely to utilize social media influences over their MBTI counterparts of introversion and judgment. The other aspect that should perceive external influences from online social networks as important would be the feeling (F) aspect, which pl aces importance on other people affected or involved in a decision as opposed to factual data and this is the case with the F type being the third most influenced by social media (Myers, 1998; MyersBriggs.org, 2013).

PAGE 78

65 MBTI A nd P erceptions Of Online P aymen t Service F eatures However, according to Jung (1921) and consequently the Myers Briggs MBTI individuals personality types are a composition of multiple aspects (Myers & McCauley, 1985). Next we examine the 16 personality types. Figure 7 shows the perc entage of each of the MBTI present in the survey response population. It is interesting to note that the two largest populations are for ESTJ and ISTJ personality types with the only other personality representing greater than 10% of the population being ESTP. This reveals that current users of online credit card payment services tend to have the STJ aspects in their personality and secondarily EST personality aspects. As mentioned in the individual personality type aspect analysis, these two partial t ypes both make sense. The STJ type likes to take the world (or electronic world) as it is, dealing in facts, and finally imposing order through their decisions, thus using an electronic resource enables the STJ to get the facts of their account and determ inistically decide how to affect their account through an online payment and impose this reality immediately (Myers, 1998; MyersBriggs.org, 2013). The EST type is similar, but substitutes interaction with the online world for the imposition of will from the J personality aspect.

PAGE 79

66 Figure 7 Percentage of MBTI personality types in survey population Table 21 reports for each of the 16 types results corresponding to those reported for the individual aspects, but the percentages are with respect to all respondents with the same MBTI (hence, ESTJ percentages represent the percentage of ESTJ respondents that selected a value of 4 or 5 for the corresponding online credit card payment service feature). The gray shading indicates MBTI personality types that did represent at least 5 percent of the overall population and thus pose an analysis limitation due to too few responses of that category. ENFJ, 2, 3% ENFP, 4, 5% ENTJ, 4, 5% ENTP, 3, 4% ESFJ, 2, 3% ESFP, 1, 1% ESTJ, 20, 27% ESTP, 9, 12% INFJ, 3, 4% INFP, 2, 3% INTJ, 4, 5% INTP, 1, 1% ISFJ, 3, 4% ISFP, 1, 1% ISTJ, 12, 16% ISTP, 3, 4% MBTI Sample Population and Percentage ENFJ ENFP ENTJ ENTP ESFJ ESFP ESTJ ESTP INFJ INFP INTJ INTP

PAGE 80

67 Table 22. Important credit card payment service features for MBTI personality types MB TI type Avoid late fees Make payments on time Tutorial availabl e Live Chat available Confirmation of payment Improv e Credit Score Heard it was a good idea on Social Media Security features Perception of Safety ESFJ 2 66.67% 66.67% 33.33% 33.33% 66.67% 33.33% 0% 66.67% 66.67% ESF P1 100% 100% 0 % 0 % 0 % 100% 0% 100% 100% EST J20 66.67% 80.95% 19.05% 9.52% 66.67% 57.14% 0% 90.48% 90.48% EST P9 33.33% 44.44% 11.11% 11.11% 44.44% 33.33% 11.11% 77.78% 33.33% ENF J2 100% 100% 0 % 50.00% 100% 0% 0% 100% 50.00% ENF P4 60.00% 80.00% 60.00% 0 % 80.00% 40.00% 20.00% 80.00% 60.00% ENT J4 100% 83.33% 16.67% 33.33% 66.67% 66.67% 0 % 66.67% 100% ENT P3 33.33% 66.67% 0 % 0 % 0 % 33.33% 0 % 66.67% 66.67% ISFJ 3 33.33% 100% 0 % 0 % 66.67% 33.33% 0 % 100% 100% ISFP 1 100% 100% 0 % 0 % 100% 0 % 0 % 100% 0 % ISTJ 12 66.67% 75.00% 25.00% 16.67% 50.00% 58.33% 0 % 100% 75.00% ISTP 3 66.67% 66.67% 0.00% 0 % 0 % 0 % 0 % 66.67% 33.33% INFJ 3 75.00% 100% 25.00% 0 % 75.00% 75.00% 0 % 75.00% 100% INF P2 100% 100% 0.00% 100% 50.00% 50.00% 0 % 50.00% 100% INTJ 4 75.00% 75.00% 25.00% 0 % 50.00% 50.00% 0 % 100% 100% INT P1 100% 100% 0 % 0 % 0 % 100% 0 % 100% 100% The questions in the survey correspond to several distinct categories: security (Security features and perception of safety), timeliness (make on time payments and

PAGE 81

68 avoid late fees), personal benefit (avoid late fees and improve credit score), ease of use ( tutorial and online chat), confirmation of a completed transaction, and external social influences (recommendation from friends or social media). From Table 22, looking at only the nonshaded rows and thus those personality types that represent at least 5 percent of the population of respondents, it may be seen that the average of the two security influences is the most important (or tied for most important) for all personality types except for ENTJ. This may not be a feature of personality type so much a s it is a necessity in todays online society with hackers and identity theft threats causing people to desire the appearance of secure online transactions. The second most influential feature category and for some personality types tied for first, is timeliness followed closely by confirmation. Thus it appears that the ability to make payments anytime and anywhere in an efficient manner is highly important t o online payment adopters with confirmation of the fact that the payment has occurred, possibly for purposes of nonrepudiation of the payment, the third most significant influence. External influences appears to not have any influence on the desire to ut ilize online credit card payment services by most personality types, with minor influence extending to those personality types that have both E and P aspects, similar to the findings for the individual aspects reported above. Since security and efficie ncy of making payments with a subsequent confirmation that the payment has in fact been made are all important features to online payment

PAGE 82

69 service users, these are features that need to be promoted by credit card payment service providers to continuation of their payment services by existing customers. The open ended questions also supported the importance of the timeliness feature of online credit card payment services. Samples of some of the open ended question responses for various respondents who pay regularly online are: I pay my credit card bills online because it is convenient, fast, and hassle free. Easy and convenient for a busy lifestyle. Its easier and faster than paying by snail mail. I pay online because of the increased access to in formation, instant payment, and to save stamps and paper. Convenience. Traditional mail, phone banking, or paying in person are all unpleasant and time consuming. The fourth item in the responses list shown above also indicates another theme in the open ended question responses that occurred a little less often than timeliness, which is personal benefit or in this case cost savings. The survey has indicated the features perceived as beneficial or necessary by online payment service users. However, a small population of credit card owners that did not pay regularly online also answered the survey and now we look at the features that this group perceived as being useful or needed as well as those features that were not perceived as beneficial and as su ch would not lead to utilization of the online credit card payment service. The MBTI personality type for the respondents that indicated they

