An Empirical Analysis of Mobile Commerce Continuance Intention [601972]

I

I
An Empirical Analysis of Mobile Commerce Continuance Intention
– A Moderated Mediation Approach

Master’s thesis
MSc in Economics, Business Administration and Marketing
Business and Social Sciences, Aarhus University

Number of characters (w/o spaces ): 182.061
Number of illustrations: 23 (18.400)
Number of pages: 91

Authors:
Kasper Urbrand Nielsen
Study no.: KN92599
Exam no.: 410758

Morten Riise Jensen
Study no.: 20119084
Exam no.: 513707

Supervisor:
Athanasios Krystallis Krontalis , PhD
Department of Business Administration , Aarhus University

Aarhus, 1st of August 2015,

Kasper Urbrand Nielsen Morten Riise Jensen

_____________________________ _____________________________

II
Acknowledgement s

Initially, we want to show our great appreciation with our supervisor, Atha nasios Krystallis
Krontalis , who in this entire process was of great assistance, and was able to push us beyond
our comfort zone. Also, we want to express our gratitude to our respective families for their
constant support and understandings throughout this entire process. Finally, we’d like to
thank all the participants who willingly helped us collect our data through our
questionnaires. Because of the high response rate, we were able to donate 622 DKK to The
Danish Cancer Society (Kræftens Bekæmpelse)1.

1 Appendix 17

III
Abstract

The features of smartphones increase continuously, and op erates, to a higher degree,
as personal a ssistant s that affects nearly every aspect of one’s every -day life . Therefore, in
a world where people feel more stressed , it thus seemed inevitable that smartphones
became a popular option for both business owners and consumers to connect. However,
one third of Danish m -commerce users have cancelled a purchase in an m -commerce
environment due to unsatisfactory experiences, whereas m any don’t return. Therefore, the
purpose of this thesis was to determine how m -vendors could improve their strategies in a
way that could increase the likelihood of maintaining existing m -commerce users. Thus, the
study set out to investigate the mechanism s influencing the relationship between prior
experiences and the intention to continue using m-commerce in the future .

ECM -IS posits that prior experiences are indeed affected by post -usage expectations
toward future usage, why this thesis aims to invest igate this fact. Therefore, drawing on the
ECM -IS, a conceptual model was established with the extensions of cognitive belief
constructs of the TAM, as well as trust and flow, with the intent to identify the underlying
mechanisms influencing this relation by the use of mediation techniques . Furthermore,
acknowledging the difference s in users’ perceptions, the thesis finally analyzed the
conditions of these proposed me diation effects established. Findings were based on valid
responses collected from Danish m -commerce users through quantitative surveys with 187
and 125 responses respectively , using non-probability sampling technique s.

Findings were that prior experiences had a large significant impact on satisfaction,
though this effect was partially mediated by the post -usage expectations of perceived ease
of use, perceived usefulness, trust and flow. Furthermore, these post -usage expectations
were highly influencing users’ continuance intention to use m -commerce. However, the
effects of perceived ease of use and trust were fully mediated by satisfaction , meaning that
m-vendors must be able to fully satisfy their users to yield the effect s of these cognitive
beliefs. Perceived usefulness and flow turned out partially mediated by satisfaction,
meaning that the effects would diminish, but not vanish, if users are not satisfied. In
addition, the mediation effect caused by satisfaction between flow and continuance
intention was moderated by users’ tendency of impulsive behaviour s. Evidently, highly
impulsive users are driven by sudden urges and current stimulus, whereas less impulsive
users use prior experiences as heuristics for future behavio ur, why the necessity of a
satisfactory experience is relatively more important for their intention to reuse the system.

Keywords: M-Commerce, ECT, TAM, Trust, Flow, Impulsiveness, Self -Efficacy , Continuance
Intention, Confirmation, Satisfaction, Moderated Mediation, Mediation .

IV
CONTENTS OF THE THESIS
– CHAPTER I – INTRODUCTION ………………………….. ………………………….. …………… 9
1.1. INTRODUCTION ………………………….. ………………………….. ………………………….. ………………… 10
1.1.2. PROBLEM STATEMENT ………………………….. ………………………….. ………………………….. ……………… 13
1.1.3. DELIMITATIONS ………………………….. ………………………….. ………………………….. ……………………… 15
1.1.4. STRUCTURE ………………………….. ………………………….. ………………………….. ………………………….. . 15
– CHAPTER II – THEORETICAL FRAMEWOR K ………………………….. ……………………. 17
2.1. MAJOR RESEARCH MODELS ………………………….. ………………………….. ………………………….. ….. 18
2.1.1. EXPECTANCY CONFIRMATION THEORY (ECT) ………………………….. ………………………….. ……………….. 19
2.1.2. EXPECTANCY CONFIRMATION MODEL – INFORMATION SYSTEMS (ECM -IS) ………………………….. ………. 20
2.2.3. TECHNOLOGY ACCEPTANCE MODEL (TAM) ………………………….. ………………………….. ………………… 22
2.2. MODEL EXTENSIONS ………………………….. ………………………….. ………………………….. ………….. 24
2.2.1. TRUST ………………………….. ………………………….. ………………………….. ………………………….. …….. 25
2.2.2. FLOW ………………………….. ………………………….. ………………………….. ………………………….. ……… 28
2.3. THE MODERATING ROLE OF PERSONAL TRAITS ………………………….. ………………………….. …………. 32
2.3.1. SELF-EFFICACY ………………………….. ………………………….. ………………………….. ……………………….. 32
2.3.2. IMPULSIVENESS ………………………….. ………………………….. ………………………….. ………………………. 35
– CHAPTER III – CONCEPTUAL MODEL & H YPOTHESES ………………………….. ………. 38
3.1. HYPOTHESIS DEVELOPMENT ………………………….. ………………………….. ………………………….. ….. 39
3.1.1. STUDY ONE ………………………….. ………………………….. ………………………….. ………………………….. . 39
3.1.1.2. Creating Satisfaction through Parallel Mediation ………………………….. ……………………… 39
3.1.1.2. TAM ………………………….. ………………………….. ………………………….. ………………………….. . 40
3.1.1.3. Trust ………………………….. ………………………….. ………………………….. ………………………….. 42
3.1.1.4. Flow ………………………….. ………………………….. ………………………….. ………………………….. . 44
3.1.2. CREATING CONTINUANCE INTENTION THROUGH SINGLE MEDIATION ………………………….. ……………….. 46
3.1.2.1. TAM ………………………….. ………………………….. ………………………….. ………………………….. . 47

V
3.1.2.2. Trust ………………………….. ………………………….. ………………………….. ………………………….. 48
3.1.2.3. Flow ………………………….. ………………………….. ………………………….. ………………………….. . 50
3.2.1. STUDY TWO ………………………….. ………………………….. ………………………….. ………………………….. 52
3.2.1.1. Moderating Role of Self -Efficacy ………………………….. ………………………….. ………………… 52
3.2.1.2. Moderating Role of Impulsiveness ………………………….. ………………………….. ……………… 53
– CHAPTER IV – METHODOLOGY ………………………….. ………………………….. ……… 56
4.1. RESEARCH DESIGN ………………………….. ………………………….. ………………………….. …………….. 57
4.2. INSTRUMENT DEVELOPMENT ………………………….. ………………………….. ………………………….. … 57
4.2.1. STUDY ONE ………………………….. ………………………….. ………………………….. ………………………….. . 57
4.2.2. STUDY TWO ………………………….. ………………………….. ………………………….. ………………………….. 59
4.3. DATA COLLECTION PROCEDURE ………………………….. ………………………….. ………………………….. 60
4.3.1. Pilot Test ………………………….. ………………………….. ………………………….. ……………………….. 60
4.3.2. STUDY ONE ………………………….. ………………………….. ………………………….. ………………………….. . 61
4.3.3. STUDY TWO ………………………….. ………………………….. ………………………….. ………………………….. 62
– CHAPTER V – DATA ANALYSIS & RESU LTS ………………………….. …………………….. 63
5.1. DATA ANALYSIS INSTRUMENTS ………………………….. ………………………….. ………………………….. . 64
5.2. SAMPLE CHARACTERISTICS ………………………….. ………………………….. ………………………….. ……. 68
5.3. REGRESSION ASSUMPTIONS ………………………….. ………………………….. ………………………….. ….. 71
5.3.1. RELIABILITY ………………………….. ………………………….. ………………………….. ………………………….. . 71
5.3.2. VALIDITY ………………………….. ………………………….. ………………………….. ………………………….. ….. 73
5.3.3. NORMAL DISTRIBUTIONS OF RESIDUALS ………………………….. ………………………….. ……………………… 74
5.3.4. HOMOSCEDA STICITY ………………………….. ………………………….. ………………………….. ………………… 75
5.3.5. INDEPENDENCE OF ERRORS ………………………….. ………………………….. ………………………….. ………… 76
5.4. MODEL FIT ………………………….. ………………………….. ………………………….. …………………….. 77
5.4.1. STUDY ONE ………………………….. ………………………….. ………………………….. ………………………….. . 77
5.4.2. STUDY TWO ………………………….. ………………………….. ………………………….. ………………………….. 78
5.5. HYPOTHESIS RESULTS ………………………….. ………………………….. ………………………….. …………. 79
5.5.1. STUDY ONE ………………………….. ………………………….. ………………………….. ………………………….. . 79
5.5.1.1. Creating Satisfaction ………………………….. ………………………….. ………………………….. ……. 80

VI
5.5.1.2. Creating Continuance Intention ………………………….. ………………………….. …………………. 81
5.5.2. STUDY TWO ………………………….. ………………………….. ………………………….. ………………………….. 84
5.5.2.1. Moderating Role of Personal Traits ………………………….. ………………………….. ……………. 85
5.5.3 SUMMARY OF HYPOTHESES TESTING ………………………….. ………………………….. ………………………….. 89
– CHAPTER VI – RECAPITULATION ………………………….. ………………………….. …….. 90
6.1. DISCUSSION ………………………….. ………………………….. ………………………….. ……………………. 91
6.1.1. ASSESSING RQ1 ………………………….. ………………………….. ………………………….. …………………….. 91
6.1.2. ASSESSING RQ2 ………………………….. ………………………….. ………………………….. …………………….. 95
6.1.3. ASSESSING RQ3 ………………………….. ………………………….. ………………………….. …………………….. 98
6.1.4. ASSESSING RQ4 ………………………….. ………………………….. ………………………….. …………………….. 99
6.2. IMPLICATIONS ………………………….. ………………………….. ………………………….. ……………….. 100
6.2.1. MANAGERIAL IMPLICATIONS ………………………….. ………………………….. ………………………….. …….. 100
6.2.2. LITERATURE IMPLICATIONS ………………………….. ………………………….. ………………………….. ………. 103
6.3. CONCLUSIVE REMARKS ………………………….. ………………………….. ………………………….. ……… 104
6.4. LIMITATIONS ………………………….. ………………………….. ………………………….. …………………. 105
6.5. FURTH ER RESEARCH ………………………….. ………………………….. ………………………….. …………. 106
REFERENCES ………………………….. ………………………….. ………………………….. ……………………….. 108
APPENDICES ………………………….. ………………………….. ………………………….. ……………………….. 120

VII
List of Figures
Figure 1 .1 Thesis Structure
Figure 2 .1 Expectancy Confirmation Theory
Figure 2.2 Expectancy Confirmation Model – Information Systems
Figure 2.3 Technology Acceptance Model
Figure 3.1 The Conceptual Model of Study One
Figure 3.2 The Conceptual Model of Study Two
Figure 5.1 The Conceptual Model – Results of Study One
Figure 5.2 The Conceptual Model – Preliminary Results of Study Two
Figure 5.3 Moderating Effect of Impulsiveness
Figure 5.4 Moderating Effect of Self -Efficacy

List of Tables
Table 2.1 Constructs and their Origin
Table 3.1 Hypotheses
Table 4.1 Origins of Measurement Items
Table 5.1 Demography Analysis
Table 5.2 Share of M -Commerce Activities
Table 5.3 Reliability and Validity Analysis
Table 5.4 Model Fit
Table 5.5 Mediation Analysis of Study One – Part I
Table 5.6 Mediation Analysis of Study One – Part II
Table 5.7 Mediation Effects Assessment – Study One
Table 5.8 Multiple Regression Results – Study Two
Table 5.9 Conditional Indirect Effect at Different Levels of Impulsi veness
Table 5.10 Summary of Hypothesis Testing

VIII
List of Appendices
Appendix 1 Questionnaire Items
Appendix 2 Independent t-test for Online and Offline Respondents
Appendix 3 Independent t-test for Early and Late Respondents
Appendix 4 Demography Analysis – Study One
Appendix 5 Demography Analysis – Study Two
Appendix 6 Scale Reliabilities – Study One
Appendix 7 Scale Reliabilities – Study Two
Appendix 8 Normality of Residuals for Satisfaction – Study One
Appendix 9 Normality of Residuals for Continuance Intention – Study One
Appendix 10 Normality of Residuals for Continuance Intention – Study Two
Appendix 11 Homoscedasticity Analysis
Appendix 12 Homoscedasticity -Consistent Regression Results
Appendix 13 Independence of Errors – Study One
Appendix 14 Independence of Errors – Study Two
Appendix 15 Questionnaire – Study One
Appendix 16 Questionnaire – Study Two
Appendix 17 Donation for The Danish Cancer Society (Kræftens Bekæmpelse)

Chapter I – Introduction

Page 9 of 155

– Chapter I –
INTRODUCTION

”There is nothing more difficult for a truly creative painter than to paint a rose, because
before he can do so he has first to forget all the roses that were ever painted.”
– Henri Matisse

Chapter I – Introduction

Page 10 of 155
1.1. Introduction

In the recent years there has been an evident aggressive growth in mobile device users
and 3G/4G mobi le internet subscriptions sold, which is consequently reflected in an
exponential growing market share of mobile commerce (m-commerce ). Indeed, the
concept ha s increased in popularity to a degree, where experts predict a bright future for
the concept. A report from Digi-capital (2014) estimate a nearly 300% increase to a market
share on 516 billion dollars in 2017. This trend is also apparent in Denmark. From 2012 to
2014 the share of Danish households having a smartphone has risen from 50% to 73% (Dst,
2015) , along with an increase in mobile wireless internet subscriptions by 24.5% from 3.2
million subscriptions in 2013 to 4 million in 2014, while also the average amount of internet
data used per subscription has increased from 3.4 GB in 2012 to 10 GB in 2014
(Erhvervsstyrelsen, 2014) . Realizing the potential in Danish m -commerce, the Danis h
telecommunication industry followed up this diffusion by investing intensively in improving
the telecommunication network with an average investment rate on 19.2% between 2008
and 2012, compared to an overall investment rate on 12.8% for Europe (Erhvervsstyrelsen,
2013) . In addition, in the effort of improving the Danish digital infrastructure, the Danish
government has supported the telecommunication industry with several initiatives (Emv,
2015) , resulting in superior network coverage and price levels compared to international
standards.
As the share of mobile internet users has increased, so has the share of users who have
conducted a purchase using a mobile internet connection. Indeed, from 2012 to 2014, the
share of Danish consumers who have purchased goods or services through a mobile device
(tabl et or smartphone) increased from 19% to 33% (DIBS, 2015) . And though the
penetration rate of m -commerce has yet to reach the same level as those of Asia or the
U.S., the business opportunities and value s of Danish m -commerce are still projected to
experience a significant increase in the coming years, due to the favorable conditions in the
Danish digital infrastructure , and the increasing saliency in socio -demographic factors
among generation Y (DIBS, 2015) . The shopping aspect of m -commerce can be described as
“any monetary transactions related to purchases of goods or services through internet

Chapter I – Introduction

Page 11 of 155
enabled mobile phones or over the mobile wireless telecommunication network” (Wong et
al., 2012, p. 25) , and though it has many similarities to conventional online shopping , it
delivers unique value s through measures not possible for other shopping methods.
One of these measures is that m -commerce breaks geographical boundaries , and
empowers users with ubiquity and immediacy, allowing them to search for information and
purchase products or services fr om anywhere at any time (Tiwari and Buse, 2007) . Online
stores therefore collide with offline stores in real time, as consumers are equipped with the
unique opportunity to compare products and prices from multiple sources directly , while
visiting physical st ores (Mahatanankoon et al., 2005) , which further allows users to make
more informed dec isions . Moreover, as opposed to regular computers, mobile devices are
designed to be “ always on” and in constant connection with the internet, which allows for
convenient and rapid access to online stores. Additionally, the built -in GPS feature enables
mob ile-vendors ( m-vendors ) to distribute special product offerings based on various
information not accessible by other shopping channels. This be the physical location of the
user (Tiwa ri and Buse, 2007) , which grants an opportunity for m -vendors to customize
marketing and offers based on users’ online check ins on social medias , announced
participation in events etc. (Mahatanankoon et al., 2005) . However, though stopping
through mobile devices seems promising, it is not without downsides.
M-vendors who are currently de livering m -commerce services are generally suffering
from low profits, shallow user bases and severe problems with high discontinue rates (Hung
et al., 2012; Lu, 2014) , since m -shopp ers are volatile and may not return, once they leave
(Chong, 2013) . This is problematic, due to the technological disparities between mobile
devices and computers that force businesses to invest significant resources to develop
software to comply with a mobile platform (Chong, 2013) . This adjustment, in turn, seems
important to retain users. In fact, every third Danish mobile shopper have cancelled a
purchase process initiated through their mobile device within the past six months , due to
an unsatisfying experience with the shopping channel (DIBS, 2015) . These unsatisfactory
experiences are often attributable to the fact that mobile devices have small screens,
inconven ient input, low multimedia processing power and poor connectivity (Lee, 2014 ). In

Chapter I – Introduction

Page 12 of 155
addition, m -shoppers operate in an online environment, which prevents them from
assessing reliable indications of actual product quality, while the lack of fac e-to-face contact
and wireless electronic nature of the operation makes the purchase subjective to concerns
about money , and personal information being d istributed to third parties without their
consent s. Thus, both greater mistrust and risk is likely to present itself within the context of
m-commerce (San ‐Martin and López ‐Catalán, 2013) . Moreover , previous studies estimate
that the costs of attracting new users are five times the costs of retaining existing ones
(Schefter and Reichheld, 2000) . It therefore seems vital that m -vendors manage to reduce
these negative measures, thereby reducing customer churn (Chong, 2013; Luqman et al.,
2014) . Also , as system users are independent individuals that are likely to have d ifferent
perceptions and orientations, these may induce considerable implications for their
respective behaviour s. In fact, previous research have clearly demonstrated that users’
decisions to accept a system is not bas ed on the same set of criteria (Cheng, 2014; Hsu et
al., 2012) . For instance, t he process of shopping via a mobile device , and maneuvering
mobile applications , often requires a certain level of user skill s and technological
comprehension . This may, in the eyes of some users, be a somewhat challenging task, why
these may find themselves constrained by their beliefs in their own abilities. At the same
time, m -vendors have tradit ionally continuously focused marketing activities , with the
intent to persuade users to conduct unplanned purchases (San ‐Martin and López ‐Catalán,
2013) . However, certain users require a more comprehensive assessment of the market,
why failure to evaluate product alternatives, price differences etc., may easily lead to an
unsatisfactory experience (Rook and Fisher, 1995) . Therefore, the inten tions to continue
using m -commerce services might likely be dependent upon personal predispositions. This
thus speaks to the fact that m -vendors may have to customize their marketing strategies to
better fulfill individual users’ needs.
Past research have mainly been focusing on investigating salient factors that facilitate
users ’ initial inte ntion to adopt m -commerce , leaving research on continuance intention
much more limited (Luqman et al., 2014) . This is evident , despite researchers for long have
called for attention to this matter (Choi et al., 2008) . The primary issue in the exiting
research has been to assess , whether the determinants recognized in adoption studies