PAGE 83

70 did not pay their credit card debt online, were varied and included at least one case of each personality type aspect. Table 22 represents the agreement or disagreement Likert responses for questions similar to what was asked of the online payment service users, but this time with a focus on what features would promote utilization by the current nonusers. If a row does not add up to 100 percent, this represents nonusers who were neutral on a specific category. Table 23 indicates some similarities between users of online credit card payment services and the non users. Feeling safe, making timely payments with co rresponding benefits, and receiving confirmation of the transaction were all important features and external influence from social media was not considered an important or attractive feature. Table 23. Feature perceptions by non users of online credit ca rd payment services Feature Feature not desired Feature desired Security 100% 0% Feel Safe 0% 100% Make payments on time 33% 50% Avoid late fees 17% 83% Improve my credit score 0% 50% Receive confirmation 17% 83% Tutorial 17% 50% Online chat available 50% 17% Hear about it on social media 83% 17% An important difference between nonusers and users of online credit card payment services is the lack of recognition of security as an important feature. All respondents who did not pay regularly online indicated that securing their personal and financial information was not important. Is this due to a lack of awareness of online financial transaction security risks, or perhaps a feeling that security will always be present and as such is not an issue. Additionally, although the nonusers indicated that

PAGE 84

71 security was not important, they did universally indicate that they wanted to feel the transaction was safe, indicating a disparity in their responses. Further research is needed to examine and help determine why nonregular users of online payment ser vices and perhaps e commerce transactions in general do not see a correlation between the security features of a web service and the consequent privacy or safety of their information. Another interesting finding from the responses shown in Table 4, is that half of the nonusers for online credit card payment services indicated that the presence of an online tutorial would increase the likelihood of their utilizing such a service, but a similar percentage indicated that presence of an online chat feature w ould not increase their willingness to utilize such online payment services. One might think that the online payment service nonusers were primarily introverted in nature and hence did not desire to interact with another person, but in fact, less than 17 percent had the I aspect on their MBTI type. A possible explanation is that these nonusers feel their chat questions would be considered nave and that the person(s) on the other side of the chat service would be judgmental and think less of them for not knowing, thus decreasing their self stature. As such, online credit card payment providers should provide easy to understand and yet thorough tutorials of their service and also promote the friendly, helpful, and customer service focused nature of the ir live chat features, to lessen fears preventing potential new users from utilizing these features to become regular users.

PAGE 85

72 CHAPTER VI LIMITATIONS AND IMPLICATIONS One of the primary limitations of this research is its' exploratory nature. There is no current known causational model for personality type's effect on online credit card payment service usage. This study is testing a hypothesized model using the sampled data. This research has a firm theoretical underpinning derived from the utilization of MBTI variables, and TAM, and DOI constructs yet the true causal pathways underlying relationships between personality type and the models constructs are only now being first tested. SEM is best used when casual paths are well known (Kline, 2012). If the use of these well founded variables and constructs had not been employed this study's confirmation of hypotheses could only be viewed as an indication that the hypothesized model cannot be found invalid. While it is felt that the employment of MBTI TAM, and DOI variables and constructs puts this study on somewhat firm ground than that, the lack of confirmed pathways between the exogenous MBTI and the endogenous TAM and DOI variables in the study points to the need for additional work focused to further confirm the model. There is concern as to the malleability of the personality as a possible limitation to any findings related to types. If an individual's personality changes under day to day variable situations, social influences, or other normal pressures then any linkages to preferences based on a measurement of personality type might change as well. In general the psychology literature finds that an individual's personality is primarily developed early on in life with some researchers consider ing it fixed prior to adolescence, while others holding that development continues into early adulthood out to perhaps 30 years of

PAGE 86

73 age (Caspi & Roberts, 2001; Fenichel, 1945; Rothbart, Ahadi, & Evans, 2000; Srivastava, Oliver, Gosling, & Potter, 2003). The current general consensus is that once it is formed, the individual's core personality is primarily stable and while it can change this will normally be a slow process over the course of one's life (Haan, Millsap, & Hartka, 1986; Heatherton & Nichols, 1994; Jones & Meredith, 1996; Mroczek & Spiro, 2003). An individual's behavior may differ in response to situational variables which is normal and expected (Gendlin, 1964, Heatherton & Nichols, 1994). These normal changes in responses are found to be withi n a behavioral range tied to higher level personality traits and not indicative of any fluctuation to the core personality type (Gendlin, 1964; Roberts & Walton, 2006; Rogers, 1957). In cases where abrupt personality change is observed it is invariably re lated to extreme trauma with related stress, and/or symptomatic of a deep rooted mental, emotional, or pathologically based personality disorder (A.P.A, 1980; Rabkin & Struening, 1976; Sanislow & McGlashan, 1998; Vitaliano, DeWolfe, Maiuro, & Katon, 1990). Due to the slow nature of any normal longitudinal changes to an individual's core personality any effect they might have on the outcome of an MBTI test would not be relevant to a study such as this, while rapid changes as noted are abnormal exceptions, and therefore outside the scope of this research. Implications F or Payment Service System Designers The goal of managers of credit card payment service systems is to reduce transaction costs while satisfying customer needs. Increasing usage is of paramou nt importance to maximize savings and gain return on the service site development investment.

PAGE 87

74 As indicated in the SEM analysis section, the MBTI type ESTJ individuals are the most likely to utilize credit card service systems for managing their accounts. This personality type indicates that being action oriented with an outer world focus (E) and a desire for information that is tangible (S) with known logical consequences (T) naturally will accept and use credit card payment service systems (Myers, 1998) Since these systems are under used, then the question becomes how to attract this natural personality type to the payment systems. The extraverted nature implies that social networking would be an ideal way to engage this type of individual, yet findings from the second approach in this study indicate that most users and nonusers alike are not largely influenced by social media recommendations. This warrants additional research to determine if there is some cognitive disconnect at play here. Additiona l research could focus on addressing implementation of design features calculated to attract personality types not currently making use of online payment services. This might take the form of using "Green" strategies in order to make online payment servic e have the perception of greater compatibility and usefulness to Feeling types. The idea of "Going Paperless" can appeal to this type as they are guided by personal values, compassion, and have a need for harmony and positive interactions (Myers, 1998; My ers & Myers, 1995). More introverted users that feel more comfortable writing a check, may want to utilize an online system as long as they do not require any interaction with or requests for help from another person. Augmenting payment sites to have sel f paced tutorials and intelligent help systems that automatically supply information about what is needed or the

PAGE 88

75 next step to be performed, might succeed in making I personality types feel more comfortable using the system on their own. Intuitive personality types want to see the big picture (Myers, 1998). Therefore advertising or messages on the payment website should enable N personality types to get a grasp of how not only the payment system but also the credit card payment process works as a whole a nd consequently how it benefits the user (e.g., through reduction in late fees and improvement in their credit rating The recommendations suggested here target the personality types that underutilize online credit card payment services. The suggestions seek to both improve utilization of personality types that are positive towards online credit card bill payments, but just underutilize this service and to try to supply personality congruent features or advertising strategies for those personality type components that typically would not intend to use a credit card payment site. Future research is needed to verify if these personality targeted recommendations will bear fruit and to identify other personality trait effective recruitment strategies. The s mall response size of Approach 2 limits the generalizability of the results of that section of the study to other areas of e commerce. Future research is needed to examine if personality may explain differential utilization of other types of e commerce si tes or even websites in general. Implications F or Research The fact that perceived compatibility was more significant in the current study than the traditional TAM constructs may reflect growing computer self efficacy. With computers now being used in elementary schools and a generation that has grown up with

PAGE 89

76 cell phones and laptops, it is not uncommon for new college graduates entering the workforce to be comfortable with, fluent in the usage of, and to some extent reliant upon information technology ( El Nasser, 2012). Thus the introduction of compatibility into the more traditional TAM model reflects the necessary adjustment of the model to account for ever growing computer competency in the work force. Therefore, although the TAM model is still thou ght to be a reliable indicator of acceptance of new applications and technology, it is imperative that new models of information technology utilization include some component like perceived compatibility to better model the technology savvy population. The utilization of compatibility in improved TAM like models is similar to prior research which has shown in some professions the compatibility of a new application with the existing work flow is of paramount importance (Richards, 2008; Sibona, Walczak, Bric key, & Parthasarathy, 2011; Walczak, 2003; Walter & Lopez, 2008). The second major finding for research is the indication of the possible impact of personality on usage of web services, specifically credit card payment sites. A wide variety of future res earch can stem from the findings of the presented research. Can this model be further confirmed? How does personality affect other information technology adoption decisions? Also, is it possible to develop a nonintrusive way to automatically assess pe rsonality traits or will users have to take a minipersonality test for websites to be able to adaptively present the best interface for each user? This study has been limited to the use of one personality indicator test. Future research should examine if other alternative personality tests such as The Big Five, or Eysencks

PAGE 90

77 Personality Questionnaire provide similar ratings and whether these profiles might be easier to implement. Alternatively, would it be possible to infer aspects of personality type b ased on the cognitive selection of website customization features, which could in turn promote compatibility through suggestions of other, as yet undiscovered, website features for the user. Individuals with the ESTJ personality traits were identified as being the most likely to utilize online credit card payment services. However this leaves out a large percentage of the population. Previously, recommendations have been presented for how to attract other personality types, but future research is needed to explicitly examine how different personality types and cultures interact with and utilize different types of web based services, especially e commerce sites and applications. Prior research has already found cultural differences in specific types of we b utilization between collectivist cultures versus individualistic cultures and this research should be continued but now emphasizing individual personality traits in addition to cultural effects (BenetMartinez & John, 1998; Kalwar, 2010; Kim, Sohn, & Choi, 2011). It is also thought that additional research should be carried out exploring the ESTJ type from a reverse perspective. Following back from the finding of ESTJ as the dominant type of adopters to their underlying motivations, preferences, and values may yield valuable insights for exploitation by designers site marketing managers, and researchers. Although a small sample size, there was some indication that non online payers were found to be concerned with safety but did not cognitively link that to online security issues. It is felt that additional inquiry into this finding would add increased granularity

PAGE 91

78 into understanding the role of risk acceptance. And as touched on in the discussion to the second approach this has implications for the need of further research in order to determine why nonregular users of online payment services do not see a correlation between the security features of a web service and the consequent privacy or safety of their information and whether this extends even to e commerce transactions in general.

PAGE 92

79 CHAPTER VII CONCLUSIONS Conclusions Approach I This study proposed that personality type has a strong influence on the decision to use online credit card bill payment s ervices and tested 12 subsequent hypotheses. The analysis shows that MBTI's types including the Extraverted and Sensing factors had the highest single direct effect on any of the latent unobserved variables of the MBTI factors. They are present in the combined indicator type s of ESTJ and ESFJ wh ich was found to account for variances in the constructs of PEU, PC, and P U and the independent variable U. The study therefore provides evidence for confirmation of hypotheses H1a through H9b, and H1 1a to H12b. Due to the disparate findings of the SEM a nd the PLS analysis's related to T and F dimension path to the PU construct full confirmation of H10a & b are not confirmed. The Judgment factor was found to have a strong effect on PEU both individually and as a member of the combined ESTJ and ESFJ type s on U, and having a contributory effect on PC. These findings yield additional evidence to the confirmation of hypotheses H1a & b, H2 a & b and H3a & b. The support for the hypotheses confirms that personality traits have a direct effect on PU, and PC and indirectly on usage online credit card bill payment services Additional research is needed to see if these findings may be generalized to ecommerce services in general.

PAGE 93

80 Finally, the presented research has verified compatibility or PC as a reliable predictor of usage of online payment services technology. This finding indicates that future adoption and technology utilization models need to utilize PC as a factor, which should improve their reliability. Future research may investigate if specific domains are more prone to compatibility requirements, such as healthcare systems (Sibona, et al., 2011; Walczak, 2003; Walter and Lopez, 2008). Conclusions Approach II The research presented in Approach 2 is an exploratory descriptive research. It should be understood that this study is based on a small sample size. Any conclusions drawn by the study should be taken with caution due to this limitation. This approach attempts to better understand how individual personality may affect which features of onli ne credit card payment systems are perceived as useful and also those perceived as unnecessary by both users and nonusers of these services. For online payment services, security, followed closely by timeliness or efficiency in making payments were the t wo most valuable features of online payment services. Additionally receiving confirmation of payment was also viewed as a valuable service. For some reason, nonusers of online credit card payment services uniformly disagreed that security was a desirable feature of such websites. However, these non users also indicated they wanted to feel safe in making an online payment should they ever use such a service, which raises issues and the need for future research about the cause of this disparity.

PAGE 94

81 The results in Table 3 may be used by payment service providers and other e commerce transaction providers to determine how to increase utilization of their services. As an example, the ESTJ personality was indicated to be the personality that most freque ntly utilizes online payment services. Noticing differences between this personality types perceptions and other personality type perceptions may enable online payment service providers to attract individuals with other personality types. For example, t he row indicating the perceptions of ENTJ (more intuitive thinkers than the ESTJ type [26]) indicates that this personality type places more emphasis on both avoidance of late fees and also on a general feeling of safety than their ESTJ counterparts. Like wise, ISTJ personality types indicated that with regard to security and timeliness of payments, they were very similar to ESTJ users, but they placed slightly more emphasis on the presence of tutorials and live chat help. The ISTJ increased emphasis on tutorials and help over their ESTJ counterparts indicates that even though they prefer to solve problems on their own, because of this they may require some additional learning or other help to feel confident in their decisions.