Chapter I – Introduction

Page 13 of 155
retain their sign ificance in also explaining post -adoption behaviour s. Researchers have , in
this relation , turned for assistance in the Expectancy Confirmation Theory (ECT), and found
evidence indicating that by providing an experience that lives up to users ’ expectations will
increase their satisfaction , which in turn increase s their intention to return (Chong, 201 3;
Hung et al., 2012; Lee, 2014 ). However, as research on post -adoption behaviour s are still in
the introduction phase (Groß, 2015) , insufficient information is available to make accurate
inferences about the specific nature of expectations that m -vendors need to accommodate
in order to satisfy their users . Moreover, since current research have been focusing on
identifying antecedents of continuance intention , little effort has been put into investigating
how the impact of antecedents is determined by providing a n overall satisfactory
experience. Furthermore, in a shopping context, some researchers have investigated the
moderating role of culture (Zhang et al., 2012) , innovativeness (Yang, 2012) , and
psychographics (Molina -Castillo et al., 2008) in consumers intention to adopt m -commerce .
However , no studies have yet examined the intervening effect of consumers ’ adherence to
buy impulsive ly, as well as the level of mobile self -efficacy among determinants dr iving
users ’ continuance intention.
A severe limitation also worth mentioning in current literature is that research amongst
European consumer s are very limited . The majority of studies published within the area is
conducted in either East Asia or the U.S, why generalization of results is often constrained
by cultural barriers (Groß, 2015; Luqman et al., 2014) . In addition, m -commerc e is an area
in constant motion. Mobile technologies , as well as telecommunication networks are
evolving rapidly , and as consumers gain more and more experience, new per ceptions and
needs may quickly emerge (Papp as et al., 2014) . Further research in this field is therefore
needed.
1.1.2. Problem Statement

As evident from the preceding introduction, the improvement in telecommunications
networks , as well as in mobile technologies , have ramped up the sales of 3G/4G enabled
mobile devices and users , are becoming more willing to accept these devices as commercial
tools. Now, m -vendors need to ensure market growth by understanding how they can retain

Chapter I – Introduction

Page 14 of 155
their users. However, current research on m-commerce post -adoption behaviour s is still in
its infancy (Groß, 2015; Luqman et al., 2014) and are lacking a better understanding of the
mechanisms that cau se users to continue using m -commerce systems. Gaining insight into
processes that stimulate the individual user’s intention to return will inevitable empower
m-vendors to recognize and deliver the needed downstream activities , and increase the
possibility of receiving a satisfying return on investment. The main purpose of this study will
therefore b e to understand:
What are the underlying mechanisms causing users to continue using m -commerce ,
and are these dependent upon users’ personal traits?
In order to investigate causalities that influence the individual users’ post -adoption
behaviour , this study draws on ECT (Oliver, 1980) and empirically tests three research
models that focus on post -adoption beliefs (Bhattacherjee, 2001) . The purpose is to
understand how a fulfillment of users ’ expectations can increase their satisfaction with m-
commerce , and to understand the mediating role of sat isfaction in driving users ’
continuance intention. Furthermore, the study seeks to reveal if the mediating role of
satisfaction is consistent for users with different personal traits. As such, the research
models are developed to test the following researc h questions:
RQ1: By what exten t do perceived ease of use, perceived usefulness, trust and flow explain
the relationship between user expectancy confirmation and user satisfaction ?
RQ2: By what extent does user satisfaction explain the effect s of perceived ease of use,
perceived usefulness, trust and flow on continuance intention?
RQ3: Assuming a mediational effect between flow and continuance intention through user
satisfaction, by what exten t is this effect dependent on users’ tendency to buy imp ulsive ly?
RQ4: Assuming a mediational effect between flow and continuance intention thro ugh user
satisfaction, by what extent is this effect dependent on users’ degree of self-efficacy ?

Chapter I – Introduction

Page 15 of 155
1.1.3. Delimitation s

In the area of m -commerce there are various definitions , including different devices, as
well as several unrelated activities that both include monetary and non -monetary activities .
Most users can, however, be divided into two categories: mobile users , with the focus on
communicat ion purposes (i.e. using the device for gaming , texting, calling etc. ) and mobile
shoppers , with a distinct focus on purchasing (buying products or services ) (Hung et al.,
2012) . This study focuses on the latter , and investigates the aspects of m-commerce that
includes the use of mobile phone applications or mobile phone browsers (e.g. Safari,
Firefox, Chrome) to purchase products or services (tickets, bets, travels, physical products,
software, subscriptions) fr om electronic retail stores. This therefore excludes activities , such
as browsing for information or scanning QR codes. By the same token, smartphone users
are through proximity payment technology (RFID, NFC) essentially offered the opportunity
to use thei r smartphone s as mobile wallet s to conduct on the spot payments , by swiping
their smartphone s over a terminal (Zhou, 2013a) . The same option that is available via
applications such as “Mobil epay” or “Swipe”. This naturally spawns a different usage
context , hence the aspect of using the smartphone as a payment method will not be
addressed. Also, it does not deal with the cost s of having access to online services , i.e.
internet s ubscription fees. Furthermore, m -commerce literature have failed to have an
explicit focus on smartphones and tablets , despite the fact that statis tics show that the
majority of m-commerce activities are conducted through these devices (DIBS, 2015;
eMarketer, 2013) . Thus, little is essentially known about the influence of these devices on
m-commerce behaviour s (Groß, 2015) . In an effort to shed some light on the area , this study
will only focus on smartphones. Table ts are , due to their sizes , not considered truly mobile ,
and are therefore excluded.
1.1.4. Structure

This study is split into six chapters (figure 1.1 ). The first chapter provides an introduction
along with the problem statement. Chapter two discusses the theoretical foundation as well
as prior research findings. Chapter three covers hypothesi s development , and presents the
conceptual models. Chapter four discusses instrument development and data collection

Chapter I – Introduction

Page 16 of 155
procedures. Chapter five descri bes the statistical techniques and methods used to test the
conceptual models , followed by analyses of the data and presentation of results. Chapter
six covers a discussion of results, implications, limitations of the study and suggestions for
future research.

FIGURE 1.1. THESIS STRUCTURE
SOURCE : OWN MAKING

Chapter II – Theoretical Framework
Page 17 of 155

– Chapter II –
THEORETICAL FRAMEWORK

“You can’t sell anything, if you can’t tell anything.” – Beth Comstock

Chapter II – Theoretical Framework
Page 18 of 155
This chapter comprises three parts. First, an examination of relevant theory and
frameworks within m -commerce , followed by model extensions that could possibly
enhance the exp lanatory value for these models and finally, the last part of the chapter
reviews findings within personal traits in the context of moderation.

2.1. Major Research Models

Several different research models have been used when studying consumer behaviour .
And although there are a difference s as to which products consumers ’ purchase and from
the channels, from where they purchase, there is generally a uniform approach. For
instance Fishbein and Ajzen's (1975 ) Theory of Reasoned Action ( TRA) and Ajzen's (1985)
further evolved Theory of Planned Behaviour (TPB) that has been applied to investigate
consumer behaviour among consumers shopping in both stationary shops, (Irianto, 2015) ,
consumers shopping through electronic commerce ( e-commerce ) (George, 2004; Lim and
Dubinsky, 2005) and m-commerce (Kim, 2010; Lin and Wang, 2006) . Still, researchers have
persistently tried to develop extensions and entire models trying to accom modate the
aspects of more specific consumer situations ; as for instance the Unified Theory of
Acceptance and Use of Technology (UTAUT) , Decomposed Theory of Planned Behaviour
(DTPB) or the Technology Acceptance Model (TAM) that, based on the characteristics of
TRA and TPB, has been customized to accommodate the alleged differences between
consumers considering buying products in general , and consumers considering ado pting a
certain technolog y (Chuttur, 2009; Davis, 1989) . Given its results in many studies, TAM has
thus heavily been applied to investigate consumers’ intention to adopt m-commerce (Lee,
2014 ; Thong et al., 2006; Zhou, 2014a) . In the context of maintain ing customers, the range
of theories is more limited. However a commonly applied theory is the ECT (E.g. Chong,
2013; Kim, 2010; Lin and Wang, 2006) .

Chapter II – Theoretical Framework
Page 19 of 155
2.1.1. Expectancy Confirmation Theory (ECT )

Measuring the intention to purchase a product , or to use a system , is a very important
aspect in conquering a market. However, an equally important aspect, if not more so, is the
aspect of measuring a consumer’s willingness to repeat that same behaviour (Thong et al.,
2006) . Note that consumers are, everything else being equal, more valuable for companies
if they re -purchase rather than abandoning the company after first b uy (Anderson and
Sullivan, 1993) . Hence, the more frequently consumers buy, the more profit suppliers will
yield from these. Given the importance, many researchers have investigated the subject in
depth. Oliver (1980) , however, was the first to pioneer with an esteemed framework that is
still hi ghly applied in contemporary research (Bhattacherjee, 2001; Thong et al., 2006) . A
prerequisite for this model is that the consumer has already purchased the good or service
before (Oliver, 1980) . The framework then proposes that the consumer’s intention to
repurchase is a comprehensive composition of (1) the perceived performance of the
product or service, (2) the (/dis)confirmation of the initial expectations prior to the purchase
and (3) the level of satisfaction (Bhattacherjee, 2001; Oliver, 1980) .

Initially, a consumer develops expectations toward a certain good. If these expectations
are sufficiently high, they will eventually likely lead t o a purchase (figure 2.1) . Subsequently,
consumer s will then evaluate their perception of the good’s performance , vis-à-vis their
initial expectations, after whi ch this difference will lead to a (/dis) confirmation of their initial
expectations. If the initial expectations are confirme d, the probability of repurchasing
intention are said to increase significantly , due to a higher degree of satisfaction . (Oliver,
1980) .

Chapter II – Theoretical Framework
Page 20 of 155

FIGUR E 2.1. EXPECTANCY CONFIRMATION THEORY
SOURCE : OWN MAKING , BASED ON OLIVER (1980)

However, a heavily debated paradox in the ECT , is that the model is not constructed as
perpetual as illustrated. The model thus ignores the antecedents of enhanced experience
with the product and other cognitive processes, which might lead to possible changes in
future expectations (Bhattacherjee, 2001) . Consequently, the initial expectations toward a
certain product will , according to the EC T, remain the same regardless of how many times
the consumer will repurchase. This might induce a problem, as the model might generate
deceptive results if not adjusted for different changes in the market, new innovations,
consumer perception s etc. (Bhattach erjee, 2001; Lee, 2014 ).

2.1.2. Expectancy Confirmation Model – Information Systems (ECM -IS)

To offset some of the aforementioned shortcomings in the ECT, Bhattacherjee (2001)
realized the need to expand the model with a link of cognition to better explain the model
within an IS context. In his modified framework (figure 2.2) , he added the cogniti on of post –
consumption expectation, represented by perceived u sefulness, trying to accommodate for
users’ needs to reevaluate expectations toward the IS, since he claims that post
expectations are essential in the case of system usage , where expectations c an change over
time. This setting further complies with the definition of expectation in the ECT, holding that
expectations equal the sum of a user’s beliefs, since perceived u sefulness is a cognitive
belief salient to IS us age (Bhattacherjee, 2001). Howev er, an important change to notice is
that while the ECT investigates both pre – and post -consumptions variables, the ECM -IS only
focuses on post -consumption variables , since the effects of the relative match between pre –
consumption expectations and perceive d performance is already accounted for in the

Chapter II – Theoretical Framework
Page 21 of 155
confirmation and satisfaction construct. This thus alters the ECT paradigm, since ECM -IS
posits that users using IS will keep updating their expectations as they get more familiar
with the system (Hong et al., 2006). In essence, ECM -IS drives on the assumption that
perceived performance is fully mediated through confirmation, and that post -consumption
expectations are modified through direct experience with the IS, which in turn functions as
vital predictors of users’ satisfaction formation and continuance usage intention
(Bhattacherjee, 2001) .

This rationale has predominantly been accepted in contemporary research, and has thus
been adopted in later studies and further expanded with other cognitive beliefs. So, though
the findings within m -commerce are rather variegated, due to the various definitions of the
term, it seems there are congruent findings from the model to explain continuance
intention. E.g. Kim (2010) , who proposed a variation of the model to explain users’
continuance i ntention to use mobile data services , Akter et al. (2013) finding the relation
within mobile applications . Or Hong et al. (2006) , whose study found that the very same
model significantly explained the users’ continuance i ntention within mobile internet in
general . These studies represent different aspects of m -commerce, and thereby indicate
congruity. Thus, it seems that the model is applicable for s tudies aiming to understand the
continuance i ntention within m-commerce in general regardless of its nature, since the
model holds that positive confirmation, corresponding cognitions and hence also
satisfaction, will impact the continuance intention (Bhat tacherjee, 2001).

FIGURE 2.2. EXPECTANCY CONFIRMATION MODEL – INFORMATION SYSTEMS
SOURCE : OWN MAKING , BASED ON BHATTACHERJEE (2001)

Chapter II – Theoretical Framework
Page 22 of 155
2.2.3. Technology Acceptance Model (TAM )

In the TRA framework from 1975, Fishbein and Ajzen proposed that the actual behaviour
of a given person could be explained by examining the person’s intention and the reasons
for it. Thus, the stronger the intention to a certain behaviour , the more likely the person
will actually behave as intended (Chuttur, 2009) . As an extension to the TRA, Davis (1989)
developed the TAM (figure 2.3) to better understand us ers’ behaviour al intention to adopt
an IS (Davis et al., 1989) . The model is thus, as opposed to the TRA and TPB, a much more
customized framework (Bhattacherjee, 2001; Thong et al., 2006) , why the model has shown
highly significant res ults throughout the years, and therefore is widely credited as well as
highly applied in contemporary studies (Chuttur, 2009) .

In the model, Davis (1989) suggested that there are two variables that best enhance the
understandings of a user’s attitudes towards the intention to adopt an IT. (1) Perceived
usefulness and (2) perceived ease of us e. Perceived u sefulness has been widely debated,
and was first introduced in a factor analysis by Schultz and Slevin (1975), stating that a
system that does not enhance a user’s performance in his/her job delivery, is not likely to
be w ell received by the user (Schultz and Slevin, 1975, cited in Longe et al., 2010) . As a result
of this analysis, Davis defined perceived u sefulness as “the degree to which a person believes
that using a particular system would enh ance his or her job performance” (Davis, 1989 , p.
320), meanin g that the system must deliver enhanced usability. Therefore, perceived
usefulness has heavily been associated with the positive influe nce on users’ overall
satisfaction with a system, given that perceived usefulness represents a behaviour al reward
(Davis, 1989) , and thereby an extrinsic motivation related to the confirmation of initial
expectations (Hung et al., 2012; Kim, 2010) .

Davis refers to the definition of “ease” , as the freedom from difficulty or great effort and
thus perceived ease of use is defined as “the degree to which a p erson believes that using a
particular system would be free of effort” (Davis, 1989 , p. 320 ). Thus, the theory builds on
the premise that achieving a desired result easier or faster, possibly enhances the

Chapter II – Theoretical Framework
Page 23 of 155
satisfaction with a system, and in turn, also the likelihood of the customer switching from
one system to another or to adopt a system in g eneral (Bhattacherjee, 2001; Hong et al.,
2006; Thong et al., 2006) .

FIGURE 2.3. TECHNOLOGY ACCEPTANCE MODEL
SOURCE : OWN MAKING , BASED ON DAVIS (1989)

However, though the model is considered valid , and has been applied extensively in
contemporary research, researchers have equally criticized the model for its shortcomings.
E.g., Legris et al. (2003) noted that a severe problem with the TAM is that analyses are
conducted with self -reported use data, meaning that the model is subject to bias as
opposed to objective actual -use data. This argument came in the aftermath of a study in La
Presse Montréal in 2000, where researchers had observed that only 67% of use rs of a public
restroom in New Orleans actually washed their hands after using the toilet. However, in
comparison, the same researchers conducted a survey among 1201 Americans, where 95%
answered that they always wash their hands after using the toilet (La Presse Montréal,
2000 ; cited in Legris et al., 2003 ). Also, researchers have pointed out the relatively low
explained variance of the model in a gen eral context , indicating that there are other
variables of significant influence, why the model should be extended with other variables
(Legris et al., 2003; Yang and Yoo, 2004) .

Chapter II – Theoretical Framework
Page 24 of 155
2.2. Model Extens ions

Hong et al. (2006) extended the applicability of ECM -IS, when studying consumers
continuance intention to use mobile internet. Their initial purpose was, however, to
investigate the explanatory power of ECM -IS by running three separate models; TAM, ECM –
IS and an extended ve rsion of ECM -IS that also includes the construct of perceived ease of
use. What they found were that the extended version of ECM -IS were almost as robust as
TAM, while having a higher overall explanatory power compared to the original versions of
ECM -IS an d TAM. This study therefore choose to adopt the ECM -IS with both construct from
TAM integrated in the framework. Moreover, it is likely that the mechanisms causing users
to continue shopping via their smartphones , are not solely driven by their perceptions of
the interaction being free of effort and useful. Thus, in order to gain a more comprehensive
understanding of the process that stimulate repetitive behaviour s, some additional
extensions are needed.

Previous studies have identified several determinan ts of users continuance intention to
use m -commerce services: Mobile affinity, mobile device experience, demographic s,
frequency of mobile use (Bigné et al., 2007) , trust , habit (Lin and Wang, 2006) , subjective
norms (Kim, 2010) , perceived cost, perceived enjoyment (Chong, 2013) and flow (Zhou,
2013a) . However, including all the construc ts identified by existing literature in one model
would be illogical because of the risk of ‘ overfitting ’ the model, thus possibly causing
difficulties of isolating the variables’ individual effects. Thus, based on the literature review,
two additional i ntrinsic motivators are derived. This being trust and f low, as they are
believed to contain a high likelihood of increasing the explanatory power of the research
models. Also, literature have called for further research to acknowledge the importance of
intrinsic motivators in a n m-commerce within the area of shopping (Hung et al., 2012) . In
addition, the latest study of Zhou (2014a) further encourages the inclusion of flow, as he
finds that flow followed by satisfaction , was the main factor driving users ’ continuance
intenti on to use mobile internet sites. The same result appeared in his earlier work in
relation to the use of mobile payment services (Zhou, 2013a) . Furthermore, the consistent

Chapter II – Theoretical Framework
Page 25 of 155
significant influence of trust on m -commerce behaviour s has also made it an object of
interest. Numerous studies have regarded t rust a critical antecedents of intention to adopt
m-commerce (Gitau and Nzuki, 2014 ; Tsu Wei et al., 2009; Vasileiadis, 2014) . Other research
have further confirmed that despite a repeating interaction with m -commerce , trust would
stay relevant and remain among determining factors (Chong, 2013; Hung et al., 2012; Zhou,
2013a) .