PAGE 95

82 R EFFERENCES Abramson, M. A., & Morin, T. C. (2003). E government 2003. Lanham, MD: Rowan and Littlefield. Adams, D. A., Nelson, R. R., & Todd, P. A. (1992). Perceived usefulness, ease of use, and usage of information technology: A replication. MIS Quarterly, 16( 2), 227247. Agarwal, R., & Prasad, J. (1997). The role of innovation characteristics and perceived voluntariness in the acceptance of information technologies. Decision Sciences 28(3), 557582. Agarwal, R., & Prasad, J. (1999). Are individual differences germane to the acceptance of new information technologies? Decision Sciences 30(2), 361391. Ajzen, I. (1985). From Intentions to actions: A theory of planned behavior In J. Kuhl and J. Beckmann (Eds.), Action control: From cognition to behavior. Berlin, Heidelber g, New York: Springer Verlag. Ajzen, I. (1991). The theory of planned behavior. Organizational Behavior Human and Decision Processes, 50(2), 179211. Ajzen, I., & Fishbein, M. (1977). Attitude behavior relations: A theoretical analysis and review of empi rical research. Psychological Bulletin 84(5), 888918. Ajzen, I., & Fishbein, M. (1980). Understanding attitudes and predicting social behavior Englewood Cliffs, NJ: Prentice Hall. Akaike, H. (1973). Information theory and an extension of the maximum likelihood principle. In Pertrov, B. N. & Csaki, F. [Eds.], Proceedings of the 2nd International Symposium on Information Theory 267281. Akinci, S., Aksoy, S., & Atilgan, E. (2004). Adoption of Internet banking among sophisticated consumer segments in an advanced developing country. International Journal of Bank Marketing 22(3), 212 232. Albert, T. C., Goes, P. B., & Gupta, A. (2004). GIST: A model for design and management of cont ent and interactivity of customer concentric web sites. MIS Quarterly 28(2), 161182. Amichai Hamburger, Y. (2002). Internet and personality. Computers in Human Behavior, 18(1), 110. Amichai Hamburger, Y., & Ben Artzi, E. (2003). Loneliness and interne t use. Computers in Human Behavior 19(1), 71.

PAGE 96

83 Amichai Hamburger, Y., Fine, A., & Goldstein, A. (2004). The impact of Internet interactivity and need for closure on consumer preference. Computers in Human Behavior 20(1), 103117. American Psychiatric As sociation (1980). Diagnostic and statistical manual of mental disorders (3rd ed.), Washington D. C.: American Psychological Association Press. Bayne, R. (1995). The Myers Briggs type indicator: A critical review and practical guide London: Chapman and Hall. Belleau, B. D., Summers, T. A., Xu, Y., & Pinel, R. (2007). Theory of reasoned action: Purchase intention of young consumers. Clothing and Textiles Research Journal 25(3), 244257. Benet Martinez, V., & John, O. P. (1998). Los cinco grandes across cultures and ethnic groups: Multitrait multimethod analyses of the Big Five in Spanish and English. Journal of Personality and Social Psychology, 75(3), 729750. Bentler, P. M. (1990). Comparative fit indexes in structural models. Psychological Bulletin, 107(2), 238246. Bernstein, D. A., Clarke Stewart, A., & Roy, E. J. (2008). Psychology (8th ed.), Boston MA: Houghton Mifflin Company. Blumberg, B., Cooper, D. R., & Schindler, P. S. (2008). Business research methods (Vol. 2). New York: McGrawHill Higher Education. Borzekowski, R., E.K. Kiser, "The Choice at the Checkout: Quantifying Demand Across Payment Instruments", International Journal of Industrial Organization, Vol. 26:889902, 2008. Brenner, V. (1997). Psychology of computer use: XLVII. Parameters of internet use, abuse and addiction: The first 90 days of the Internet Usage Survey Psychological Reports 80(3), 879882. Brickey, J., Walczak, S., & Burgess, T. (2010). A comparative analysis of persona clustering methods. AMCIS 2010 Proceedings Paper 215. Browne, M. W., & Cudek, R. (1989). Single Sample cross validation indices for covariance structures. Multivariate Behavioral Research, 24(4), 445455. Browne, M. W., & Cudeck, R. (1993). Alternative ways of assessing model fit In Bollan K. A., and Long J. S. EDs. Testing Structural Equation Models, Newbury Park: CA, Sage. Byrne, B. M. (2010). Structural equation modeling with AMOS. (2nd ed.). Ottawa, OT: Routledge.

PAGE 97

84 Carlson, J. (1985). Recent assessments of the Myers Briggs type indicator. Journal of Personality Assessment, 49(4), 356. Carmines, E. G., & McIver, J. P. (1981). Analyzing models with unobserved variables. In Bohrnstedt, G. W. and Brogatta, E. F. [Eds.] Social Measurement: Current Issues Beverly Hills: CA, Sage. Caspi, A., Roberts, B. W. (2001). Personality development across the life course: The argument for change and continuity. Psychological Inquiry 12(2), 4966. Chau, P. Y. K. (1996). An empirical assessment of a modified technolog y acceptance model. Journal of Management Information Systems 13(2), 185204. Cooper, A. (1999). The Inmates are Running the Asylum: Why high tech products drive us crazy and how to restore the sanity Indianapolis: SAMS, a division of Macmillan Computer Publishing. Davis, F. D. (1989). Perceived usefulness, perceived ease of use, and user acceptance of information technology. MIS Quarterly 13(3), 319340. Davis, F. D., Bagozzi, R. P., & Warshaw, P. R. (1989). User acceptance of computer technology: A c omparison of two theoretical models. Management Science 35(8), 982 1003. Dharwada, P., Greenstein, J., Gramopadhye, A., & Davis, S. (2007). A case study on use of personas in design and development of an audit management system. Proc. Human Factors and E rgonomics Soc. 51st Ann. Meeting 469473. Didio, L. (1998). Beta testers endorse e checks. Computerworld, 32, 57. Digman, J. M. (1990). Personality structure: Emergence of the five factor model, Annual Review of Psychology 41(1), 417440. Dishaw, M. T., & Strong, D. M. (1999). Extending the technology acceptance model with tasktechnology fit constructs Information & Management, 36(1), 921. Edwards, J. R., & Bagozzi, R. P. (2000). On the nature and direction of relationships between constructs and measures. Psychological Methods 5(2), 155174. Emergenetics (2013). Case Studies. [Online] Retrieved from http://www.emergenetics.com/results/case studies. Emergenetics (2013). Clients. [Online] Retrieved from http://www.emergenetics.com/ results/client s.

PAGE 98

85 El Nasser, H. (2012). The office is shrinking as tech creates workplace everywhere. USA Today [Online] Retrieved from http://www.usatoday.com/money/workplace/story/20120605/tech creates workplace everywhere/55405518/1. Eysenck, H. J., & Eysenck, S. B. G. (1975). Manual of the Eysenck personality questionnaire. London: Hodder and Stoughton. Federal Reserve Bank of Boston (2010). The survey of consumer payment choice [Online]. Retrieved from http://www.bostonfed.org/economic/cprc/SCPC/index.htm. Feni chel, O. (1945). Father absence and the personality development of the male child. Fishbein, M., & Ajzen, I. (1975). Belief, attitude, intention, and behavior: An introduction to theory and research. Reading, MA: AddisonWesley. Fisher, J. C., & Pry, R. H. (1971). A simple substitution model of technological change. Technological Forecasting and Social Change, 3(1), 7588. Fliegel, F. C., & Kivlin, J. E. (1966). Attributes of innovations as factors in diffusion. American Journal of Sociology 72(3) 235248. Friedman, B., Kahn, P. H., & Howe, D. C. (2000). Trust online. Communications of the ACM 43(12), 3440. Fornell, C., Larcker, D. F. (1981) Structural equation models with unobservable variables and measurement error: Algebra and statistics. Journal of Marketing Research, 18(3), 382388. Fuchs, R. (2001). Personality traits and their impact on graphical user interface design: Lessons learned from the design of a real estate website. In 2nd Workshop on Attitude, Personality and Emotions in Use r Adapted Interaction. Garson, G. D. (2012). Structural equation modeling (2012 ed.). Asheboro, NC: Statistical Publishing Associates. Garson, G. D. (2001). Reliability Analysis Reliability analysis faculty and staff North Carolina State University [Online] Retrieved from http:// faculty.chass.ncsu.edu/garson/PA765/reliab.htm. Garson, G. D. (2012). Structural equation modeling (2012 ed.). Asheboro, NC: Statistical Publishing Associates. Gefen, D., Karahanna, E., & Staub, D. W. (2003). Trust and TAM in online shopping: An integrated model. MIS Quarterly 27(1), 5190.

PAGE 99

86 Gefen, D., Straub, D. W., & Boudreau, M. C. (2000). Structural equation modeling and regression: Guidelines for research practice. Communications of the Association for Information Systems 4(7), 277. Gendlin, E. T. (1964). A theory of personality change. In P. Worchel & D. Byrne (EDs.), Personality Change (pp. 100148). New York, NY: John Wiley & Sons. Gregg, D. G. & Walczak, S. (2010). The relatio nship between website quality, trust, and price premiums at online auctions. Electronic Commerce Research 10(1), 125. Guadagno, R. E., Okdie, B. M., & Eno, C. A. (2008). Who blogs? Personality predictors of blogging. Computers in Human Behavior 24(5), 19932004. Hair, J. F, Black, B., Babin, B., Anderson, R. E., & Tatham, R. L. (2005). Structural equation modeling: An introduction, multivariate data analysis (Sixth ed. pp. 729793). Boston: Pearson. Hamburger, Y. A., & BenArtzi, E. (2000). The relat ionship between extraversion and neuroticism and the different uses of the internet, Computers in Human Behavior 16(4), 441449. Haan, N., Millsap, R., & Hartka, E. (1986). As time goes by: Change and stability in personality over fifty years. Psychology and Aging, 1(3), 220 232. Hayashi, F., & Klee, E. (2003). Technology Adoption and Consumer Payment Choice http://www.bostonfed.org/economic/cprc/SCPC/index.htm, Accessed on May 10, 2012. Heatherton, T. F., & Nichols, P. A. (1994). Conceptual issues in assessing whether personality can change. In J. L. Weinberger (ED.) Can Personality Change? (pp. 318). Washington D.C: American Psychological Association Press. Hills, P., & Argyle, M. (2003). Uses of the internet and their relationships with individual differences in personality. Computers in Human Behavior 19(1), 5970. Hoffman, D. L., Novak, T. P., & Peralta, M. (1999). Building consumer trust online, Communications of the ACM 42(4), 8085. Hsu, C., & Lu, H. (2004). Why do people play online games? An extended TAM with social influences and flow experience. Information Management, 41(7), 853 868. Hulland, J. (1999). Use of partial least squares (PLS) in strategic management research: A review of four recent studies. Strategic Management Journal 20(2), 195204.

PAGE 100

87 Identity Theft Assistance Center. (2013). More than 12 million identity fraud victims in 2012 according to latest javelin strategy & research report [Online]. Retrieved from http://www.identitytheftassistance.org/pageview.php?cateid=47. Jones, C. J., Meredith, W. (1996). Patterns of personality change across the life span. Psychology and Aging 11(1), 5765. Jung, C. G. (1921, 1971). Collected Works: Psychological types (Vol. 6). Princeton, N.J.: Princeton University Press. Jung, C. G. (1933). Modern man in search of a soul New York: Harvest book. Kalwar, S. K. (2010). Comparison of human anxiety based on different cultural backgrounds. Cyberpsychology, Behavior, and Social Networking 13(4), 443446. Kim, Y., Sohn, D., & Choi, S. M. (2011). Cultural difference in motivations for using social network sites: A comparative study of American and Korean college students. Computers in Human Behavior 27(1) 365372. King, W. R., & He, J. (2006). A Meta analysis of the technology acceptance model. Information and Management, 4(6), 740755. Kirton, M. J. (1976). Adaptors and Innovators: A description and measure. Journal of Applied Psychology 61(5), 622629. Kline, R. B. (2012). Assumptions in structural equation modeling. In R. Hoyle (Ed.), Handbook of structural equation modeling (pp. 111 125), New York: Guildford. Lai, V. S., & Li, H. (2005). Technology acceptance model for internet banking: an invariance analysis. Information and Management 42(2), 373386. Langdon, C. S., & Shaw, M. J. (2000, January). The online retailing challenge: Forward integration and e backend development, In ECIS (p p. 10251028). Lederer, A. L., Maupin, D. J., Sena, M. P., & Zhuang, Y. (2000). The technology acceptance model and the world wide web, Decision Support Systems 29(3), 269 282. Lee, H. H., & Chang, E. (2011). Consumer attitudes toward online mass custom ization: An application of extended technology acceptance model. Journal of Computer Mediated Communication, 16(2), 171200. Lee, M. K., & Turban, E. (2001). A trust model for consumer internet shopping. International Journal of Electronic Commerce 6(1), 7591.