2.2.1. Trust

In literature , trust is regarded as a broad concept since it have been accommodated by
a range of different definitions depending on perspective and research context (Gefen et
al., 2003a; McKnight et al., 2002) . This study will , however, treat trust as a set of trusting
beliefs hold by the user, and therefore describe trust as the willingness of users to leave
themselves vulnerable to the actions of others, which is based on the expectations toward
the other party’s future behav iour (Mayer et al., 1995) . Thus it is the belief that the trusted
party will not take advantage of the situation and will behave in a dependable, ethical ly and
socially appropriate manner (Gefen et al., 2003a) . In this perspective, users ’ level of trust
with in m-vendors is believed to emerge from their perception s of specific attributes offered
by the m -vendor, which subsequently influence their attitude s and behaviou ral intentions.
This approach does also allow for a more rigorous integration with different behavioural
theories such as TAM and ECM, since it essentially follows the same logic as TRA (Fishbein
and Ajzen, 1975) , stating that individuals ’ beliefs indirectly influence behavioural intentions
through attitude , and are therefore more aligned with the theoretical foundation of these
models.

Mayer et al. (1995) argue that individuals’ level of trust is reflected by their belief s
concerning the other parties ’ benevolence, ability and integrity. Although Mayer et al.
(1995) discuss trust within organizations, t heir operational ization of trust is also heavily
applied within IS research (McKnight et al., 2002) . However, some researchers have also
included the dimension of predictability as a part of trust, while arguing that individuals who

Chapter II – Theoretical Framework
Page 26 of 155
perceive the other person to be predictable in his/her behaviours, may also be more willing
to depend on that person (Gefen and Straub, 2004; McKnight et al., 1998) . At the same
time, it seems reasonable to believe that m -vendors who demonstrate predictable and
consistent behaviours, e.g. who always deliver goods or services on time, would be
considered more trustworthy.

According to the definition of Mayer e t al. (1995) , ability can be defined as the extent to
which the user presumes m -vendor s to possess the sufficient knowledge and competencies
in order to fulfil their task. Benevolence is referred to as user caring, motivation to act in
their users’ best interest and their willingness to put their users’ interests above their own.
Integrity is defined as the m -vendors’ distance to any deceptive behaviour and their ability
to keep promis es. Finally, predictability is, according to Gefen et al. (2003a) , related to the
users ’ perception about the m -vendors’ behavioural consistency.

When investigating the saliency of trust within IS research, it becomes evident that
several researchers regard trust as a particularly critical element within online exchanges
(Gefen et al., 2003a; McKnight et al., 2002) , as it is suggested that online users generally
stay away from e-vendors, they do not trust (Jarvenpaa et al., 2000; Liu et al., 2005) . Some
researchers ev en argue that facilitating trust is essential for e -vendors to succeed within e-
commerce (Gefen, 2002; Kim et al., 2008) . Several studies have noted that the saliency of
trust generally increases in situations where consumers are facing uncertain situations (Lin
et al., 2014; Siegrist et al., 2005) . M-commerce is arguably an area that is attached with
many uncertainties. In fact, Vasileiadis (2014) suggests that the inherent nature of m –
commerce with its constant and rapidly evolving state , is associated with a relatively higher
degree of ri sks perceived by users compared to e -commerce and traditional offline
channels. Specifically, the possibility of tracking users ’ location and users ’ preferences raises
a major privacy concern that questions the benevolence of m -vendors (Joubert and Van
Belle, 2013) . Users may also suffer from a lack of trust in the technology they use. Users’
access m -commerce services via smartphones from wireless connections in different places,
which evokes not only transactions concerns, but also privacy concerns , since users may

Chapter II – Theoretical Framework
Page 27 of 155
feel vulnerable to hackers and malicious software (i.e. viruses, malware etc.) (Ghosh and
Swaminatha, 2001) . A commonly known phenomenon is blue -snarfing , where intruders
hack the Bluetooth system in the smar tphone , and thereby gain access to personal
information. Also, the wireless 3G/4G connection is often unstable , and users may therefore
be worried about the consequences of a lost connection during a transaction with the m –
vendor (Lim, 2003) . Additionally, as similar to e -commerce, m -commerce includes the
process of purchasing products/services in a virtual environment that consequently inhibit s
the users from accessing reliable indication on product/service quality. Research have
demonstrated that intangibility is closely correlate d with perceived risk (De Ruyter et al.,
2001) . By the same token , users may also be concerned about expenses they may endure if
they cancel , or need to return a prod uct (Vas ileiadis, 2014) . Thus, having favourable
perceptions concerning the m -vendors ’ trustworthiness , diminish the importance of
perceived risks (Lin et al., 2014) and increase the willingness to enter a vulnerable position ,
despite risks of receiving a negative outcome (Mayer et al., 1995) .

Several studies conducted within the m -commerce domain have empirically confirmed
the saliency of trust in influencing users ’ satisfaction (Lin and Wang, 2006; San ‐Martin and
López ‐Catalán, 2013) . For example, Lee and Chung (2009) incorpo rated trust within the
DeLone and McLean’s IS success model , and demonstrated that users ’ degree of trust was
a strong predictor of users ’ degree of satisfaction with mobile banking services. Chong
(2013) extended the ECM, revealing that trust was a key construct explaining users ’
satisfaction with Chinese m -commerce services. Researchers have also found a direct
connection between trust and users ’ behavioural intentions. For instance, the exploratory
analysis conducted by Sadi and Noordin (2011) found trust to be a n important construct
driving users ’ intention to adopt m -commerce in Malaysia . The longitudinal study of Lin et
al. (2014) integrated trust in the ECM -IS and Valence Theory, uncovering that pre – and post –
trust were critical factors affecting intention to use mobile banking, since pre -trust
diminished the saliency of perceived risk and enhanced the degree of perceive d benefit,
while post -trust, through a confirmation of trusting beliefs, would further influence future
behaviours. Additional research have found a significant relationship between trust and

Chapter II – Theoretical Framework
Page 28 of 155
continuance intention to use mobile payment services (Zhou, 2013a) and m-health (Akter
et al., 2013) . Factors such as syst em quality, information quality, service quality, perceived
risks and confirmation of expectations have been identified to influence users ’ trust in m –
commerce (Akter et al., 2013; Alsajjan, 2014; Lee and Chung, 2009; Vasileiadis, 2014; Zhou,
2013a) .

2.2.2. Flow

The origin of f low theory is found in the research papers of Csikszentmihalyi within the
human psychology domain, where he developed this theory by studying and interviewing
individuals that exhibited a high commitment and devotion toward an activity. E.g.
professional chess players playing chess or rock climbers climbing a mountain
(Csikszentmihalyi, 1975) . From his results, he conceptualized a particularly and extremely
gratifying state of mind that occurred when an individual pa rticipated in an activity with
total immersion, while experiencing a range of different positive characteristics, such as:
loss of self -consciousness, loss of time sense, sense of effortless control of the situation,
total centring of attention or an embra cement of the autotelic nature of the activity
(Csikszentmihalyi, 1997, 1975) . Achieving such state of mind is what Csiksze ntmihalyi
described as gaining “flow experience ”, or as his subjects verbalized as “ being in the f low”
(Csikszentmihalyi, 1975) . He de fined this phenomena as the “ holistic sensation present
when we act with total involvement ” (Csikszentmihalyi, 1975 p. 43) . Hereby,
Csikszentmihalyi – simply put – names th e feeling that occur when we are fully immersed
and engulfed in an activity, which in turn fills us with enjoyment and fulfilment.
The relevancy of f low t heory in an m -commerce context emerges from the work of
Hoffman and Novak (1996) , as they extended the applicability of the flow construct in order
to study and explain online experiences in a computer -mediated environment. Though
fitted to an online context, the definition proposed by Hoffman and Novak, (1996) still
draws on the fundamentals of the f low construct, as they define online flow experience as:
“The s tate occurring during network navigation, which is: (1) characterized by a seamless
sequence of responses facilitated by machine interactivity, (2) intrinsically enjoyable, (3)

Chapter II – Theoretical Framework
Page 29 of 155
accompanied by a loss of self -consciousness, and (4) self -reinforcing ” (Hoffman and Novak,
1996 p. 57) . This means that users could gain a state of mind similar to what
Csikszentmihalyi (1975) identifies by navigating in a network, as for instance web surfing or
browsing a webpage (Hoffman and Novak, 1996 ). However, in order for users to experience
online flow, Hoffman and Novak (1996) , similar to Csikszentmihalyi (1975) , propose that
two primary conditions need to be present. First, the users must fully focus their attention
on the interaction to a degree where they filter out any background noise and irrelevant
thoughts and secondly, strike a balance between skills and c hallenges. Within the
conceptual framework of Hoffman and Novak (1996) , the degree of users attention is
viewed as a consequence of content characteristics (interactivity, vividness) and
involvement, whi lst the degree of involvement is determined by the navigation process
characteristics (goal -driven, experiential -driven). Additionally, website performance and
prior website experiences have also been suggested to be factors contributing to online
flow exp erience (Skadberg and Kimmel, 2004) . Similar to Csikszentmihalyi (1975) , Hoffman
and Novak (1996) emphasize the importance of the ratio between users’ skill levels and the
challenges faced by users when navigating a network. If the users ’ skill level surpasses the
challenges they face, they will experience boredom. If the challenges faced by users surpass
their skill level, they will experience anxiety. Only when users possess a high perceived level
of skills and sense of control congruent with an equally high level of perceived severity of
the task at hand that evokes arousal, they will expe rience online f low (Novak et al., 2000) .
The interest in the flow construct is a consequence of its affect. The possibility of
harvesting intrinsic rewards, such as enjoyment and fulfilment when users experience flow
is suggestively correlated with a range of influential f actors affecting online behaviour s
(Hoffman and Novak, 2009, 1996; Novak et al., 2000; Siekpe, 2005; Skadberg and Kimmel,
2004) . For example, Skadb erg and Kimmel (2004) demonstrated that flow had a significant
positive influence on users learning abilities when browsing websites, meaning that the
presence of flow would increase the information processed by the user. In the meantime,
Korzaan (2003) revealed that f low was leading to a more exploratory behaviour when
shopping online and therefore also increased time spend on the website. Both studie s
conclude that the outcome of f low was significantly related to users ’ attitude toward the

Chapter II – Theoretical Framework
Page 30 of 155
activity. Novak et al. (2000) also further validated and empirically tested t heir 1996
framewo rk by finding f low to be stimulating positive affects (e.g. satisfaction) , and argued
that flow experience can mitigate users ’ price sensitivity. In essence, the influential role of
flow in a n online context builds on the augmentation that elements of the flow construct
are vital precursor for a pleasant and enjoyable online experience (Hoffman and Novak,
2009; Koufaris, 2002; Siekpe, 2005) . Researchers propose that the ratio b etween failure and
success of online marketers are mediated by their ability to facilitate and create exciting
online e xperiences that promote online f low (Bilgihan et al., 2014; Hoff man and Novak,
1996) . Bilgihan et al. (2014) opined that unsatisfactory online experiences are globally
accountable for a substantial loss in revenue. This proposition is well grounded since several
empirical studies have provided evidence that highlight the magnitude of online flow , and
its ability to influen ce users’ attitudes and behaviour al intentions. For instance, some
studies concluded that flow significantly influenced attitude towards online shopping
(Korzaan, 2003) , satisfaction with online shopping (Hsu et al., 2012; Rose et al., 2012;
Sharkey et al., 2012) , attitude with instant messaging (Lu et al., 2009) , attitude towards
website search engines (Chung and Tan, 2004) satisfaction with online financial services
(Lee et al., 2007a; Xin Ding et al., 2010) , satisfaction with e -learning systems (Cheng, 2014)
while other research found flow to also stimulate intention to purchase online (Sharkey et
al., 2012; Siekpe, 2005) and intention to re -visit webpage (Koufaris, 2002; Nel et al., 1999;
Siekpe, 2005) .
Although research of flow in an m -commerce context i s limited, the research that do
exist provide similar results. The studies mainly published by Tao Zhou clarified the
importance of flow in this setting by finding flow to be significantly related to both users ’
satisfaction with and continuance i ntention to use mobile payment s ystem (Zhou, 2013a) ,
mobile internet sites (Zhou, 2014a, 2013b, 2011) , mobile social network services (Gao and
Bai, 2014; Zhou et al., 2010) , intention to use mobile TV (Zhou, 2013c) and continuance
intention to use mobile internet sites (Zhou, 2014b) . Thus, studies seem to ag ree that the
intrinsic rewarding state created by experiencing flow can play a significant role in attitude
formation and s atisfaction evaluation. Also, as found in the very basic s of Csikszentmihalyi
(1975) , users experiencing flow will be likely to engage in repetitive behaviour s, as they will

Chapter II – Theoretical Framework
Page 31 of 155
be drawn to re -experience such gratifying state produced by f low.
In literature, flow is illustrated as an elusive concept. Researchers seem to agree o n the
conceptual definition of f low provided by Csikszentmihalyi (1975) . However, the intuitive
translation into a more universal operational definition seems much more complex, why
anteceden ts, as wel l as consequences of f low, differ depending on research context and
researcher (Obadă, 2013) . Consequently, this have also created diversity in relation to
measurement approache s. For example, Korzaan (2003) measures f low as a direct
unidimensional construct by providing subjects w ith a narrative description, followed by
three questions. Other research employ a deriv ed unidimensional measurement that
aggregates antecedents of flow into an overall measurement (Skadberg and Kimmel, 2004;
Zhou, 2014b) , while Koufaris (2002) approaches flow as a multidimensional construct ,
consisting of three separate dimensions. For simplicity reasons, and based on the
recommendation of Hoffman and Novak (2009) , approaching flow as a derived
unidimensional construct seems more fitted with the purpose of this study. Hoffman and
Novak (2009) note that a serious disadvantage of this approach is the fact th at it smears the
distinctions between consequences and antecedents, meaning that jus tifying which items
to reflect f low become somewhat more challenging. However, in order to align this study
with the definition provided by Hoffman and Novak (1996) , the a ntecedents chosen to
represent f low are based on the primary antecedents suggested in their framework. This
being perceived control, enjoyment and focused attention. Within a n IS context, perce ived
control captures the extent of users ’ perceived level of control over the their actions and
over the environment , in which they interact (Koufaris, 2002) . Perceived enjoyment reflects
users ’ level of intrinsic enjoyment or pleasure associated with the interaction, whereas
focused attention re flects users ’ immersion and measures their ability to focus their
attention on the interaction at hand (Koufaris, 2002; Zhou, 2014b) . These factors are also
congruent with frequently used measures of flow in m -commerce research (Zhou, 2014a,
2013a)

Chapter II – Theoretical Framework
Page 32 of 155
2.3. The Moderating R ole of Personal T raits

Research have r epeatedly shown that individual characteristics should not be ignored
when trying to gain a dee per understanding of underlying factors that drives users ’
intention to use technologies (Khedhaouria et al., 2014; Venkatesh et al., 2003) . Whether it
be a product, service or shopping method, individual s tend to differ in the amount of value
they place on the attributes related to the giving object or activity, which ultimately affects
their behavioural intentions. Thus, what is an important attribute for one individual is not
necessarily important for an other. Research within IS have often accused factors such as
age, gender and experience for playing a noticeable part in users ’ distribution of value
between attributes related to e -commerce (Pappas et al., 2014; Venkatesh et al., 2003) .
However, recent research in m -commerce suggest that also personal traits , such as users
adherence to buy impulsive ly (San ‐Martin and López ‐Catalán, 2013) and users ’ level of self –
efficacy (Yang, 2012) play an intervening role in users ’ perception of m -commerce services .
Users ’ degree of impulsiveness and self -efficacy will therefore be assessed .

2.3.1. Self -Efficacy

In a social cognitive learning perspective, human functioning is viewed as product of a
dynamic interrelationship among behaviour s, environment and personal factors (cognitive,
affective and biological events) (Bandura, 1986) . A concept that is coined reciprocal
determinism, m eaning that humans are able to interpret the result of their own behaviour s
that may influence their surrounding environment and cognition that subsequ ently
facilitate future behaviour s (Wood and Bandura, 1989) . It takes an inside out and outside in
approach to learning and behavioural changes , since it runs on the idea that learning and
change in behaviour can be extracted purely from expectancies (e.g. beliefs) and through
vicarious learning (Bandura, 2001; Bandura et al., 1961) , as opposed to traditional learning
theories (e.g. classical conditioning theory ). Social cognitive theory is founded on a human
agency perspective (Bandura, 1989) . In this sense, in order for individuals to function
successfully within the reciprocal framework , they exercise certain capabilities, which allow

Chapter II – Theoretical Framework
Page 33 of 155
them to contribute t o their own motivation, behaviour and development course (Bandura,
1986) . Such capabilities are referred to as: symbolize, forethought, vicarious learning, self –
regulation and self -reflection , whereas the self -reflection capability is regarded as a critical
factor in the social cognitive theory , since individuals , through self -reflection , analys e their
own cognition and self -beliefs (Bandura, 199 9). Within the self -reflection construct lays the
concept of self -efficacy, which is also a precursor for self -regulation that operate s in the
very foundation of the human agency perspective (Bandura, 1999, 1993) .

Self-efficacy can , within a computer usage context , be defined as users ’ judgment about
his or her capability to undertake a behaviour with confidence in successfully achieving a
desired outcome (Compeau and Higgins, 1995, p. 191; based on Bandura, 1986) . It is ,
according to Bandura (1997) , an important personal factor that functions as an intrinsic
motivator , and is hypothesized to influence , and to b e influenced by , the individual’s
behaviour and environment. The self -efficacy construct therefore offers a connection
between self-perception and individual behaviour (Chii and Braun, 1995) . Digging deeper
into the mechanism of self -efficacy , it reveals the potential saliency in the current research
domain . According to self -efficacy theory, beliefs about one ’s self -efficacy influence human
functioning by affecting how individual s feel, think, motivate themselves and behave
(Bandura, 1997) . Thus, also assisting in dividuals in deciding which activit ies to pursuit, how
much effort to allocate and their degree of persistency (Bandura, 1991) . It is suggested that
individuals with high se lf-efficacy will tend to view difficult tasks as challenges that should
be mastered rather than threats to be avoided (Bandura, 1994) . According to Bandura
(1977) , the most influential source , from which individual s judge their level of self -efficacy ,
is derived from direct authentic experiences. A sequence of successful experiences raises
self-efficacy appraisals , whereas failures lower them (Bandura, 1977) . Thus, in the centre of
the self -efficacy theory lies the belief that individuals ’ behaviours are often better explained
by their expectancies and beliefs about their own capabilities , more than what they are
actually capable of doing (Bandura, 1997, 1994, 1986) .

Chapter II – Theoretical Framework
Page 34 of 155
According to Compeau and Higgins (1995) the construct of self -efficacy is, within a
computer usage context , measured as a derived unidimensional construct with three
distinct but interrelated dimensions: strength, generalizability and magnitude. Self -efficacy
strength reflects the l evel of confidence the user has in accomplishing difficul t computing
tasks. Gener alizability reflect the degree to which the expectation is generalizable to a
specific domain , while users with a high self -efficacy magnitude imagine themselves to be
capable of accomplishing difficult computing tasks with little o r no support from others. This
operational definition is directly in line with the most -often applied definition within m –
commerce research (Trivedi and Kumar, 2014; Wang et al., 2006; Yang, 2012, 2010) .