PAGE 101

88 Legris, P., Ingham, P., & Collerette, P. (2003). Why do people use information technology? A critical review of the technology acceptance model. Information and Management 40(3), 191204. Likert, R. (1932). A technique for the measurement of at titudes. Archives of Psychology 22(140), 155. Lu, H. P., & Gustafson, D. H. (1994). An empirical study of perceived usefulness and perceived ease of use on computerized support system use over time. International Journal of Information Management 14(5) 317329. McCrae, R. R., & John, O. P. (1992). An introduction to the five factor model and its applications. Journal of Personality, 60(2) 175 215. McElroy, J. C., Hendrickson, A. R., Townsend, A. M., & DeMarie, S. M. (2007). Dispositional factors in internet use: Personality versus cognitive style. MIS Quarterly 31(4), 809820. Mershon, B., & Gorsuch, R. L. (1988). Number of factors in the personality sphere: Does increase in factors increase predictability of real life criteria?. Journal of Personality and Social Psychology 55(4), 675680. Moon, J. W., & Kim, Y. G. (2001). Extending the TAM for a world wide web context. Information and Management 28(4), 217230. Mroczek, D. K., Spiro III, S. (2003). Modeling intra individual change in personality traits: Finding from the normative aging study. Journal of Gerontology: Psychological Sciences 58B(3), 53 65. Mulder, S. & Yaar, Z. (2007). The user is always right: A practical guide to creating and using personas for the web. San Francisco, CA: New Riders Publishing. Myers, I. B. (1962). The Myers Briggs type indicator manual. Princeton. New Jersey: The Educational Testing Service. Myers, I. B. (1998). Introduction to Type: A guide to understanding your results on the MBTI instrument (6th ed.). Mountain View, CA: CPP, Inc. Myers, I. B., & McCauley, M. H. (1985). Manual: A guide to the development and use of the Myers Briggs type indicator Pala Alto, CA: Consulting Ps ychologists Press. Myers, I. B., & Myers, P. B. (1995). Gifts differing understanding personality type. Mountain View, CA: CPP. MyersBriggs.org (2013). MBTI basics [Online]. Retrieved from http://www.myersbriggs.org/my mbtipersonality type/mbtibasics/

PAGE 102

89 Nath, R., Schrick, P., & Parzinger, M. (2001). Banker's perspectives on Internet banking. E service Journal 1(1), 2136. Nilson Staff. Master Card and Visa US. (2010). The Nilson Report 942, 710. Norman, D. A. (1999). The Invisible Computer: Why good products fail, the personal computer is so complex, and information appliances are the solution (2nd. ed.). Cambridge, MA: The MIT Press. Ozdemir, S., P. Trott, "Exploring the Adoption of a Service Innovation: A Study of Internet Banking Adopters and NonAdopters", Journal of Financial Services Marketing, Vol. 13:284299, 2009. Parthasarathy, M. P., & Bhattacherijee, A. (1998). Understanding post adoption behavior in the context of online services Information Systems Research, 9(4), 364379. Paunonen, S. V., & Ashton, M. S. (2001). Big five factors and facets and the prediction of behavior. Journal of Personality and Social Psychology 81(3), 524539. Pavlou, P. A. (2003). Consumer acceptance of electronic commerce: Integrating trust and risk with the technology acceptance model. International Journal of Electronic Commerce 7(3), 101134. Pruitt, J. & Adlin, T. (2006). The persona lifecycle: Keeping people in mind throughout product design. San Francisco, CA: Morgan Kaufmann. Pruitt, J., & Grudin, J. (2003). Personas: Practice and theory. DUX 03 Proceedings of the 2003 conference on designing for user experiences, ACM New York, NY, USA. Ostlund, R. L. (1974). Perceived innovation attributes as predictors of innovativeness. Journal of Consumer Research 1(2), 2329. Quenk, N. L. (2009). Essentials of MyersBriggs Type Indicator Assessment (2nd ed.). Hoboken, NJ: John Wiley and Sons. Rabkin, J. G., & Struening, E. L. (1976). Life events, stress, and illness. Science, 194, 1013 1020. Richards, D. A. (2008). Call for change in the call center. Journal of Organizational and End User Computing, 20(1), 6475. Ringle, C. M., Wende, S., & Will, S. (2005). SmartPLS 2.0 (M3) beta, Hamburg. [Online] Retrieved from http://www.smartpls.de.

PAGE 103

90 Roberts, B. W., Walton, K. E. (2006). Patterns of meanlevel change in personality traits across the life course: A metaanalysis of longitudinal studies. Psychological B ulletin 132(1), 125. Rogers, C. R. (1957). The necessary and sufficient conditions of therapeutic personality change. Journal of Consulting Psychology 21(2), 95103. Rogers, E. (2003). Diffusion of Innovation ( 5th ed.). New York, NY: Free Press. Rothba rt, M. K., Ahadi, S. A., & Evans, D. E. (2000). Temperament and personality: Origins and outcomes. Journal of Personality and Social Psychology 78(1), 122135. Sanislow, C. A., & McGlashan, T. H. (1998). Treatment outcome of personality disorders Canadian Journal of Psychiatry 43(3), 237250. Schell, B. H. (2007). The Internet and society: a reference book. Santa Barbara, CA: ABC CLIO. Shah, R., & Goldstein, S. M. (2006). Use of structural equation modeling in operations management research: Looking back and forward. Journal of Operations Management. 24(2), 148169. Sibona, C., Walczak, S., Brickey, J., & Parthasarathy, M. (2011). Patient perceptions of electronic medical records: physician satisfaction, portability, security, and quality of care. International Journal of Healthcare Technology Management 12(1), 6284. Srivastava, S., Oliver, J. P., Gosling, S. D., Potter, J. (2003). Development of personality in early an d middle adulthood: Set like plaster or persistent change? Journal of Personality and Social Psychology 84(5), 10411053. Sybil, B. G., Eysenck, G., Eysenck, H. J., & Barrett, P. (1985). A revised version of the psychoticism scale. Personality and Indivi dual Differences 6(1), 2129. Tabachnick, G., & Fidell, L. S. (2007). Structural equation modeling. In J. B. Ullman (Ed.), Using Multivariate Statistics (5th ed., pp. 676780). Boston: Pearson. Tuten, T.L., & Bosnjak, M. (2001). Understanding differences in web usage: The role of need for cognition and the five factor model of personality. Social behavior and personality 29(4), 391. Tornatzky, L. G. & Klein, L. (1982). Innovation characteristics and innovation implementation: A meta analysis of findings. IEEE Transactions On Engineering Management 29(1), 2845. Torres, Z. (2013). Douglas district pilots personality tests with color coded profiles. The Denver Post (August 27), 1A, 9A.