Individuals may arguably perceive technologies as daunting challenges , since a
successful use may often be believed to require a certain degree of skill level and mental
effort. It is the refore expected that individual s’ decisions to use technologies may be guided
by their level of self -efficacy, which consequently forms different perceptions and
behaviours. In fact, users ’ level of self -efficacy in technology use is apparent ly of great
significan ce in nurturing and promoting the use of technologies. For example , users ’
perceived level of self -efficacy have been found to be a significant predictor of users ’
decision to use online s hopping services (Vijayasarathy, 2004) , online music services
(Bounagui and Nel, 2009) , continue using websites (Wangpipatwong et al., 2008) and more
importantly, to influence users ’ decision to use m -commerce services (Trivedi and Kumar,
2014; Wang et al., 2006) . The study of Hernández et al. (2010) reported that online sho pping
frequency affected users level of self -efficacy , while Compeau and Higgins (1995) found that
users with a high level of self-efficacy used computers more frequently and experienced
less computer anxiety. More relevant to this study is the diversities in users ’ perceptions of
determinant factors driving users ’ behavioural intentions. Current m -commerce literature
seem to have de voted less attention to examine the moderating effect of self -efficacy. The
study of Yang (2012) however, shed some light on the area, finding that users ’ level of self-
efficacy positively moderated the relations hip between enjoyment and attitude towards
mobile shopping . Yang’ s study also revealed that increased self -efficacy led to increased
control and consequently a higher intention to adopt mobile shopping . In the meantime,

Chapter II – Theoretical Framework
Page 35 of 155
Jaradat and Faqih (2014) demonstrated that users with a high level of self-efficacy were
more likely to perceive mobile payment as useful , and thus more likely to use it when
compared to users with a low level of self-efficacy. In summary, the evidence presented
abov e indicate that mechanisms of self -efficacy play an important role in motivating users
to engage in m -commerce activities , and high sens e of self -efficacy strengthen positive
perception and orientation of m -commerce.

2.3.2. Impulsiveness

The concept of impulsiveness has been heavily debated throughout the years, and has
thus been subject for changes in definitions along the way. In the early fifties, impulsiveness
was primarily regarded as signs of immaturity, primitivism, foolishness and other similar
social deviations (Park and Choi, 2013) , whereas the concept has evolved into a more
complex construct in contemporary literature. However, the consensus of general
characteristics of impulsiveness remain the same; that impulsiveness covers purchases with
low or non -existing prior planning, whi ch in turn is concluded to be irrational buying
behaviour (Etzioni, 1986; Park and Choi, 2013) . The interesting aspects of impulsiveness,
and the aspects that cause divergence between different studies are, however, what
activates this behaviour and which consequences , it has. Cobb and Hoyer (1986) regard
impulsive behaviour as the d ecision of buying a product made inside the store. Thus, the
consumer is assumed to have no intentions or plans for buying the product in question
before entering the store, but simply consciously experienced a latent need being brought
to life when being presented with the product. That is, a need that the consumer had not
previously recognized.

This study, however, ado pts the definitions of impulsiveness from Rook and Fisher (1995
p. 306) , defining that buying impulsively is “ a consumer’s tendency to buy spontaneously,
non-reflectively, immediatel y, and kinetically ”, which is acknowledged by several studies
within e -commerce (e.g. Chih et al., 2012; Parboteeah et al., 2009) as well as m -commerce
(e.g. San‐Martin and López ‐Catalán, 2013) , though findings wit hin the context of m –

Chapter II – Theoretical Framework
Page 36 of 155
commerce are rather limited. Rook (1987) and Rook and Fisher (1995) recognize that
impulsive purchasing behaviour refers to a more spe cific range of phenomena rather than
just unplanned purchase, thereby challenging the theories of Cobb and Hoyer (1986) by
distinguishing between the two. The difference, according to Rook (1987) , is that i mpulsive
purchasing behaviour occurs when consumers experience sudden powerful and persistent
urge to buy something immediately, as opposed to unplanned purchasing that is
characterized as more ordinary and tranquil. Impulsive behaviour is thus also, to a greater
extent, caused by emotional feelings for the product in the current cognitive state. An
apparent weighting factor of enhancing the chance s of impulsive behaviour among
consumers, is to increase the intensity of advertising of the product (Arens and Rust, 2012) ,
provided that these create positive feelings to the brand in question, and that these
associative feelings act as cues for rewards if purchasing the products. According to
Parboteeah et al. (2009) , these feelings might likely be induced by the consumer’s
enjoyment in a given situatio n when exposed to a product within e -commerce. Their
findings were that perceived enjoyment was in fact the primary explanatory variable of the
urge to buy impulsively. These findings are furthermore supported by Chih et al. (2012) ,
finding that exposing consumers to hedonic consumption needs in e -commerce will
enhance the positive affect s for the customers. That is, hedonic cons umptions needs being
exposure to product characteristics that enhance the positive affect (Chih et al., 2012) vis-
à-vis the consumer s’ mood (Parboteeah et al., 2009) or associative feelings (Arens and Rust,
2012) . In fact, several studies (e.g. Chih et al., 2012; Flight et al., 2012; R ook and Fisher,
1995) suggest that failure to induce positive affects, will severely impair the chances of
consumers repurchas ing. Rook and Fisher (1995) found that, in the case of consumers
experiencing negative affects, the general buying behaviour was significantly impaired, and
some impulsive consumers even managed to reject the need for impulsive shopping when
their normative evaluations were sufficiently negative.

In a virtual store context, such as m -commerce, there’s generally tradition for
persuading users t o engage impulsively (San ‐Martin and López ‐Catalán, 2013) . However,
acting impulsively severely increases the chances of overall dissatisfaction with the process,

Chapter II – Theoretical Framework
Page 37 of 155
as there are none or low planning prior to the purchase , and a low degree of other
considerations (Rook and Fisher, 1995) . However, though the outcome of a negative affect
has proven to be imperative for most people s’ intention to engage in a similar behaviour
prospectively , George and Yaoyuneyong (2010) stress that highly impulsive people are more
risk tolerant, and are therefore less likely to let prior bad experiences impact future
behaviour .

Construct Conceptualization Object Study
Flow A holistic sensation present when we
act with total involvement Human psychology Csikszentmihalyi
(1997 )
Perceived Ease
of Use The degree to which a person believes
that using a particular system would be
free of effort User acceptance of
IT Davis (1989 )
Perceived
Usefulness The degree to which a person believes
that using a particular system would
enhance his or her job performance’ User acceptance of
IT Davis (1989 )
Trust The willingness of a party to be
vulnerable to the actions of another
party based on the expectation that the
other will perform a particular action
important to the trustor, irrespective of
the ability to monitor or control that
other party Organiza tional Trust Mayer et al. (1995)
Confirmation Users’ perception of the congruence
between expectation of a system and
its actual performance Continuance
Intention of IT Bhattacherjee (2001 )
Satisfaction Users’ affect with (feeling about) prior
use Continuance
Intention of IT Bhattacherjee (2001 )
Continuance
Intention Users’ intention to continue using a
system Continuance
Intention of IT Bhattacherjee (2001 )
Self-Efficacy Belief about one’s ability to perform a
specific behaviour with confidence in
achieving positive task outcomes Information systems Yang (2010 ),
adopted from
Compeau and
Higgins (1995 )
Impulsiveness A consumer’s tendency to buy
spontaneously, non -reflectively,
immediately, and kinetically Buying behaviour Rook and Fisher
(1995 )
TABLE 2.1. CONSTRUCTS AND THEIR ORIGINS

Chapter III – Conceptual Model & Hypotheses
Page 38 of 155

– Chapter III –
CONCEPTUAL MODEL & HYPOTHESES

“Yes, I sell people things they don't need. I can't, however, sell them something they don't
want. Even with advertising. Even if I were of a mind to.“ – John O'Toole

Chapter III – Conceptual Model & Hypotheses
Page 39 of 155
3.1. Hypothesis Development

In this chapter, a hybrid conceptual model is developed based on the ECM -IS and TAM,
given that literature have found significant convergence between the explanatory powers
of some variables in both contexts. Furthermore, the model is extended with trust and flow,
as both variables have frequently been positively associated with important user beliefs
regarding continuance usage intention of IS .

3.1.1. Study O ne

The hypotheses development of study one is two -fold. The first set of hypotheses
concerns the mechanisms of inflicting satisfaction. The second set regards the measures to
enhance the likelihood of users’ continuance intention.

3.1.1. 2. Creating S atisfaction through P arallel Mediation

The relationship between confirmation and satisfaction

According to the ECM -IS, user s’ degree of satisfaction is derived from a function of
expectations and expectancy confirmation, meaning that users’ evaluation of m -commerce
performance weighted against their initial pre -adoption expectations, will determine the
users’ degree of satisfaction with m -commerce usage (Bhattacherjee, 2001; Oliver, 1980) .
Confirmation is therefore positively connected with satisfaction because a confirmation of
expectations of the m -commerce interaction would mean a fulfilment of expected benefits
(Bhattacherjee, 2001) . For instance, mobile banking users attributed t heir s atisfactio n with
the banks’ ability to provide accurate, complete and relevant information (Lee and Chung,
2009) . Similarly, mobile users are more likely to be satisfied with mobile carriers, if they are
able to provide a high network quality and competent customer service (Lim et al., 2006) .
Additionally, in consumer research, Oliver and DeSarbo (1988) found that consumers with
high initial expectations would be more likely to form a high degree of satisfaction , if their
expectations were co nfirmed and subsequently, Brown et al. (2012) found, by drawing on

Chapter III – Conceptual Model & Hypotheses
Page 40 of 155
prospect theory, that negative disconfirmation consequences outweigh the consequences
of positive confirmation in relation to users ’ attitude formation towards a n intranet -system.
Thus, it seems reasonable to say that that users’ degree of satisfaction depends on the
degree of (/dis)confirmation of expectations. Several m -commerce studies have supported
this link between confirmation and satisfaction (Kim, 2010; Lee, 2014 ; Lin et al., 2014; Thong
et al., 2006)

3.1.1.2. TAM

The relationship between confirmation and TAM

The variable of perceived usefulness covers the extent to which degree a system
enhances the task performance for a user (Davis, 1989) . In an m -commerce context it thus
refers to the users’ enhanced performance by using m -commerce as opposed to doing the
same task without it (Chong, 2013; Lee, 2014 ). However, the definition is slightly different
when combined with the ECT, since the user, as opposed to the situation in the original
TAM, has already used the system before. Lin et al. (2014) and Bhattacherjee (2001) refer
to the variable as a post -adoption expectation, meaning that the already experienced
usefulness will affect the perceived usefulness for future use. Researchers (e.g. Chong,
2013; Hong et al., 2006; Thong et al., 2006) have heavily hypothesized that confirmation of
pre-consumption expectations will impact the cognitive attitudes regarding future use. Such
cognitive attitudes have, in a technological cont ext, often been hypothesized to be aspects
of the TAM. E.g. Thong et al. (2006) , who found significant correlations between
confirmation and perceived usefulness as well as perceived ease of use. In other contexts,
there have been somewhat more divergent results as to what these correlation represent.
For instance in the context of m obile internet, Hong et al. (2006) found relatively little
correlation between confirmation and perceived ease of use as well as an almost non –
existing correlation between confirmation and perceived usefulness. Chong (2013) on the
other han d, found highly significant correlations between confirmation and the two
discussed variables in his study regarding m -commerce. The reason for these differences

Chapter III – Conceptual Model & Hypotheses
Page 41 of 155
could possibly be explaine d by the difference in context, which in turn impedes a congruent
continuity of results.

The relationship between TAM and satisfaction

Davis et al., (1989) found that perceived usefulness and perceived ease of u se had a
major significant impact on users’ affect/attitude toward the ad option of technology in the
TAM (Davi s et al., 1989). However, though the concept was developed for adoption of a
technology, Thong et al. (2006) argue that since s atisfaction is a type of affect, perceived
usefulness and perceived ease of u se can equally be applied as indicators for the satisfaction
of an already tested technology. Thus, the more useful (Bhattacherjee, 2001; Lin et al.,
2014) and free of effort (Chong, 2013; Hong et al., 2006) m-commerce has been for the
respective user so far, the higher the deg ree of post -adoption expectation will be, and thus
also the satisfaction with m -commerce in respect to the ECT.

The mediating role of TAM

In general, it seems there are substantial evidence in literature suggesting the effects
betwe en confirmation and s atisfaction. There have, however, been divergent views as to
what nature the correlation has. Several studies have thus hypothesized different cognitive
components to represent the extent of correlations between the two variables. One of the
more frequent assumptions is that elements from the TAM represent comprehensive
explanatory value in the correlations, as for instance Thong et al. (2006) , who acknowledge
the importance of including elements from the TAM to better explain the elements that
influence the variance of s atisfaction, though no actual mediation effects were examined.
That particular study wa s, however, investigated in depth, as Al-Jabri (2015) suggested the
necessity to clarify the nature of correlations be tween external variables (i.e. training and
communica tions) and u ser satisfaction. By implementing perceived ease of use and
perceived usefulness a s mediators, which in his study are called ease of use and benefits,
he identified part ial mediation, suggesting that p erceived ease of use and perceived

Chapter III – Conceptual Model & Hypotheses
Page 42 of 155
usefulness both play a significant role when explaining the significant relation between
training and s atisfaction (Jabri, 2015) . Burton -Jones and Hubona (2006) proposed a similar
layout, using the same mediators between system experience and u sage volume within IS
acceptance, though he failed to identify the effects (Burton -Jones and Hubona, 2006) .

As the ECT builds upon the premise that the users’ have already used the system at least
once before, theory has it that they have a certain degree of experience, and that this
experience will evolve in the event of multiple usage. Moreover, as the experie nce changes
so do the relating variables (Bhattacherjee, 2001) , why c onfirmation in the ECT, to some
degree, can be compared to both system experience and system training, why we propose:

H1a: Users’ extent of Perceived Ease of Use will mediate the relationship bet ween
Confirmation and Satisfaction (CON  PEOU  SAT)

H1b: Users’ extent of Perceived Usefulness will mediate the relationship between
Confirmation and Satisfaction (CON  PU  SAT)

3.1.1.3. Trust

The relationship between confirmation and trust

The definition of trust operationalized in this study reflects the users ’ confidence in m –
vendor s’ trustworthiness, which allows them to enter a vulnerable position in which they
are depend ent on the actions of the m -vendors (Mayer et al., 1995) . It is the users ’
expectations that m -vendors can be relied upon to fulfil their promises and will not engage
in opportunistic behaviours (Gefen et al., 2003a) . Gefen and Straub (2004) suggest that e –
service providers can improve customers’ trusting beliefs by providing trustworthy
attributes (integrity, predictability, ability and benevolence) that are consistent with
customers’ expectations. Individual s’ level of trust increases when the other party displays
behaviours or other indicators that matches their expectations (Johnson and Grayson, 2005;

Chapter III – Conceptual Model & Hypotheses
Page 43 of 155
Zhou, 2013a) . It is therefore reasonable to expect tha t regardless of prior experiences, a
confirmation of m -vendors’ trustworthiness will lead to favourable trusting beliefs.
Empirical evidence supporting this relation are found in m -commerce (Chong, 2013; Lin and
Wang, 2006) and in mobile apps (Akter et al., 2013) .

The relationship between trust and satisfaction

The underlying mechanism in which trust suggestively influence satisfaction is based
directly on the fundamental logic of Ajzen (1985) and Oliver (1980) , expecting that positive
trusting beliefs will lead to attitude/satisfaction. Pavlou and Fygenson (2006) opined that
favourable trusting beliefs create positive perception s about the outcome of e -vendors ’
actions that consequently evoke positive attitudes. Specifically, trust mitigate risk
perceptions, e.g. beliefs about being exploited or mistreated by the e -vendor, which in turn
positively affect users ’ attitude (Jarvenpaa et al., 2000) . Ziaullah et al. (2014) even advocate
that trust is fundamental for creating satisfied and committed customers. Meanwhile , in
the longitudinal study of Kim et al. (2009) , it was concluded that users ’ trust with in e-
vendors was a critical facto r that not only guided their initial purchase decision s, but also
impacted their evaluation s of the experience s and further shaped their long -term decisions.
Additional m -commerce research have provided support for the correlation between trust
and satisfa ction (Chong, 2013; Lin and Wang, 2006; San ‐Martin and López ‐Catalán, 2013) .
It is therefore reasonably to believe that a fulfilment of users ’ trusting beliefs will positively
influence their satisfactio n with m -vendors.

The mediating role of trust

The studies of Hung et al. (2012) and Thong et al. (2006) encouraged re searchers to look
beyond the extrinsic constructs of TAM and thereby also investigate intrinsic variables,
through which the effect of confirmation on satisfaction could be mediated. Building on
these studies , and as indicated by the finding presented so f ar, this study infer s that the
effect s of a positive confirmation of one ’s expectations, e.g. m -vendor s behave as expected

Chapter III – Conceptual Model & Hypotheses
Page 44 of 155
of them, will translate into a higher degree of trust that subsequently positively influence
users ’ degree of satisfaction. In an off line banking context, the study of Carlander et al.
(2011) found that the positive effect derived from delivering a n acceptable degree of service
quality positively influenced customers ’ degree of satisfaction. T his effect was , however,
fully explained by customers’ degree of trust in the bank. Therefore , based on the presented
evidence, it is expected that:

H1c. Users’ exten t of Trust will mediate the relationship between Confirmation and
Satisfaction (CON  TRU  SAT)

3.1.1.4. Flow

The relationship between confirmation and flow

Recall that flow is referred to as the pleasant feeling and enjoyment derived from a
match between skill level and challenge that allowed for immersion and acting with a sense
of total control (Hoffman and Novak, 1996) . Thus, by incorporating the flow construct within
the ECM -IS framework, it implies that a confirmation of users’ expectations towards m –
commerce will have a direct impact on their degree of flow elicited from the interaction.
Support for this propositio n is found within an e -learning context. Specifically, Cheng (2014)
connected the flow construct with ECM theory , and demonstrated that a confirmation of
nurses ’ expectations towards e -learning systems were in fact directly related to the amount
of flow experienced. Similar, Sørebø et al. (2009) found a significant positive relationship
between expectancy confirmation and teachers ’ level of intrinsic motivation towards e –
learning.

The relationship between flow and satisfaction

According to Csikszentmihalyi and LeFevre (1989, p. 816) , individuals engaging in
activities in which they experienced flow reported to “ feel more active, alert, concentrated,

Chapter III – Conceptual Model & Hypotheses
Page 45 of 155
happy, satisfied, and creative ”, and according to Hoffman and Novak (1996 p. 58) , in a
computer -mediated environment, “ resulting in a state of mind that is extremely gratifying ”.
This makes it reasonable to believe that flow experience may in fact influence users ’ degree
of satisfaction with m -commerce. By the same token, the m -commerce activities
investigated in this study bears both hedonic and utilitarian motives , e.g. shopping, ticketing
etc., why users may expect to obtain a pleasant and enjoya ble experience in order to be
fully satisfied. Relating flow theory to an m -commerce context , it implies that some
predefined conditions might be necessary in order for users to gain an ‘ optimal experience’
(Hoffman and Novak, 1996) . For example, users are required to possess a certain degree of
knowledge and skills prior to the m -commerce interaction bef ore flow is to be experienced,
meaning that the user’s knowledge about m -commerce and restrictions of the smartphone
will not impede the m -commerce experience. E.g. small screens, small bottoms, insufficient
understanding of mobile security , complex interfaces or general unfamiliarity with mobile
use might lead an unskilled user to feel a lack of control. Research have showed that
experienced customers tend to feel more in control when shopping online and are therefore
more satisfied (Pappas et al., 2014) . Furthermore, if the users are vulnerable to distractions
while using m -commerce , it may prevent them from fully focusing on th e interaction th at
may lead to dissatisfactory . In online banking, Lee et al. (2007 ) argued that lacking fa ce-to-
face contact and the general distracting environment associated with using a computer , e.g.
pop-ups, e -mails, instant messages etc. , diminished the customers’ ability to focus on the
interaction and therefore caused the customer to be dissatisfied. This study therefore
expect s to find that users experiencing flow through m -commerc e interactions will
consequently show positive perceptions towards the m -comme rce experience that may
account for a significant proportion of users ’ overall evaluation of the m -commerce
experience , i.e. influence their degree of satisfaction. Previous studies within m -commerce
were also found to deliver evidence for this relationship, in mobile internet sites (Zhou,
2014a, 2013b, 2011) and in mobile payment services (Zhou, 2013a)

Chapter III – Conceptual Model & Hypotheses
Page 46 of 155
The mediating role of flow

In essence, a realization of expected benefits of m -commerce usages , i.e. a confirmation
of users expectations would spawn a positive psychological state (Bhattacherjee, 2001) and
increase the level of flow perceived by users (Cheng, 2014) , which would further transl ate
into a n increased level of satisfaction (Hsu et al., 2 012; Novak et al., 2000; Zhou, 2014a) .
Hence, it is suspected that flow might facilitate some of the effect between confirmation
and satisfaction, meaning that users ’ flow experience will only impact their degree of
satisfaction, if their initial expectations regarding flow towards m -commerce are confirmed,
while users whose expectations are not confirmed , will be unl ikely to form a high degree of
satisfaction. This leads to the following hypothesis:

H1d: Users’ e xtent of Flow will mediate the re lationship between Confirmation and
Satisfaction (CON  FLO  SAT)

3.1.2. Creating Continuance I ntention through Single Mediation

The relationship between satisfaction and continuance intention

In accordance with ECM -IS users decision to re -use m -commerce should be guided by
their initial degree of satisfaction with the system (Bhattacherjee, 2001) . A more intuitive
explanation for this relationship is found within the construct of satisfaction. According to
Choi et al. (2008) , satisfaction within m -commerce is represented by an aggregation of
positive, negative, or indifferent feelings acc umulated through multiple interactions with
m-commerce. This is, however, similar to traditional offline satisfaction, meaning that
satisfaction is expressed through an affective s tate, which is suggestively influencing
behaviour al intentions directly and indirectly through attitude (Oliver, 1980) . Marketing
literature genera lly agree that consumers’ degree of satisfaction holds a high explanatory
power in relation to one ’s decision to patronize goods or services (Oliver and Bearden, 1985;
Swan and Trawick, 1981) . Numerous studies wit hin IS also support this relationship (Chen

Chapter III – Conceptual Model & Hypotheses
Page 47 of 155
and Chou, 2012; Halilovic and Cicic, 2013; Mohamed et al., 2014) . Consequently, this study
infers that an equivalent relationship is present within an m -commerce context. P revious
m-commerce research do also support this relation (e.g. Hong et al., 2006; Hung et al., 2012;
Kim, 2010) .

3.1.2.1. TAM

The relationship between perceived ease of use and continuance intention

In the context of continuance usage intention, research stress that a technology that is
perceived easier to use will lead to a higher probability of continuing using a technology, as
opposed to a technology that is perceived more advanced (Davis, 1989; Thong et al., 2006) .
This claim is furthermore supported by Hong et al., (2006), claiming th at users’ perception
of easiness will continuously be enhanced, as the users will gain more experience from the
usage of the system, hence creating more familiarity with the system (Hong et al., 2006;
Thong et al., 2006) . Therefore, given fact that users will gain more experience by using the
system more frequently, Bhattacherjee (2001) advocated for the necessity to include
addition al attitudes in the context of continuance i ntention, given that this situation will
indeed expand the paradigms of usage (Bhattacherjee, 2001) . Perceived ease of u se has
thus frequently been subject for postulations regarding its proclaimed direct imp act on
continuance i ntention. However, though several studies have found a significant direct
impact in various extents (e.g. Hong et al., 2006; Thong et al., 2006) , other studies have
failed to identify this correlation (e.g. Chong, 2013; Zhou, 2011) . Thus, it seems there are
inconsistent evidence whether or not this link exists. However, realizing the absence of
impact, Zhou (2011) proved that his definition of perceived ease of u se had a significant
impact on c ontinuan ce i ntention if mediated by s atisfaction. Though there are
inconsistencies, it’s our assumptions that:

H2a. Users’ extent of Satisfaction will mediate the relationship between Perceived Ease
of Use and Continuance Intention (PEOU  SAT  CI)

Chapter III – Conceptual Model & Hypotheses
Page 48 of 155
The relationship between perceived usefulness and continuance intention

Though perceived u sefulness was originally intended to prove the impact on the
intention to adopt a technology, comprehensive research indicate that the variable is also
viable to impact the intention of continuance us age, as people generally strive for rewards
and to utilize a situation as much as possible no matter the timing (Bhattacherjee, 2001) .
Consequently, the linkage between perceived usefulness and continuance i ntention has
been proposed in various studies. For instance in the context of IS (Bhattacherjee , 2001) ,
data services (Kim, 2010) and within m -commerce (Chong, 2013; Lin et al., 2014) . However,
the study of this impact has shown divergent results, why this proclaimed linkage is rather
questionable.

When demonstrating the direct correlations between perceived usefulness and
continuance i ntent ion, Bhattache rjee (2001) further found that satisfaction with an IS
mediated an indirect influence between the two variables. Lin and Wang (2006) support
this thesis , claiming that s atisfaction with an IS is determined by two aspects: c onfirmation
and perceived value , thus leading us to hypothesize that:

H2b. Users’ extent of Satisfaction will mediate the relationship between Perceived
Usefulness and Continuance Intention (PU  SAT  CI)

3.1.2.2. Trust

The relationship between trust and continuance intention

According to Mayer et al. (1995) when the degree of t rust exceeds a threshold value of
perceived risks, the person will be motivated to enter a vulnerable position, even though
risks are present. Researchers within IS therefore generally agree that trust can directly
influence behavioural intentions since it , as mentioned earlier, diminishes risk perceptions
(Kim et al., 2008) . Liu et al. (2005) proposed an online “ privacy -trust -behaviour model ” in

Chapter III – Conceptual Model & Hypotheses
Page 49 of 155
which trust is place d as a primary factor driving users ’ re-purchase intention. Ziaullah et al.,
(2014) stressed that online customers prefer to revisit e -vendors that they trust the most.
Likewise, the study of Gefen et al. (2003b) revealed that repeating customers tended to
have a high er degree of trust with the e -vendor. As revealed in the literature review, this
connection between trust and continuance intention has also been confirmed within m –
commerce (Akter et al., 2013; Hung et al., 2012; Zhou, 2013a) .

Since the current m-commerce researchers generally agree that trust is an important
antecedents of both user satisfaction and continuance intention (Akter et al., 2013; Chong,
2013; Lin and Wang, 2006) , this study infers that some of the effect inflicted by trust on
continuance Intention, may in fact be caused by satisfaction. In other words, users ’ trusting
beliefs may influence their continuance intention through an affective state of mind , and
may therefore also explain the relationship between trust and continuance Intention. The
study of Lin and Wang (2006) dealt with this area , as they reported an indirect effect of trust
on users ’ intention to patronize m -commerce services that operated through satisfaction.
It though remains unclear if t he indirect effect was significant , since mediation analysis was
not their primary objective. In a n offline service context, Caruana (2002) demonstrated that
customers ’ satisfaction with a firm mediated the relationship between service quality and
service loyalty. This is interesting since service quality taps into similar dimension as found
in trust, such as promise keeping and consistency. In consonance with such finding , it is
expected that:

H2c. Users’ extend of Satisfaction will mediate the relationship between Trust and
Continuance Intention (TRU  SAT  CI)

Chapter III – Conceptual Model & Hypotheses
Page 50 of 155
3.1.2.3. Flow

The relationship between flow and continuance intention

According to Hoffman and Novak (1996) , experiencing flow will motivate users to repeat
past behaviours , since users will be drawn to re -experience the intrinsically rewarding state
produced by flow. This theory was ev idently also true within m -commerce , since it was
concluded from the literature review that flow contained the ability to not only influence
users ’ intention to adopt m -commerce (Zhou, 2013c) , but also shaped their post -adoption
intentions (Zhou, 2013a) . Additionally, considering flow as separate constructs reveals that
dimensions representing flow in this study independently have been found to directly
influence behavioural intentions. For example, in a n e-commerce context, Koufaris (2002)
found that users ’ level of enjoyment experienced when buying books online was positive ly
related to their intention s to return. Identical results are also found within m -commerce
(E.g. Chong, 2013; Cyr et al., 2006) . Moreover, Kim (2010) clarifies the importance of
matching users ’ skill level with challenges faced when using mobile data services, since he
demonstrated that users ’ perceived b ehavioural control were just as important a
determinant of users ’ continuance intention as their perceived usefulness. In addition, Lu
et al. (2009) found that users ’ perceived ability to concentrate on instant messages services
were positively related to their intention to use it.

As presented in the literature review, some studies have also investigated the flow
construct in coherence with both satisfaction and continuance intention (Gao and Bai, 2014;
Hsu et al., 2012; Zhou, 2014a, 2013a) . These studies all concluded that flow positively
influenced both users ’ satisfaction and their intention to re -engage in the given activity.
Additionally, the gravity of flow and its ability to facilitate behaviours and foster positive
affects are directly highlighted in the study of O’Cass and Carlson (2010) , since they
distinguished consequences of flow , finding that users experiencing flow in a sports teams
website , expressed a higher degree of satisfaction, aroused feelings, website loyalty and
intention to engage in positive word of mouth. In essence, users expect to experience a

Chapter III – Conceptual Model & Hypotheses
Page 51 of 155
pleasant and compelling experience as well as perceived control, enjoyment and focused
attention when using m -commerce (Zhou, 2014a) . A confirmation of t hese expectations will
likely lead to favourable flow perceptions, which increase sa tisfaction and ultimately cause
a greater intention to patronize m -commerce services, why we propose that:

H2d: Users’ extent of Satisfaction will mediate the relationship between Flow and
Continuance Intention (FLO  SAT  CI)

FIGUR E 3.1. THE CONCEPTUAL MODEL OF STUDY ONE

Chapter III – Conceptual Model & Hypotheses
Page 52 of 155
3.2.1. Study Two

The hypotheses of this study is meant as extensions to study one. Study two thus builds
upon some elements of study one, and hypothesize that the projected findings are affected
by users’ personal characteristics.

3.2.1 .1. Moderating R ole of Self-Efficacy

The construct of self -efficac y reflects individuals ’ perception of their abilities to perform
a specific activity (Compeau and Higgins, 1995) . As already stated, conducting m -commerce
related activities is far from hassle -free (Shao Yeh and Li, 2009) . Some users may be more
familiar with m -commerce services , and thus have more confidence in their abilities to
produce a positive outcome when using m -commerce services , why such users may
approach the challenge of m -commerce with greater intrinsic mot ivation, hence be more
committed and devote more effort in completing the activity , compared to users with a low
degree of self-efficacy. This eventually leaves room for a more favourable perception of m –
commerce properties. Findings from the literature re view clearly demonstrated that users
with a high degree of self-efficacy have stronger perception s of intrinsic and extrinsic factors
associated with m -commerce , such as enjoyment, control and usefulness compared to
users with a low degree of self-efficacy (Jaradat and Faqih, 2014; Yang, 2012) . Both
perceived usefulness and perceived control are important constructs influencing users ’
satisfaction with m -commerce (Zhou, 201 4a). In addition, users ’ self-efficacy was found to
facilitate users ’ decision s to use m -commerce (Trivedi and Kumar, 2014; Wang et al., 2006) .
So, it is expected that the magnitude of users ’ degree of satisfaction influences continuance
intention differently according to users ’ degree of self-efficacy. In essence, users with a high
degree of self-efficacy may be more likely to act on information related to their degree of
satisfaction , wherea s users with a low degree of self-efficacy may hesitate to act on such
information due to a n insufficient belief in one ’s abilities to produce a positive outcome.
Thus, users with much self -efficacy are expected to show a significant ly stronger
relationshi p between satisfaction and continuance intention , compare d to users with low
self-efficacy.

Chapter III – Conceptual Model & Hypotheses
Page 53 of 155
Assuming that user s’ level of self -efficacy moderates the relationship between
satisfaction and continuance intention, it is also plausible that the indirect effect of flow ,
through satisfaction , on continuance intention could be conditioned by users ’ level of self –
efficacy, thus demonstrating the anticipated situation of moderated mediation. Specifically,
users experiencing flow when using m -commerce services would increase their degree of
satisfaction, which subsequently increase s their continuance intention. This effect is ,
however, expected to be stronger for users with a high level of self -efficacy. For this purpose
the following hypothesis is constructed:

H3a. The indirect effect between Flow and Continuance Intention through Satisfaction
will be moderated by the degree of Self -Efficacy (FLO  SAT  CI)

3.2.1.2. Moderating R ole of Impulsiveness

Within online shopping, much research have been conducted on impulsive behaviour
among users. Impulsiveness has thus been acknowledged as an antecedent of repurchase
intentions within e -commerce (Drossos et al., 2014) . Another important antecedent is that
users are generally satisfied with prior experiences (Chih et al., 2012; Flight et al., 2012) .
However, it seems that the relative importance of satisfaction is indeed dependent by the
degree of impulsiveness. In the case of users with less impulsive tendencies, several studies
argue (e.g. Arens and Rust, 2012; Chih et al., 2012; Parboteeah et al., 2009) that ensuring a
satisfactory buying process for users is far more important than focusing on inducing urgent
positive feelings toward a product or service ; at least when trying to enhance continuance
intention. For highly impulsive people, on the other hand, these urgent needs launched by
sudden induced positive af fects are evidently a viable option in the persuasion process, why
the overall satisfaction is relatively less important due to the higher risk tolerance . It thus
seems plausible that vendors must focus on striking balance between inducing impulsive
behavi our, and creating positive affects according to users’ degree of impulsiveness, not to
impair the chances of continuance intention, since especially non -impulsive users incline to

Chapter III – Conceptual Model & Hypotheses
Page 54 of 155
use past experiences as heuristics for future behaviour (Flight et al., 2012; Hsu et al., 2012;
Rook and Fisher, 1995) .

Parboteeah et al. (2009) argue that one of the most important influents of impulsive
behaviour is perceived enjoyment. Theref ore, since flow consists of users’ enjoyment while
using m -commerce services, flow can assumedly induce impulsive behaviour . This thesis is
furthermore supported by Hsu et al. (2012) , whose study found flow to significantly
influence impu lsive buying behaviour within the context of internet shopping.
Consequently, according to contemporary findings, experiencing flow will likely cause both
a direct positive effect on continuance intention, as well as an indirect effect through
satisfaction, though this effect might diverge significantly when considering the users’
degree of impulsiveness, why is it the assumption that:

H3b. The indirect effect between Flow and Cont inuance Intention through
Satisfaction will be moderated by the degree of I mpulsiveness (FLO  SAT  CI)

FIGURE 3.2. THE CONCEPTUAL MODEL OF STUDY TWO

Chapter III – Conceptual Model & Hypotheses
Page 55 of 155
Hypotheses Relation
H1a Users’ extent of Perceived Ease of Use will mediate the relationship
between Confirmation and Satisfaction CON  PEOU  SAT
H1b Users’ extent of Perceived Usefulness will mediate the relationship
between Confirmation and Satisfaction CON  PU  SAT
H1c Users’ extend of Trust will mediate the relationship between
Confirmation and Satisfaction CON  TR  SAT
H1d Users’ extend of Flow will mediate the relationship between
Confirmation and Satisfaction CON  FL  SAT
H2a Users’ extent of Satisfaction will mediate the relationship between
Perceived Ease of Use and Continuance Intention PEOU  SAT  CI
H2b Users’ extent of Satisfaction will mediate the relationship between
Perceived Usefulness and Continuance Intention PU  SAT  CI
H2c Users’ extend of Satisfaction will mediate the relationship between
Trust and Continuance Intention TR  SAT  CI
H2d Users’ extent of Satisfaction will mediate the relationship between Flow
and Continuance Intention FL  SAT  CI
H3a The indirect effect between Flow and Continuance Intention through
Satisfaction will be moderated by the degree of Self -Efficacy. FL  SAT  CI
H3b The indirect effect between F low and Continuance Intention through
Satisfaction will be moderated by the degree of I mpulsiveness. FL  SAT  CI
TABLE 3.1. HYPOTHESES

Chapter IV – Methodology
Page 56 of 155

– Chapter IV –
METHODOLOGY

“If you’re a good marketing person, you have to be a little crazy.” – Jim Metcalf

Chapter IV – Methodology
Page 57 of 155
The following chapter reviews the blueprint of the data collection. It thus assesses the
approach of the stud ies as well as the establishment of the data collection methods and the
instruments used

4.1. Research Design

The purpose of this thesis is to clarify the nature of the relationship of , and among ,
factors driving users ’ continence intention to use m -commerce , through testing a
conceptual model that is based on findings reported by previous research , and thereby
produce results that can pot entially be applied to the general population of m -commerce
users. This study therefore used a conclusive research design and collect ed cross -sectional
data through an online , as well as an offline , survey consisting of highly structured
questions. This ap proach allows for a fast, inexpensive and geographically wide data
collection (Malhotra and Birks, 2006) , while also, the quantitative nature supports large
sample sizes and generalization of results (Adams et al., 2007) . The data were collected
among individuals , who were current ly m-commerce users at the time of the survey , and to
the degree possible, users representing the age groups reflecting the proportion of m –
commerce users in the general population .

4.2. Instrument Development

4.2.1. Study One

The conceptual model of study one consists of s even constructs: Confirmation (CON) ,
Perceived Ease of Use (PEOU) , Perceived Usefulness (PU) , Trust (TR), Flow (FL), Satisfaction
(SAT) and Continuance Intention (CI). To ensure content validity, the items chosen to
represent the constructs were derived from pre-existing literature2. These items are
therefore pre -validated and already customized into an m -commerce context. The items

2 Appendix 1

Chapter IV – Methodology
Page 58 of 155
chose n to represent satisfaction were originally adopted from Bhattacherjee (2001) .
However, due to the importance of an accurate translated presentation to the Danish
population, the scale was rather problematic , as the items are developed by words rather
than sentences. This way, this study failed to present the items in the way intended by the
author. Instead, this study adopted the three items scale of satisfaction from Lee (2014 ),
which seemed more appropriate to apply .