PAGE 104

91 United States Department of Health and Hu man Services. (2013) [Online]. Retrieved from http://aspe.hhs.gov/poverty/13poverty.cfm. US Census Bureau. (2011). Statistical abstract of US, 2011 [Online]. Retrieved from http://www.census.gov/prod/2011pubs/11statab/pop.pdf US China Economic and Security Review Commission. (2010). 2010 Annual Report [Online]. Retrieved from http://www.uscc.gov/annual_report/2010/10_annual_report.php. US Federal Reserve System. (2011). G.1 9 Report of consumer credit [Online]. Retrieved from http://www.federalreserve.gov/releases/g19/Current. Venkatesh, V., & Davis, F. D. (2000). A theoretical extension of the technology acceptance model: Four longitudinal field studies. Management Science, 46(2), 186204. Venkatesh, V., & Morris M. G. (2000). Why don't men ever stop to ask for directions? Gender, social influence, and their role in technology acceptance and usage behavior. MIS Quarterly, 24(1), 115139. Venkatesh, V., Morris, M. G., Davis, G. B., & Davis, F. D. (2003). User acceptance of information technology: Toward a unified view. MIS Quarterly 27(3), 425478. Vijayasarathy, L. R. (2004). Predicting consumer intentions to use online shopping: The case for augmented technology acceptance model. Information and Management, 41(6), 747762. Vinzi, V. E., Chin, W. W., Henseler, J., Wang, H. (2010). Handbook of part ial least squares: Concepts, methods and applications, New York: Springer. Vitaliano, P. P, DeWolfe, D. J., Maiuro, R. D., Russo, J., & Katon, W. (1990). Appraised changeability of a stressor as a modifier of the relationship between coping and depression: A test of the hypothesis of fit. Journal of Personality and Social Psychology 59(3), 582592. Walczak, S. (2003). A multiagent architecture for developing medical information retrieval agents. Journal of Medical Systems 27(5), 479498. Walczak, S. & Gregg, D. G. (2009). Factors influencing corporate online identity : A new paradigm. Journal of Theoretical and Applied Electronic Commerce Research, 4(3), 1729. Walter, Z., & Lopez, M. (2008). User acceptance of information technologies: An examination of physicians. Decision Support Systems 46(1), 206215.

PAGE 105

92 Wang, Y. S., Wang, Y. M., Lin, H., & Tang, T. (2003). Determinants of user acceptance of Internet banking: an empirical study. International Journal of Service Industry Management 14(5), 501 519. Wheeler, P. R., Hunton, J. E., & Bryant, S. M. (2004). Accounting information systems research opportunities using personality type theory and the Myers Briggs Type Indicator, Journal of Information Systems 18(1), 119. Wu, J., & Wang, S. (2005). What drives mobile commerce? An empirical evaluation of the revised technology acceptance model. Information and Management, 42(5), 719729. Yousafzai, S. Y., Pallister, J. G., & Foxall, G. R. (2005). Strategies for building and communicating trust in electronic banking: A field experiment. Psychology & Marketing, 22(2), 181201.

PAGE 106

93 APPENDIX A Survey Instruments Survey I Constructs 7 Point Likert Scale 1 2 3 4 5 6 7 Wholly Agree Mostly Agree Somewhat Agree Neither Agree nor Disagree Somewhat Disagree Mostly Disagree Wholly Disagree Section A Rate your agreement with each statement using the 7 point scale above. Perceived Usefulness 1. I would find online credit card payment services to be useful. 2. I would find online credit card services to be of value 3. I believe online credit card payment Services would help me make my payments on time. 4. Using the online payment services would help to improve my credit score. 5. Paying my credit card online has no real benefit for me. Perceived Ease of Use 1. I find making online payments easy. 2. I think my providers online credit card payment systems is user friendly. 3. My credit card providers website is easy to navigate. 4. I think it is hard to make a payment using my Credit Card Providers website. Perceived Compatibility 1. Us ing an online credit card payment service fits well with the way I pay my bills. 2. Web based credit card payments work with my financial management style. 3. Making credit card payments online conflicts with my lifestyle. 4. The system I use to pay my bills is a g ood fit with online credit card payments. 5. Online credit card services are very compatible with how I manage my bills.

PAGE 107

94 Intent to Use 1. I make some of my credit card payments online. 2. I pay my credit cards bills online. 3. I seldom use online services to pay my credit card bills. 4. I will be making my credit card payments online. 5. I make use of online credit card payment Services. Section B Personality Types MBTI Personality Self Scorable Form M The MBTI Personality Self Scorable Form M is a proprietary copy written instrument, and is available for purchase from CPP, Inc. www.cpp.com Survey II Section 1 demographics 1. Are you male or female? 2. Which category below includes your age? 1. 17 or younger 2. 18 20 3. 21 29 4. 30 39 5. 40 49 6. 50 59 7. 60 or older 3. What is the highest level of school you have completed or the highest degree you have received? 1. Less than high school degree 2. High school degree or equivalent

PAGE 108

95 3. Some college but no degree 4. Associate degree 5. Bac helor degree 6. Graduate degree 4. Which of the following categories best describes your employment status? 1. Employed, working 1 39 hours per week 2. Employed, working 40 or more hours per week 3. Not employed, looking for work 4. Not employed, NOT looking for work 5. Retired 6. Disabled, not able to work 5. Are you now married, widowed, divorced, separated, or never married? 1. Married 2. Widowed 3. Divorced 4. Separated 5. Single never married 6. 6. How much money did YOU personally earn in 2012? This includes money from jobs; net income from business, farm, or rent; pensions; dividends; interest; social security payments; and any other money income received by You. Please report the total amount of money you earned do not subtract the amount you paid in taxes or any deductions listed on your tax return. 1. $0 $9,999 2. $10,000 $19,999 3. $20,000 $29,999 4. $30,000 $39,999 5. $40,000 $49,999 6. $50,000 $59,999 7. $60,000 $69,999 8. $70,000 $79,999 9. $80,000 $89,999 10. $90,000 $99,999

PAGE 109

96 11. $100,000 or More 7. Do you pay your bills yourself? 1. Yes 2. No Logic: If #2.(No) skip to end of survey. Section 2 & 3 contain 20 questions comprising a short form deriving the MBTI type for the respondent. 28. Do you pay your credit card bills online? 1. Never pay online. 2. Seldom pay online 3. Often pay online 4. Always pay online Logic: If 1 or 2 skip to Question 42 If 3 or 4 continue to Question 29 Questions 29 41 are answered only by those answering 3 or 4 to question 28 29. When I pay my credit card bills online I prefer to 1. Pay using my card issuers/providers own pa yment site 2. Pay through a third party such as the online bill payer services of my checking/savings Bank account or Credit Union website. The following questions use a 5 point Likert Scale 1. Disagree Totally 2. Disagree Somewhat 3. Neither Disagree nor Agree 4. Agree Somewhat 5. Agree Totally