A self-report method was applied on items used to test the conceptual model , and were
measured on a seven -point -likert -scale from 1 ( strongly disagree ) to 7 ( strongly agree ).
However , though this measurement method grants access to phenomenological data , it
also comes with validity concern s related to self -deception, self-serving biases (Robins et
al., 2007) , as well as the risk that respondents don’t fully comprehend the meaning of the
objective of the study . Thus, to ensure a consistent view on m -commerce -relate d activities,
a thorough definition was provided to the respondent s at the beginning of the survey.
Furthermore, t he two first questions were designed as screening questions , were
respondents were asked if they currently posse ss a smartphone , and if they h ave previously
conducted any m -commerce activities related to the definition provided at the beginning .
Thus, r espondents that did not answer affirmative ly were excluded from the survey. In
addition, h alfway through the survey, the respondent s were asked to report which kinds of
m-commerce activit ies they have been conducted within the p ast six months. This question
was constructed as a multiple choice question , were respondents could choose several
options. These specific options were extracted from a lar ge quantitate survey among Danish
consumers conducted by DIBS (2015) . Additionally, respondents had the opportunity to
report other options if their latest m-commerce activit ies were not listed. From this point
on, respondents were told to base their subsequent answers on their latest m -commerce
experience , provided that this experience happened within the past six months . Attaching
a specific m -commer ce experience should increase the likelihood that answers be coherent
and consistent. This method should also aid the respondents in retrieving the relevant
information needed , and also increase the chances of providing answers that are based on
the best m emorized experiences.

Chapter IV – Methodology
Page 59 of 155
4.2.2. Study Two

The aim of study two was to pick up, where study one ‘left off ’. Thus, study two was
basically an extension of some of the aspects of the first study, why the conceptual model
builds upon the same premises and thus have an equivalent instrumental approach. Study
two concentrates on the proposed mediation effect between Flow (FL) and Conti nuance
Intention (CI) through Satisfaction (SAT) , why these variables are also present in the
conceptual model of this study . The purpose is to exhaust the information regarding this
phenomena in relation to moderation theory to determine, if personal traits have influence
on the proposed effects extracted from study one . Therefore, study two expands with two
additional variables: Self -Efficacy (SE) and Impulsiveness (IMP)

However, this study was based on another dataset that was not related to that of study
one. There fore, a new set of respondents was needed after finishing study one. The
instrumental approach of study two was, however, equivalent to that of study one . The
difference is the exclusion of Confirmation (CON), Perceived Ease of Use (PEOU) , Perceived
Usefulness (PU) Trust (TR) , and some of the initial questions regarding income, smartphone
frequency usage and m -commerce experience , since these variables were not relevant in
the context of this study. On the other end, the survey differs with the two additional
variables measured on a seven -point -likert -scale equivalent to the other variables. The
questionnaire was developed with the same theoretical evidence, though the questions are
fewer due to the deletion of variables. All variables from both studies are displayed in table
4.1.

Chapter IV – Methodology
Page 60 of 155
Constructs Measurement
items Sources
Confirmation 3 Bhattacherjee (2001)
Perceived Ease of Use 4 Thong et al. (2006) , adopted from Davis et al. (1989)
Perceived Usefulness 4 Thong et al. (2006) , adopted from Davis et al. (1989)
Trust 4 Zhou (2013a) , adopted from Gefen et al. (2003)
Flow 3 Zhou (2014b) , adopted from Lee et al. (2007 )
Satisfaction 3 Lee (2014) , adopted from Spreng and Olshavsky (1993)
Continuance Intention 3 Thong et al. (2006) , ado pted from Bhattacherjee (2001)
Self-Efficacy 3 Yang (2010)
Impulsiveness 3 Rook and Fisher (1995)
TABLE 4.1. ORIGINS OF MEASUREMENT ITEMS

4.3. Data Collection P rocedure

4.3.1. Pil ot Test

Before publishing the final questionnaire to the general population, a pilot test was
conducted to accommodate potential errors (Martin and Polivka, 1995) . Especially since the
items de fining the constructs originate from external literature in English, a contrario
Danish, in which the questionnaires in this paper was published in. It was thus of utmost
importance to ensure accurate presentation of the items in the context, in which they were
designed. The pilot test was conducted among 20 individuals, half of which had extensive
experience with m -commerce and half of which were rather ine xperienced within the field.
This way, the study only used respondents qualified to participate in the final
questionnaires, and further used experience as a measure to take into account some of the
proposed important divergences among the entire populatio n, which was important for two
reasons: Firstly because the respondents in the pilot test should, to the highest extent
possible, be similar to the respondents responding to the final questionnaire (Conrad and
Schober, 2000 , cited in Malhotra and Birks, 2011 ) and secondly, since the feedback from
the pilot test should comprise all elements of the presentation. And given the
comprehensive variety of elements, it seemed appropriate to include as many people of
different backgrounds as possible.

Chapter IV – Methodology
Page 61 of 155
Using the feedback, several changes were made. The most severe issue was the
translation of the items from English into Danish. Si nce some of the variations of different
words and formulations don’t exist in the same extent in Danish, some questions within the
same construct seemed rather indifferent. As for instance in the construct of Satisfaction
(SAT), several respondents reveale d confusion about the differences between the
questions, which ultimately could lead to issues within the factor, since absence of diversity
between the items would most likely cause the respondents to rank equally on the items
unheeded of the actual diffe rence. To rectify this problem, the three items scale from Lee
(2014 ) was adopted instead, as mentioned. Lastly, the respondents further argued that
there were insufficient options for educational level as well as occupations, which were also
revised.

4.3.2 . Study O ne

To increase population validity , two different non -probability sampling techniques were
deployed: convenience sampling and snowball sampling. Data were collected in a social
media environment , where respondents were encouraged to share the survey in their
respective networks (n = 148) . This offered an increasing probability of randomization , since
exposure increased drastically for every time a person shared the survey link. Additional
data were collected at different bus stops in Aarhus , were respondents were asked if they
would fill out a questionnaire , while they waited for their transport ation (n = 39) . To address
ecological validity of the offline survey , two independent t-tests were conducted ; one for
each dependent variable ( DV) in the framework . The point of the tests was to find out
whether or not there were significant differences in responses between respondents
answering online, and respondents answering the questionnaire face to face with the
authors. However, these tests did no t show any significant differences betwe en the two
groups3 (p > .05) .

3 Appendix 2

Chapter IV – Methodology
Page 62 of 155
To decrease non -response bias , respondents were informed that each completed survey
would trigger a 2 DKK donation to a charity organization. At the end of the questionnaire,
the respondents had the opportunity to vote for the organizati on that he or she though
should have the accumulated gathered amount.

Data were collected over a three -week timespan . By the cut -off date , a total of 523
questionnaires were di stributed , whereas 204 were completed and returned , giving a
response rate on 39 percent . Based on Mahalanobis ’ distance (< df = 6 ; p = .05 = 12.592)
and Cook's distance (< 1) , 17 observati ons were identified as outliers (Cook and Weisberg,
1982; Field, 2009) , why they were removed from the mix . Thus, 187 valid responses from
Danish m -commerce users were acquired .

4.3.3 . Study T wo

Since the independent t-tests of study one illustrated response homogeneity between
the respondents from the social media and from the street , it seemed acceptable to simply
focus on obtaining responses from one group, why all respondents were obtained
conveniently through social medias , using the snowball sampling technique the same way
as in study one .

As in study one, an extra incentive was given to answer the questionnaire with the
promise of a 2 DKK donation to charity for each usable completed questionnaire. The survey
was active for seven days and resulted in 253 distributions and 138 completed
questionna ires for data analysis , giving a response rate on 55 percent . However, based on
the computations of Mahalanobis ’ distance ( < df = 4 ; p = .05 = 9.488) and Cook’s distance
(< 1), 13 observations were considered outliers, why the final number of useable
obse rvations resulted to 125.

Chapter V – Data Analysis & Results
Page 63 of 155

– Chapter V –
DATA ANALYSIS & RESULTS

“Customers can’t always tell you what they want, but they can always tell you what’s
wrong.” – Carly Fiona

Chapter V – Data Analysis & Results
Page 64 of 155
The following chapter consists of four parts. Initially, an assessment of the statistical
techniques and the software used , followed by an initial analysis of the sample
characteristics. Third, a thorough assessment of necessary assumptions to figure out, if the
data is in fact valid for further analyses. Finally, the hypothesi s results are examined.

5.1. Data Analysis I nstruments

The statistical software SPSS version 21.0 was used to conduct all statistical analyses in
this study. OLS (Ordinal Least Square) regression was used to calculate model coefficients.
The SPSS macros INDIRECT and MEDIATE ( Preacher and Hayes, 2008; 2014) incorporated
with the three -step approach suggested by Baron and Kenny (1986) , were used to examine
the existence of mediation in the data. The macro MODMED , along with the method of
Preacher et al. (2007) of analyzing interactions and conditional indirect effects , were used
to test the moderated mediation al effects . For testing the significance of these indirect
effects , as well as conditioned indirect effects, the recommended bootstrap bias -corrected
method was utilized (Preacher and Hayes, 2008, 2004) . The following section seeks to
elaborate and support the methodologically approach taken , as well as the statistical
techniques used to test and analyze the proposed research models

OLS regression is a powerful method for revealing relationships between variables and
the magnitude hereof . It follows the mathematical technique termed method of least
squares (Field, 2009) . The technique basically seek s to fit a regression line between the
observed data points that is based on the smallest possible distance from all the dat a points
to the regression line (residuals). Thus, the l east square method calculates the intercept and
slope of the line with the lowest possible sum of squared differences (Field, 2009) . With a
representative regression line , these regression models can fairly accurate ly predict the
value of the DV at a giving value of an independent variable ( IV), by estimating the strength
of the effect and how much of the variance in the DV is accounted for by the IV (Field, 2009) .
Multiple regression is a natural extension to OLS regression , since it allows for the
simultaneous inclusion of several IV s, which provides additional information , such as which

Chapter V – Data Analysis & Results
Page 65 of 155
of the IVs is the best predictor of the DV. This method have also been widely utilized in
several disciplines, since it all ows for a comprehensive testing of the relative importance of
several observable as well as unobservable latent variables in explaining a given outcome.
E.g. the study of Stefl -Mabry (2003) used multiple regression to understand the importance
of different information sources (WOM, newspapers, internet etc.) in explaining the degree
of consumers ’ informati on satisfaction. However, though multiple regression provides
several useful pieces of information about the relationship between variables, it does not,
by itself , yield a deeper understanding of the underlying mechanism s and process es in
which these effects occur (Tabachnick and Fidell, 2007) . As this was of great interest in this
thesis , the models presented were therefore hypothesiz ed as causal models , meaning that
the IVs (X) were assumed to be causing the mediators (M) that in turn caused the DV (Y).
This is basically what, in literature terms , is defined as a mediation model (Rose et al., 2004) .
Mediation analysis essentially extends the comprehensiveness of multiple regression , and
utilize s its properties as a way of means in gaining the unique opportunity of understanding
the underlying nature of the relationships that may exist between variables.

In order to test the proposed mediation models, this study adopted, to a certain degree,
the methodological approach f or evaluating path coefficients suggested by Baron and
Kenny (1986) . This approach is also known as the ca usal-step approach that inv olves four
steps, and is one of the most widely used method s for testing the existence of mediation
(MacKinnon et al., 2002; Rose et al., 2004) . The first step is to prove a significant relationship
between the IV (X) and the DV (Y) (c-path ) through simple regression . The next step is to
show a significant relationship between the mediation variable M an d X (a-path ), whereas
in the third step, the rel ations hip between M and Y is investigated (b-path ). This third step ,
however, requires a multiple regression of both X and M onto Y, since Y and M may be
correlated , since they are both presumed to be impacted by X . Finally, to establish the
meditational effect of M in the relationship between X and Y , the path between X and Y
when controlling for M ( c’-path ) is exanimated. If the c’ -path resolves to zero , or becomes
insignificant, M is suggested to be fully mediating the effect s between X and Y. If c’ ,

Chapter V – Data Analysis & Results
Page 66 of 155
however , is significant ly smaller after introducing M, but still greater than zero, a partial
mediation of M is evident (Baron and Kenny, 1986; Kenny et al., 1998) .

There seems to be an on -going debate in litera ture regarding the necessity of
performing all four steps, since step two and three essentially covers step one, as they
reveal the overall effect and therefore also the c -path. It is also argued that it is possible
that the overall effect woul d not be observed in the presence of suppressors or multiple
mediators that might cancel each other out (Wu and Zumbo, 2008) . Furthermore, step four
essentially diagnose the nature of mediation based on coefficients and s ignificance level .
This is , however , not sufficient for claiming that the indirect effect is significant (Baron and
Kenny, 1986; Frazier et al., 2004; Kenny et al., 1998) . Thus, literature offer several
alternatives for testing the indirect effect in a statistical significant way . E.g. MacKinnon et
al. (2002) compare 14 methods for testing the indirect effect , which t hey further categorise
within three main categories allocated according to method of performing the statistics : (1)
Causal steps, (2) differences in coefficients and (3) product of coefficients method.
According to Wu and Zumbo (2008) , the most commonly applied method is the Sobel ’s test,
which is in the coefficient family . This is , however, not the most recommended approach.
Especially not in situation s where sample sizes a re small to medium (Preacher and Hayes,
2004) . The main criticism surrounding the Sobel ’s test is that it directly tests the significance
of ab-paths against a norm al Z-distribution (Wu and Zumbo, 2008) . However, as also argued
by MacKinnon et al. (2004) , the product of ab rarely follows a normal distribution, which is
suggestively also the reason for its low statistical power (MacKinnon et al., 2002) .

Literature seems to support the use of resembling methods when testing mediation in
smaller sample sizes (< 500) (Preacher and Hayes, 2004; Shrout and Bolger, 2002; Wu and
Zumbo, 2008) . The study of Fritz and MacKinnon, (2007) compared seven methods , and
found the bias -corrected bootstrap method to be the one that required the smallest sample
size to achieve a statistical power of .8. Additional evidence is found in MacKinnon et al.
(2004) , since they found that the bias -corrected bootstrap method produced more accurate
confidence intervals compar ed to methods that assume a normal distribution in the product

Chapter V – Data Analysis & Results
Page 67 of 155
of two variables, even after the distribution is adjusted, and also when compared to
percentile bootstrap, bootstrap -t, bootstrap -Q and the Monte Carlo method. In light of such
clear evidence , and when considering the sample sizes in this study, the bias -corrected
bootstrapping method was therefore concluded to be the most appropriate method for
testing the significance of indirect effects under the conditions of this thesis .

The bootstrap bias-corrected method essentially takes a large number of samples from
the original sample size ( n), and adds replacements to that derived sample ( ny). The indirect
effect of ab is then computed for the new sample with replacements ( ny). This process is
run multipl e times – up to 10.000 times – each time with a unique sample size. The mean of
the calculated indirect effect is then extracted across the samples , which further provides
an empirical sample distribution , in which the standard deviation of the mean can be
estimated (Preacher and Hayes, 2004) . However, this distribution is often asymmetrical
(MacKinnon et al., 2004) . The bias -corrected bootstrapping , however, unlike other
bootstrapping method s, corrects for skewness in the distribution, meaning that the
confidence interval is more likely to be centred around the true parameter value (Fritz and
MacKinnon, 2007) . Thus, the conclusion whether or not the indirect effect is significant ,
relies on the presence of zero in the confidence interval (Preacher and Hayes, 2004) .

Concerning hypothese s H 3a and H 3b it was suspected that the mediated effect was
depend ing linearly upon two moderating variables (W). For this purpose , the SPSS
MODMED macro developed by Preacher and colleagues , together with the
methodologically approach for addressing conditional indirect effects published in Preacher
et al. (2007) , were implemented. Variables were centred before entering the model in order
to reduce multicollinearity (Preacher et al., 2007) , while the significance of the conditioned
indirect was tested through the bias -corrected bootstrap approach . Recent behavioural
research see m to be favouring this method (Hmielowski et al., 2014) , since the MODMED
macro efficiently allows research to probe whether or not the mediated effect is consistent
and significant across different level s of the moderator. The MODMED macro esse ntially
produces coefficient of the indirect effect when the moderat or is set to the sample mean

Chapter V – Data Analysis & Results
Page 68 of 155
and +/ -1 standard deviation. The conditioned indirect effect was , however, only tested after
observing an interaction effect (W x M) on Y , since this is a crit ical assumption for claiming
the existence of moderated mediation ( Preacher et al. 2007) .

5.2. Sample Characteristics

The obtained sample from study one showed a relatively even distribution of males
(45%) and females (55%). The largest group represented in the sample, was between by 21 –
30 (75%) and respondents were most likely either student s (48%) or tenured (41%). Most
respondents (79%) reported having, at a minimum, completed a high school diploma and
having between one and four years of m -commerce experience (78%). The numbers did
also indicate, respondents were relatively frequent smartphone users, since 89%, at the
time of the survey, reported to be using their smartphone on a daily basis. 58% reported to
be using their smartphone on average between one and three hours per day, while 32%
used their smartphone more than three hours per day on average. Demo graphic from study
two showed similar results with an almost even participation of females (53%) and males
(47%) while the majority were between 21 -30 years old (71%). Few had an educational
degree below high school (19%) and most were either student (36%) or tenured (45%). The
demographic of the resp ondents are presented in table 5.1 .

Chapter V – Data Analysis & Results
Page 69 of 155
Study one (n = 187) Study two (n = 125)
Variable Option Count Percentage Count Percentage
Gender Male 84 44.92 % 59 47.20 %
Female 103 55.08 % 66 52.80 %
Age 18 – 20 14 7.49 % 7 5.60 %
21-30 141 75.40 % 89 71.20 %
31-40 17 9.09 % 14 11.20 %
41-50 6 3.21 % 6 4.80 %
> 50 9 4.81 % 9 7.20 %
Highest completed
education Primary school 9 4.81 % 11 8.80 %
Craftsman 16 8.56 % 12 9.60 %
High school 47 25.13 % 25 20.00 %
Professions bachelor 28 14.97 % 29 23.20 %
Bachelor 48 25.67 % 23 18.40 %
Master 25 13.37 % 14 11.20 %
Phd. 0 0.00 % 0 0.00 %
Other 14 7.49 % 11 8.80 %
Primary occupation Unemployed 7 3.74 % 4 3.20 %
Part time job 4 2.14 % 9 7.20 %
Tenure 77 41.18 % 56 44.80 %
Student 89 47.59 % 45 36.00 %
Self-employed 4 2.14 % 3 2.40 %
Other 6 3.21 % 8 6.40 %
Smartphone usage (avg.
hours per day) < 1 20 10.70 %
1 – 3 108 57.75 %
3 – 5 44 23.53 %
5 – 7 12 6.42 %
> 7 3 1.60 %
M-Commerce experience
(years) < 1 13 6.95 %
1 – 2 44 23.53 %
2 – 3 48 25.67 %
3 – 4 53 28.34 %
4 – 5 10 5.35 %
5 – 6 7 3.74 %
> 6 12 6.42 %
TABLE 5.1. DEMOGRAPHY ANALYSIS4

In relation to study one , there seems to be a general agreement in terms of the usability
of the smartphone to facilitate different kinds of shopping activities , since 99% of the
respondents were found to be involved with more than one shopping category.
Furthermore, it appears that the most t ypical activity involves shopping for physical
products (99%), software downloading (75%) and/or using mobile tickets services (77%).
The same three activities did also appear to be the most popular m -comme rce activity in
study two . Table 5.2 presents the m -commerce categories include in this study.

4 Appendix 4 and 5

Chapter V – Data Analysis & Results
Page 70 of 155

Study one Study two
Count Percentage Count Percentage
Shopping via M -Commerce
(Past six months) Subscriptions 43 22.99 % 56 44.80 %
Software download 143 76.47 % 70 56.00 %
Betting / casino 40 21.39 % 32 25.6 0 %
Physical products 185 98.93 % 77 61.60 %
Mobile ticketing 142 75.94 % 79 63.2 0 %
Mobile travel ordering 42 22.46 % 17 13.6 0 %
TABLE 5.2. SHARE OF M-COMMERCE ACTIVITIES

In summary, the typical m -commerce user in this study seems to be between 21 -30 years
old, well educated, has between one and four years of m -commerce experience , and uses
his/her smartphone on a daily basis. The m -commerce activities typically consist of
shopping for physical products , purchasing tickets and/or downloading paid software. The
majority with such characteristics , as represented in the sample , corresponds relatively well
to the average m -commerce user, since numbers from DIBS (2015) indicate that the most
active Danish m -commerce user is between 25 -44 years old , and is, according to Ipsos
(2013) , a frequent smartphone user. There is suggestively no difference related to gender
in ter ms of transaction frequencies. O nly the type s of transaction s conducted differ
between male s and females (DIBS, 2015) . Note, however, that since m -commerce in
Denmark is still in its infancy, the descriptive statistics generated, that are available to the
general pub lic, are therefore based on more general tendencies that do not have explicit
focus on smartphones. However, when revie wing studies that are geographically dispersed ,
it seems that researchers generally agree that the common m -commerce user is well
educated and roughly around 21 -40 years old (Chong, 2013; Groß, 2015; Hung et al., 2012;
San‐Martin and L ópez‐Catal án, 2013) . Additionally, due to the relative ly low response rate
experienced in the online survey in study one , a test for non -response biases was conducted
by comparing early respondents to late respondents. It is argued that late respondents will
have similar characteristic as non -respondents (Rogelberg et al., 2003) . T-statistic5 showed
no significant different between the groups ( p > .05). Based on these indications it is
plausible that the samples obtained reflect the average Danish m -commerce users ’
perceptions of m -commerce services.

5 Appendix 3

Chapter V – Data Analysis & Results
Page 71 of 155
The sample size necessary to generate sufficient statistical power , is difficult to estimate ,
because under the current conditions, the exact population is practically impossible to
obtain , why it is also impossible to assess sampling errors. One way to estimate population
size could be to base it on the number of smartphone owners. T his is , however, as according
to Lin and Wang (2006) , just as wrong as basing the population of e -commerce users on the
number of computer owners . However, according to Tabachnick and Fidell (2007) , the rule
of thumb regarding sample sizes is that the number of respo ndents should preferably
exceed > 50 + 8m , where m is the number of IVs. Thus, in the sec ond part of the model , with
6 IVs, the sample size should preferably be equal to, or exceed, 98 respondents, which was
also the case.

5.3. Regression A ssumptions

According to Osborne and Waters (2002) , ensuring that the postulated regression model
is not violating some predefined assumptions is of high importance, since violation can
eventually cause unreliably and misguiding results , such as an over – or underestimation of
causal effects (Field, 2013; Osborne and Waters, 2002; Tabachnick and Fidell, 2007) .
However, since not all assumptions suggested by the literature are equally important
(Osborne and Waters, 2002) , and that complying w ith some assumptions makes other
assumptions redundant, the following section will only briefly analyze assumptions that are
highly problematic to violate within multiple linear regression. Further, a s both studies rely
on the same set of assumptions, these are analysed simultaneously.

5.3.1. Reliability

To measure the internal consistency within each construct, Cronbach’s Alpha is often
used as the indicator (Knapp, 1991) because of its abilities to provide knowledge about
inter -correlations of the items of each construct. That is, measuring whether or not there is
compliance between the items within each constructs , and that the items thus explain the
same thing , though w ith some degree of uniqueness (Streiner, 2003) . A rule of thumb is

Chapter V – Data Analysis & Results
Page 72 of 155
that Cronbach’ s Alpha must be at least .7 to indicate good internal consistency (Lance,
2006) .

In study one , all variables except for flow – that has a Cronbach’ s Alpha at .65 – qualify
the . 7 lower boundary and do generally represent h igh alphas . However, t hough it is
generally considered beneficial to achieve as high alphas as possible, Streiner (2003)
stresses that an a lpha higher than .95 indicates a lack of uniqueness, meaning that one or
more items may be entirely redundant. None of the construc ts within study one fall under
that assumption. Nor do the variables in study two , though continuance i ntention fails to
comply with the lower limit ( α = .59), which may be due to the low communalities (.47) of
CI2, in conjunction with the low numbers of items to offset this sufficiency.

The method has, however, been subject to a lo t of criticism. For instance, Cronbach’s
Alpha does not only take into account the magnitude of correlations between the items,
but also the number of items in an aggregated approach . Cronbach’s Alpha can thus be
increased simply by increasing the number of items (Streiner, 2003) , regardless of their
loadings on the factor in question . Thus, it seems that for using Cronbach’ s Alpha, another
set of a ssumptions should be adhered for it to be reliable.

To accommodate s ome of these issues, the Composite Reliability (CR) score is used as a
more accurate indicator of reliability . CR takes into account the different loadings, thereby
acknowledging the hete rogeneity between the items. Thus, CR is aware of different loadings
and is not subject to the same underestimated bias , as what is the case for Cronbach’s Alpha
(Fornell and Larcker, 1981; Raykov, 1997) . The lower limit for CR is also .7 (Fornell and
Larck er, 1981) , and i s met by all constructs of both study one6 and two7, why the results of
Cronbach’s Alpha are considered less relevant . It is there fore the assumption that the
constructs of both studies are reliable.

6 Appendix 6
7 Appendix 7

Chapter V – Data Analysis & Results
Page 73 of 155
5.3.2. Validity

To measure the overall validity, convergent and discriminant validity have been
analysed. Convergent validity refers to the fact that the measurement items that are
supposed to measure the same thing, in fact do measure the same thing. On the other hand,
the purpose of discriminant validity is to display that measures that are not supposed to be
related , are in fact no t related.

To qualify for convergent validity, Chin et al. (1997) suggest t hat if all factor loadings are
.60 or higher , the construct qualifies for convergent validity . The lowest loading in both
study one and study two is interestingly FL1 with a loading of .63, thus satisfying convergent
validity for both study one and study two . Another assessment is by calculating the Average
Variance Extracted ( AVE) for each construct. An AVE on at least . 5 equally qualifie s for
convergent validity, as an AVE on more than .5 indicates that the items of the construct
explains more than 50% of the variance in the variable (Fornell and Larcker, 1981) . All AVEs
in both studies qualifies the .5 limit , thus qualifying for convergent validity in this method
as well.

To assess the discriminant validity, Fornell and Larcker (1981) and Chin et al. (1997)
agree on the method of comparing the shar ed variances with the respective √AVEs of each
construct , referring to their average factor loadings . Table 5.3 displays the individual √AVEs
for each construct compared with their inter -construct correlations. The table confirms
discriminant validity by showing that each factor’s √AVE is higher than their correlated
counterparty. For all variables in both studies , discriminant validity was satisfied .

Chapter V – Data Analysis & Results
Page 74 of 155
Study one (n = 187)

Factor CON PEOU PU TR FL SAT CI
CON .86
PEOU .45 .91
PU .56 .57 .81
TR .57 .59 .57 .86
FL .60 .49 .47 .63 .78
SAT .68 .63 .66 .74 .66 .95
CI .58 .60 .66 .64 .64 .75 .86
α1 .83 .93 .81 .89 .65 .94 .80
CR2 .90 .95 .89 .92 .82 .96 .89
AVE3 .75 .83 .66 .75 .61 .89 .74

Study two (n = 125)

Factor FL SAT CI SE IMP
FL .90
SAT .46 .94
CI .42 .57 .88
SE .37 .60 27 .90
IMP -.02 -.20 .04 -.15 .92
α1 .71 .85 .59 .75 .79
CR2 .85 .91 .81 .86 .88
AVE3 .66 .78 .60 .66 .71
Diagonal elements represent the square root of average variance extracted (√AVE). Off -diagonal
elements represent the shared variances ( Pearson’s inter -construct correlations)
1α = Cronbach’s Alpha
2CR = Composite Reliability
3AVE = Average Variance Extracted
TABLE 5.3. RELIABILITY AND VALIDITY ANALYSIS

5.3.3. Normal D istributions of Residuals

To reduce the chances of distorting the final results, it is of importance that the residuals
of the DVs are approximately normally distributed (Osborne and Waters, 2002) . According
to Tabachnick and Fidell (2007) , a v iable assessment of normality is through a visual
inspection of a histogram, QQ -plot and by calculating Skewness and Kurtosis (Osborne and
Waters, 2002; Tabachnick and Fidell, 2007)

First, analysing the skewness and kurtosis , it is important that both have values w ithin
the absolute value of 1 (Adams et al., 2007; Pallant, 2007) . Assessing the DVs for the two
models of study one , both the construct of satisfaction and that for continuance i ntention ,

Chapter V – Data Analysis & Results
Page 75 of 155
the assumption was me t with respective values at ( S: -.254 ; K: .126 ) and (S: -.035 ; K:
.111)8+9. Note, that data cleaning was already performed at this point . The assumptions was
equally qualified for study two at the values of (S: -.138 ; K: -.352)10 for continuance
intention .

Furthermore, looking at the QQ -plot, the residuals are subjectively considered to be
closely aligned around the line for all DVs of both studies , which also indicat es normally
distributed residuals. Lastly, the DVs of both studies were analysed visually using the
histogram, which, if the residuals are normally distributed, should follow the normal
distribution curve. Both studies met this assumptions too, though there are small deviations
indicating a few outliers. These are, howev er, not considered severe , why it is fair to say
that both studies meet the normality assumptions under these circumstances too .

5.3.4. Homoscedasticity

The model s further assume homoscedasticity, meaning that the variance of residuals
needs to be simila r at different levels of the predictors (Osborne and Waters, 2002) . For
study one , assessing the first part of the model, having satisfaction as the DV, indicated
slight deviation from this assumption , which was confirmed by conducting a Breusch -Pagan
test (p-value = .04) , thereby proving heteroscedasticity (Breusch and Pagan, 1979) . In worst
case scenario, this could imply that the coefficients associated with the predictors could be
inaccurate. The K oenker test, howe ver, showed the opposite result and thereby insinuated
homoscedasticity ( p-value = .60 ). Following the adjusted error technique through the
homoscedasticity -consistent standard error (HCSE) approach (robustness test) in Hayes and
Cai (2007) , it revealed that the p-values of the independent predictors in this model did not
change significantly after adjusting their standard errors to be homoscedasticity
consistent11, thus only a mild degree of heteroscedasticity was present. However, Osborne

8 Appendix 8
9 Appendix 9
10 Appendix 10
11 Appendix 12

Chapter V – Data Analysis & Results
Page 76 of 155
and Waters (2002) argue that small degrees of heteroscedasticity only have small effects on
significance tests. E.g. Hayes and Cai (2007) analysed the implications for heteroscedasticity
and found that though a lack of homoscedasticity cau sed some problems in hypothese s
testing , far from all tests were impaired by it.

When conducting the analysis using continuance i ntention as the DV in the second part
of the model , both the Koenker – (p-value = .00) and Breusch -Pagan (p-value = .00) test
showed heteroscedasticity at a higher degree12. Using the robustness test it displayed that
the direct effects of perceived ease of u se went from significant to insignificant . The
remaining constructs did not significantly deviate from their initial values . These results will
be taking into consideration when conducting the multiple regression.

In study two , both the Breusch -Pagan test and Koenker test were conducted. Both tests
proclaim homoscedasticity by rejecting the null hypotheses at p-values = .14 and .29
respectively.

5.3.5. Independence of E rrors

The final assumption is that the errors must be independent from one another (Field,
2013; Osborne and Waters, 2002) . Most cases of non -independence of errors are, however,
caused in situation of high variability. For instance in qualitative studies , where the
interviewer’s experience with the questionnaire and subject in general evolves over time,
why the initial respondents might exhibit more variability in their responses (Tabachnick
and Fidell, 2007) . In both studies in this paper , all responses are obtained using the same
respective questionnaire s with no significant varying approaches and experiences. Th us,
independence of errors is initially assumed.

To test the independence of errors, a Durbin -Watson test was conducted. The Durbin –
Watson test tests whether or not adjacent residuals are correlated , and varies from 0 -4,

12 Appendix 11

Chapter V – Data Analysis & Results
Page 77 of 155
where values close to 2 indicate uncorrelated residuals , hence independent of errors (Field,
2013) . Thus, models with values exceeding 3 or is less than 1 are more likely to caus e
declination of the final regression. In study one, the Durbin -Watson tests indicated
inde pendence of errors at the (2.03) and (2.05 ) for satisfaction and continuance i ntention
as DVs respectively13. In study two , assessing continuance i ntention, a Durbin -Watson score
at (1.771) was observed14.

5.4. Model Fit

Prior to running the regression analyses , the following section will seek to determine,
how well the models fit the observed data. The goodness -of-fit approach is commonly
applied, using the R2 to gauge the magnitude of the relationships (Field, 2009) .

5.4.1. Study O ne

The R2 (.71) for satisfaction was found to be significant different from zero, F(5, 181) =
90.325, p =.00. Altogether, 71% of the variability of satisfaction was accounted for by the
five IVs. The R2 (.66) for continuance i ntention was also significant, F(6, 180) = 57.401 p =
.00. This indicate d that 66% of the variability of continuance i ntention could be explained
by the six IVs. Because proving the existence of mediation effects in the proposed research
model required the application of several single and multiple r egressions , that naturally
spawn different R2 depending on t he model used, a n overview of the R2 values , together
with Anova results generated by each applied model , are therefore presented in table 5.1.
Based on these results, the single regression models ( df = 1) applied indicated a medium to
low degree of model fit with R2 values ranging from a minimum of .20 to a maximum of .46.
Nevertheless, all models were found to be significant bette r predictor of their respective
outcome variable when compared to models using the means as “ best guess ” (Field, 2009;

13 Appendix 13
14 Appendix 14

Chapter V – Data Analysis & Results
Page 78 of 155
Tabachnick and Fidell, 2007) . It is therefore concluded to be reasonable to continue the
analysis .

5.4.2. Study Two

Initially, for s atisfaction the R2 (.21) is significantly different from zero F(1, 123) = 36.964,
p = .00, thus flow accounts for 21% of the overall variability in the average users’
satisfaction. Looking further at the continuance intention, the model includes both the
proposed moderating variables of self -efficacy and impulsiveness, and their proposed
moderating ef fect in two different m odels. Thus, when implementing s elf-efficacy and the
hypothes ized moderation interaction on satisfaction along with f low, the variability of
continuance intention is R2 (.37) and is significant at F(4, 120) = 17.461, p = .00. The sam e
test was conducted, replacing the effects of self-efficacy with those of i mpulsiveness,
resulting i n a significant variability of continuance intention of 40%, R2 (.40), F(4, 120) =
19.365, p = .00 as presented in table 5.4.

Chapter V – Data Analysis & Results
Page 79 of 155

Study one

Variables R2 df F-ratio P-value
Perceived Ease of Use1 .20 1 46.029 .00
Perceived Usefulness1 .31 1 83.191 .00
Trust1 .33 1 90.748 .00
Flow1 .36 1 103.524 .00
Satisfaction1 .46 1 159.794 .00
Satisfaction2 .71 5 90.324 .00
Satisfaction3 .69 4 98.926 .00
Continuance intention4 .62 4 73.594 .00
Continuance intention5 .66 5 69.252 .00

Study two

Variables R2 df F-ratio P-value
Satisfaction6 .21 1 36.964 .00
Continuance intention7 .37 4 17.461 .00
Continuance intention8 .40 4 19.635 .00
1IV = CON
2IVs = CON, PEOU, PU, TR, FL
3IVs = PEOU, PU, TR, FL
4IVs = PEOU, PU, TR, FL
5IVs = PEOU, PU, TR, FL, SAT
6IVs = FL
7IVs = FL, SAT, SE , INT(X* M1)
8IVs = FL, SAT, IMP , INT(X* M2)
TABLE 5.4. MODEL FIT

5.5. Hypothesi s Results
5.5.1. Study O ne

The ana lysis of study one is two -fold. First, the objective is to explore the antecedents
of users’ satisfaction by investigating the first part of the conceptual model involving a
parallel mediation. Thus, the first part is to examine and segregate the d irect effects from
the indirect effects toward user satisfaction, distinguishing between prior experiences and
future expectations. Secondly, the study seeks to explain to what degree, and in which
nature, satisfaction and its underlying cognitive antecede nts can influence users’
continuance intention to use m -commerce. Thus, by conducting a single mediation test
between the cognitive variables of the model and continuance intention, using satisfaction
as the mediator, the study strives to clarify which ass umptions must be taken into

Chapter V – Data Analysis & Results
Page 80 of 155
considerations , when trying to adjust users’ behaviour within m -commerce usage. All
results are primarily reported with unstandardized coefficients.

5.5.1.1. Creating S atisfaction

Initially, a significant path was found between confirmation and satisfaction (B = .71, p
= .00). Also, significant paths were demonstrated between confirmation and perceived ease
of use (B = .38, p = .00), perceived u sefulness (B = .49, p = .00), Trust (B = .57, p = .00) and
flow (B = .56 , p = .00 ). Additionally, perceived e ase of use (B = .20, p = .00), perceived
usefulness (B = .23, p = .00), Trust (B = .31, p = .00) and flow (B = .18, p = .01) were positively
related to satisfaction . Given these results , the meditational roles of the cognitive variables
were tested using bootstrapping method with bias corrected confidence estimates
(MacKinnon et al., 2004; Preacher and Hayes, 2004) . In this study, the 95% confidence
interval (CI) estimating the indi rect effects were derived from 5,000 bootstrap resamples
(Preacher & Hayes 2008). Since zero was not present in any of the mediators’ estimated CI s,
the mediation analysis therefore confirms the accumulated mediating role of perceived
ease of u se, perceived u sefulness , trust and flow in the relationship between confirmation
and satisfaction (B = .47 ; CI = [.37 ; .57]). In light of these results, the expectations related
to hypothesis H1a, H 1b, H 1c and H 1d are therefore accepted. The CIs of the estim ated
mediation effects are shown in table 5.7 .