PAGE 110

97 30. One reason I pay my credit card bills online is because it helps me avoid late fees. 31. One reason I pay my credit card bills online is it helps me make my payments on time. 32. One reason I pay my credit card bills online is t hat my Facebook, or Twitter, or other friends recommend paying on line. 33. I feel the site I make my payments on is secure enough to protect my identity and information 34. One reason I pay my credit card bills online is that there is a self guided tutor ial on how to correctly make the payment. 35. One reason I pay my credit card bills online is that I feel it will help improve my credit score because I have fewer late payments. 36. One reason I pay my credit card bills online is that I receive an email and/or confirmation code confirming that my credit card bill has been paid. 37. One reason I pay my credit card bills online it that my payment site provides a live chat feature allowing me to ask questions or get directions. 38. One reason I pay my cre dit card bills online is that I feel it is a safe way to make my payments. The next questions are using the respondents own words. 39. On the site where you make your payments what features do you find useful? 40. What features might your online credit c ard payment site offer that would increase your usage of online payments? 41. Why do you pay your credit card bills online? Questions 42are answered only by those answering 3 or 4 to question 28

PAGE 111

98 The following questions use a 5 point Likert Scale 1. Disagree Totally 2. Disagree Somewhat 3. Neither Disagree nor Agree 4. Agree Somewhat 5. Agree Totally 42. I would be more willing to pay my credit card bills online if I knew it would help me avoid late fees. 43. I would be more willing to pay my credit card bil ls online if I knew it would help me make my payments on time. 44. I would be more willing to pay my credit card bills online if my Facebook and Twitter or other friends told me they pay online. 45. I would be more willing to pay my credit card bills online if I knew my identity and information would not be stolen. 46. I would be more willing to pay my credit card bills online if there was a self guided tutorial on how to correctly make the payment. 47. I would be more willing to pay my credit card bills online if I thought it would improve my credit score because I would make fewer late payments. 48. I would be more willing to pay my credit card bills online if I could chat with a person online to help explain the process. 49. I would be more will ing to pay my credit card bills online if I received an email and/or confirmation code confirming that my credit card had been paid. 50. I would be more willing to pay my credit card bills online if the site made me feel safe making my payment.

PAGE 112

99 The next q uestions are using the respondents own words. 51. Why don't you pay your credit card bills online? 52. Explain any features, benefits, or other factors which would make you more likely to pay your credit card bills online.

PAGE 113

100 A PPENDIX B SEM TESTS SEM Regression Weights/ Path Beta Coefficients Relationship Std. Est. Est. C.R. P PEU < --Value_EI .0708 .0144 3.3161 .0009 PEU < --Value_SN .0830 .0183 3.5531 .0004 PEU < --Value_FT .0522 .0117 2.2954 .0031 PEU < --Value_JP .0232 .0048 2.3850 .0171 PC < --Value_EI .0714 .0238 3.3685 .0008 PU < --Value_EI .0271 .0092 2.5320 .0113 PC < --Value_SN .0346 .0127 2.6355 .0084 PU < --Value_SN .0022 .0008 1.9780 .0479 PC < --Value_FT .0783 .0288 3.5075 .0005 PU < --Value_FT .0215 .0080 2.4077 .0164 PC < --Value_JP .0646 .0220 3.2327 .0012 PU < --Value_JP .0348 .0120 2.7046 .0068 U < --PC .7386 .6991 15.9889 .0000 U < --PU .2854 .2663 6.1542 .0000 U < --EU .0986 .1534 2.1574 .0310 PU < --EU .5359 .8934 8.7963 .0000 SEM Test Beta Coefficients for Additional Tested Variables Estimate C.R. P PC < --Gender .032 1.477 .139 PU < --Gender .033 1.876 .061 PEU < --Gender .016 1.275 .202 PC < --Age .014 1.282 .199 PU < --Age .006 .6287 .530 PEU < --Age .008 1.151 .250 PC < --Income .004 1.362 .173 PU < --Income .005 1.524 .125 PEU < --Income .000 .037 .970

PAGE 114

101 T Test for Gender PLS Construct Correlations PEU PU PC U PEU 1.000 PU .558 1.000 PC .651 .771 1.000 U .585 .688 .796 1.000 Group Statistics Gender N Mean Std. Deviation Std. Error Mean U_Value 0 208 6.0308 1.14825 .07962 1 228 5.8772 1.18879 .07873 Independent Samples Test Levene's Test for Equality of Variances t test for Equality of Means F Sig. t df Sig. (2 tailed) Mean Differe nce Std. Error Differenc e 95% Confidence Interval of the Difference Lower Upper U_ Val ue Equal variances assumed 2.367 .125 1.369 434 .172 .15358 .11215 .06684 .37400 Equal variances not assumed 1.372 432. 577 .171 .15358 .11197 .06650 .37365

PAGE 115

102 A PPENDIX C Survey II Variable Key Variable Name Data Set Description Gender All Age All Education All Employment All Marital All Income All Total_E All Sum of all E data points Total_I All Sum of all I data points Total_S All Sum of all S data points Total_N All Sum of all N data points Total_F All Sum of all F data points Total_T All Sum of all T data points Total_J All Sum of all J data points Total_P All Sum of all P data points Value_EI All (2*(Total_I/5) 1) Centralizing Algorithm Value_SN All (2*( Total_N/5) 1) Centralizing Algorithm Value_FT All (2*( Total_T/5) 1) Centralizing Algorithm Value_JP All (2*( Total_P/5) 1) Centralizing Algorithm Payonline Online Makes payments online CCorBank Online Makes payments through Bank or Credit Card website Latefees Online Pays online to avoid late fees OnTime Online Pays online to be on time SocialMedia Online Pays online due to influence from social media sites Security Online Feels payment site is secure Tutorial Online Pays online because there is tutorial available CreditScore Online Pays online to improve credit score Confirmation Online Pays online for real time payment confirmation OLHelp Online Pays online because there is online help available Safeway Online Feel paying online is a safe way to make payment Latefee_ck NotOline Likely to pay online if it helped avoid late fees Ontime_ck NotOnline Likely to pay online if it helped to pay on time Soc ialMedia_ck NotOnline Likely to pay online if social media friends advised to Idsecurity_ck NotOnline Likely to pay online if knew ID were secure Tutorial_ck NotOnline Likely to pay online if knew there were online tutorial CredScore_ck NotOnline Likely to pay online if knew it to improve credit score OLChat_ck NotOnline Likely to pay online if live person chat available explaining process OLCnfirm_Email_ck NotOnline Likely to pay online if knew a payment confirmation email or code is provided Feelsa fepay_ck NotOnline Likely to pay online if site made them feel safe making payment

PAGE 116

103 A PPENDIX D Conceptual Model Construct Detail Detail of PU Detail of PEU

PAGE 117

104 Detail of PC