In addition, through a pairwise comparison of the mediators’ strength, no results were
found indicating a significant difference in regards to the estimated indirect effects, i.e.
none of the mediators showed a significant greater impact when compared to o ne another.
Furthermore, the results from the mediation analysis indicated that the direct effect of
confirmation on satisfaction decreased, but remained signific ant (B = .24, p = .00) when
controlling for perceived ease of u se, perceived u sefulness , trust and flow, thus indicating
partial mediation.

Chapter V – Data Analysis & Results
Page 81 of 155
Steps of parallel mediation B s.e. 95% CI β R2
Step 1
c-path
DV: Satisfaction
IV: Confirmation .71** .06 [.60 ; .82] .68** .46**
Step 2
a-path
DV: Perceived Ease of Use .38** .06 [.27 ; .50] .45** .20**
DV: Perceived Usefulness .49** .05 [.37 ; .60] .56** .31**
DV: Trust .57** .06 [.45 ; .70] .60** .33**
DV: Flow .56** .06 [.45 ; .67] .60** .36**
IV: Confirmation
Step 3
b-path
DV: Satisfaction .71**
M: Perceived Ease of Use .20** .07 [.07 ; .33] .16**
M: Perceived Usefulness .23** .06 [.11 ; .36] .20**
M: Trust .31** .06 [.18 ; .43] .29**
M: Flow .18** .06 [.06 ; .30] .16**
IV: Confirmation
Assessing
c'-path
DV: Satisfaction .71**
M: PEOU, PU, TR, FL
IV: Confirmation .24** .06 [.13 ; .36] .24**
** p < .01
* p < .05
CI = Confidence interval
TABLE 5.5. MEDIATION ANALYSIS OF STUDY ONE – PART I

5.5.1.2. Creating Continuance I ntention

In the simple mediation part of the model, the approach is slightly differ ent. As
established prior to this chapter , there were significant heteroscedasticity within this part
of the model. Therefore, an ordinary testing through OLS could in worst case lead to biased
and inconsistent results (Hayes and Cai, 2007) , why the HCSE approach seems more viable,
as it does not automatically assume homoscedasticity as OLS normally does. More
specifically, the HCS E approach calculates the standard errors more accurately according to
the lev el of h eteroscedasticity in the dataset, and does thus not calculate these in an
erroneous way (Hayes and Cai, 2007) .

Chapter V – Data Analysis & Results
Page 82 of 155
Significant impacts on c ontinuance intention were found from p erceived ease of u se (B
= .22, p = .02), perceived u sefulness (B = .41, p = .00), trust (B = .17, p = .04) and f low (B =
.34, p = 00). Furthermore, s atisfaction was found significant towards continuance i ntention
(B = .36, p = .00). Thus, since the four cognitive variables were found significant toward
satisfaction in the parallel mediation analysis, it justified the necessity for testing the
existence of mediational eff ect conceived by s atisfaction.

Steps of single mediation B s.e. 95% CI β R2
Step 1
c-path
DV: Continuance Intention .62**
IV: Perceived Ease of Use .22* .10 [.07 ; .38] .18*
IV: Per ceived Usefulness .41** .09 [.26 ; .55] .34**
IV: Trust .17* .08 [.02 ; .31] .15*
IV: Flow .34** .08 [.20 ; .48] .30**
Step 2
a-path
DV: Satisfaction .69**
IV: Perceived Ease of Use .20* .07 [.07 ; .33] .16*
IV: Perceived Usefulness .31** .06 [.18 ; .44] .26**
IV: Trust .35** .06 [.23 ; .48] .34**
IV: Flow .27** .06 [.15 ; .39] .24**
Step 3
b-path
DV: Continuance Intention .66**
M: Satisfaction .36** .10 [.20 ; .52] .35**
IV: Perceived Ease of Use
IV: Perceived Usefulness
IV: Trust
IV: Flow
Assessing
c’-path

DV: Continuance Intention .66**
M: Satisfaction
IV: Perceived Ease of Use .15ns .08 [.00 ; .30] .12ns
IV: Perceived Usefulness .29** .08 [.14 ; .44] .24**
IV: Trust .04ns .08 [-.11 ; .18] .03ns
IV: Flow .24** .07 [.10 ; .38] .21**
** p < .01
* p < .05
ns = not significant
CI = Confidence interval
TABLE 5.6. MEDIATION ANALYSIS OF STUDY ONE – PART II

Chapter V – Data Analysis & Results
Page 83 of 155
In the effort of decrypting the medi ational effects, support for H 2b and H 2d was
established , since the direct effects of perceived u sefulness (B = .29, p = .00) and flow (B =
.24, p = .00) were still significant after introducing satisfaction as a mediator, thus displaying
a partial mediation effect on ( B = .12; CI = [.0 5 ; .20 ]) and (B = .10 ; CI = [.04 ; .20 ]) respectively ,
using the bias corrected bootstrapping approach. The same test caused trust to become
insignificant ( B = .04 , p = .64 ), thus indicating full mediation with a significant mediational
effect on (B = .13; CI = [.06 ; .22 ]), thereby supporting H 2c. The same effect appeared when
testing perceived e ase of use (B = .15, p = .11 ), indicating full mediation with a significant
mediational effect on (B = .07; CI = [.02 ; .17]) , thus finally supporting H 2a. To illustrate the
difference in approach, the OLS approach illustrated partial mediation between perceived
ease of use and continuance intention through satisfaction, as the direct effect was still
significant after introducing s atisfaction to the construct. The mediated effects are
exhibited in table 5.7 .

Direct effects Mediated effects Total effects
CON SAT CI (c’) CI SAT1 BCBI CI2 BCBI SAT CI (c)
CON .24** .01ns .78**
PEOU .38** .20** .15ns .08* [.02 ; .15] .07* [.02 ; .17] .22*
PU .49** .23** .29** .11* [.05 ; .19] .12* [.05 ; .20] .41**
TR .57** .31** 04ns .18* [.09 ; .28] .13* [.06 ; .22] .17*
FL .56** .18** .24** .10* [.02 ; .19] .10* [.04 ; .20] .34**
SAT .36**
** p < .01
* p < .05
1 Mediated effects of CON on SAT through PEOU, PU, TR and FL
2 Mediated effects of PEOU, PU, TR, FL on CI through SAT
BCBI = Bia s Corrected Bootstrap Intervals
Bootstrap sample = 5,000
TABLE 5.7. MEDIATION EFFECTS ASSESSMENT – STUDY ONE

The results of the hypotheses are implemented in the conceptual model in figure 5.1.
The values noted above the lines ap ply to the total effect s between the respective variables.
The value s below the line s apply to the direct effect, when deducting the mediated effects
caused by the mediators specified in the model. Note that some paths represent both
significant as well as insignificant values, w hy no divergences have been made in design of

Chapter V – Data Analysis & Results
Page 84 of 155
these lines in accordance with significan ce level. Instead, significant levels are indicated by
(** < p = .01 ; * < p = .05). The final outcomes of hypotheses related to study one are
presented in table 5.10 after the analysis of study two.

** p < .01
* p < .05
FIGURE 5.1. THE CONCEPTUAL MODEL – RESULTS OF STUDY ONE

5.5.2. Study Two

The main objective of study two is to investigate the moderated mediation effect
between flow and continuance intention through s atisfaction , as exhibited in study one .
However, since study two is based on a new set of observations, the analysis will initially
demonstrate if the mediation effect proven in study one is in fact present in the new data
material , since that is an essential assumption for the further analysis. Another assumption
is that the b -path in the construct, being the direct effect between s atisfaction and
continuance intention, is also significantly moderated by the same factors proposed in the
moderated mediation part of the analysis. Therefore, prior to the analysis of moderated

Chapter V – Data Analysis & Results
Page 85 of 155
mediation, the study will first exhibit the possibility of moderation between user
satisfaction and continuance intention.

5.5.2.1. Moderating Role of Personal T raits

First, assessing the mediation analysis between flow and continuance intention through
satisfaction (figure 5.2) , the path between flow and continuance intention was significant
(B = .33, p = .00). Furthermore, significant paths were found between flow and satisfaction
(B = .38, p = .00), as well as satisfaction towards continuance intention (B = .46, p = .00).
Thus, when introducing satisfaction, the effect between flow and continuance intention
decreased significantly (B = .16, p = .02), leaving a significant mediational effect on (B = .17;
CI = [.10 ; .26]), thereby supporting the findings of study one.

FIGURE 5.2. THE CONCEPTUAL MODEL – PRELIMINARY RESULTS OF STUDY TWO

Next, a test was conducted regarding the hypothesized interaction of satisfaction and
self-efficacy as well as impulsiveness respectively. Multiple regressions were conducted
including the interaction variables of ( M x W ), to find whether these were in fact significant
for further analysis. The results showed a significant interaction in the case of impulsiveness

Chapter V – Data Analysis & Results
Page 86 of 155
(Bimp = -.10, p = .03) as opposed to the case of self -efficacy that turned out insignificant (B se
= .07, p = .33 ). Multiple regression results are presented in table 5.8.

Satisfaction Continuance Intention1 Continuance Intention2
Variables B s.e. p-value B s.e. p-value B s.e. p-value
Constant -1.96** .36 .00 4.95** .36 .00 5.03** .35 .00
Flow (X) .38** .07 .00 .16* .07 .02 .16* .07 .03
Satisfaction (M) .56** .10 .00 .52** .09 .00
Self-efficacy (W 1) -.12ns .08 .14
Int se (M x W 1) .07ns .07 .33
Impulsivenss (W 2) .08* .04 .04
Int imp (M x W 2) -.10* .04 .03
1 Self-Efficacy
2 Impulsiveness
TABLE 5.8. MULTIPLE REGRESSION RESULTS – STUDY TWO

The graphical illustration in figure 5.3 reveals that users’ impulsiveness interacts with
the positive relationship between satisfaction and continuance intention. That is, testing
the difference between users’ with high , medium and low degree of impulsiveness
respectively, where high and low being +/-1 standard deviation of 1.54 apart from the
medium level, represented by the mean of 3.46 on the seven -point -likert -scale. Specifically,
users with low degree of impulsive buying te ndencies show a greater continuance intention
when having a high degree of satisfaction, as opposed to users with a medium or high level
of impulsive tendencies according to the coefficient of the slope.

Chapter V – Data Analysis & Results
Page 87 of 155

FIGURE 5.3. MODERATING EFFECT OF IMPULSIVENESS (W2)

In the case of self-efficacy, the figure 5.4 confirms the initial assumption that the
relationship between satisfaction and continuance intention is stronger for users with a high
degree of self -efficacy, since the correlation eventually surpasses that of the users with a
low or medium degree of s atisfaction as illustrated in figure 5.3. However, though the
results correspond affirmatively with the hypothesis, the results are statistically insignificant
and must therefore be interpreted with caution .

Chapter V – Data Analysis & Results
Page 88 of 155

FIGURE 5.4. MODERATING EFFECT OF SELF-EFFICACY (W 1)

Finally, the findings of the moderated mediation analysis support hypothesis H 3d,
meaning that impulsiveness significantly moderates the mediated effect between flow and
continuance intention through satisfaction. This mediated effect on continuance intention
is thus stronger for individuals with low impulsiveness (B imp = .25, CI = [.15 ; .40]), compared
to those with medium (Bimp = .19, CI = [.12 ; .30]) and high impulsiveness (Bimp = .14, CI = [.07
; .23]) tend encies as illust rated in table 5.9 . So, provided that a user experiences flow, and
that the same user’s experience with m -commerce is generally satisfactory, the chances
that this particular user will use the system in the future increases significantly the less
impulsive s/he is. Meanwhile, H 3a is re jected, given the insignificant moderation results of
path b.

Degree of impulsiveness (W) Effect Bootstrap s.e. BCBI
Low -1SD .25* .06 [.15 ; .40]
Medium Mean .19* .05 [.12 ; .30]
High +1SD .14* .04 [.07 ; .23]
* Significant at 95% level
DV = Cont inuance Intention
Mean = 3.46
SD = 1.54
Bootstrap sample = 5,000
TABLE 5.9. CONDITIONAL INDIRECT EFFECT AT DIFFERENT LEVELS OF IMPULSIVENESS

Chapter V – Data Analysis & Results
Page 89 of 155
5.5.3 Summary of Hypotheses T esting

All hypotheses , except H 3a, are accepted, as shown in table 5.10 .

Hypotheses Relation Result
H1a Users’ extent of Perceived Ease of Use will mediate the
relationship between Confirmation and Satisfaction CON  PEOU  SAT Accepted
H1b Users’ extent of Perceived Usefulness will mediate the
relationship between Confirmation and Satisfaction CON  PU  SAT Accepted
H1c Users’ extend of Trust will mediate the relationship
between Confirmation and Satisfaction CON  TR  SAT Accepted
H1d Users’ extend of Flow will mediate the relationship
between Confirmation and Satisfaction CON  FL  SAT Accepted
H2a Users’ extent of Satisfaction will mediate the relationship
between Perceived Ease of Use and Continuance Intention PEOU  SAT  CI Accepted
H2b Users’ extent of Satisfaction will mediate the relationship
between Perceived Usefulness and Continuance Intention PU  SAT  CI Accepted
H2c Users’ extend of Satisfaction will mediate the relationship
between Trust and Continuance Intention TR  SAT  CI Accepted
H2d Users’ extent of Satisfaction will mediate the relationship
between Flow and Continuance Intention FL  SAT  CI Accepted
H3a The indirect effect between Flow and Continuance
Intention through Satisfaction will be moderated by the
degree of Self -Efficacy. FL  SAT  CI Rejected
H3b The indirect effect between Flow and Continuance
Intention through Satisfaction will be moderated by the
degree of Impulsiveness. FL  SAT  CI Accepted
TABLE 5.10. SUMMARY OF HYPOTHESIS TESTING

Chapter VI – Recapitulation
Page 90 of 155

– Chapter VI –
RECAPITULATION

“Don’t find customers for your products, find products for your customers.” – Seth Godin

Chapter VI – Recapitulation
Page 91 of 155
6.1. Discussion

This thesis aims to investigate constructs influencing users’ continuance intention
towards using m -commerce. However, the thesis attempts not only to identify, which
factors have influence, but also to what extent, and how, these factors constitute an imp act.
Finally, due to the fact that results drawn from multiple regression are based on an overall
average among the respondents, this thesis hypothesize that the presumed significant
differences between respondents’ personal traits , will ultimately affect the outcome of
their respective continuance intentions. In regard of these considerations, a conceptual
model was developed to investigate which constructs have direct influence on continuance
intention, and also how much of these effects – if any – are me diated by other constructs.
Therefore, the objective was to determine, which attributes m -vendors should allocate
resources to, in the effort of yielding the maximum effects of these proposed mediators.
Finally, the thesis tried to prove, through moderated mediation analyses, that specific
personal traits could have a significant impact on the effects drawn for the average
respondents.

6.1.1. Assessing R Q1

By what extent do perceived ease of use, perceived usefulness, trust and flow explain the
relationsh ip between user expectancy confirmation and user satisfaction?

According to the ECM, confirmation of prior expectations towards using information
systems, such as m -commerce, has a significant impact on the overall satisfaction with the
system. This pheno mena has thus also been proven in many studies, e.g. in the context of
information systems (Bhattacherjee, 2001) , e-commerce (Chang and Chou, 2011) and m –
commerce (Chong, 2013; Lee, 2014 ). The results in this thesis exhibit an equivalent
outcome, illustrating that confirmation explains 46 percent of the variance in user
satisfaction , and that the amount of this variance is explained by a linear coefficient at .71,
indicating high cohesiveness between the two. This means that the more a user’s positive

Chapter VI – Recapitulation
Page 92 of 155
expectations are confirmed following the use of a system, the more positive an attit ude,
the user will form, and thus also a greater satisfaction with the system. However, the further
developed ECM -IS posits that the same confirmation is also severely correlated with
cognitive beliefs towards the potential continuance usage of the system, and that these also
hold significant influence toward users’ satisfaction. Therefore, one of the main objectives
of study one was to investigate the possible mediational effects between confirmation and
satisfaction, thereby proclaiming that the cohesiven ess between the two, is in fact
dependent on underlying mechanisms to be fulfilled. Other studies have previously
acknowledged this possibility, why cognitive constructs have been implemented in various
conceptual models , trying to explain this phenomenon, though only few have actually
investigated the possible mediational effects caused by these. Thus, after scrutinizing
relevant literature, this study implemented four cognitive belief variables that are generally
acknowledged in this context: perceived ea se of use (Burton -Jones and Hubona, 2006;
Chong, 2013) , perceived usefulness (Bhattacherjee, 2001; Hong et al., 2006; Lin et al., 2014;
Thong et al., 2006) , trust (Alsajjan, 2014; Gefen et al., 2003a; Zhou, 2013a) and flow (Gao
and Bai, 2014; Zhou et al., 2010) . Under these circumstances, the model’s overall
explanation of the variance of satisfaction increased to 71 percent, indicating that the
model is now significantly more explanatory for users’ satisfaction. To investigate this
objective, however, it was necessary to assess the direct effects from confirmat ion to these
proposed mediators, furthermore the effects from the mediators towards satisfaction.

Perceived ease of use and perceived usefulness have empirically well -documented
effects on users’ attitudes toward the adoption of a technology. Therefore, these constructs
are generally considered natural extension to the ECM -IS, as satisfaction with a system can
arguably be compared to the positive attitude towards adoption of a system, since
satisfaction is a type of affect. Several studies have thus also found significant correlations
between perceived ease of use (e.g. Hong et al., 2006; Thong et al., 2006) , perceived
usefulness (Bhattacherje e, 2001; Chong, 2013; Lee, 2014 ; Lin et al., 2014; Zhou, 2014b) with
satisfaction in a continuance usage context. This study confirms these prior results, since
both PEOU (B = .20**) and PU (B = 31**) significantly influence satisfaction. Thus, according

Chapter VI – Recapitulation
Page 93 of 155
to the definition of PEOU adopted in this thesis, it seems that the easier the system is to
use, the more satisfactory the process will likely be for the user. In the case of PU, results
show that enhancing the performance of a desired action through m -commerce , compared
to other types of shopping (e.g. e -commerce, visiting physical stores etc.), m -vendors will a
paribus induce a higher degree of user satisfaction. Furthermore, this study also found that
confirmation of prior expectations significantly influence post -consumption beliefs,
including PEOU (B = .38**) and PU (B = .49**), which is also consistent with previous studies
(e.g. Chong et al., 2012; Thong et al., 2006) . Thus, based on whether or not the users’ initial
expectations of m -commerce are confirmed after the usage, this could likely alter their
future perceptions of the easiness and usefulness of m -commerce.

Trust is another well -researched construct that seems to be significantly important for
users’ satisfaction within many different contexts, as for instance within mobile apps (Akter
et al., 2013) , e-shopping (Chen and Chou, 2012) and m -commerce (Hung et al., 2012; Lin
and Wang, 2006) . Evidently, users trusting the companies, from which they purchase, or
trusting the systems they use, is of significant importance. This study provides empirical
support for these prior findings, since trust is found to significantly influence satisfaction (B
= .31 **), and is therefore no exception for the empirical learnings in the literature so far.
Thus, m -vendors acting in a trustworthy manne r that, in the definition adopted in this study,
includes integrity, predictability, ability and benevolence, enhance the likelihood of users
having a satisfactory experience severely. In fact, the influence of trust is the most
influential antecedent of u ser satisfaction, based on the evidence presented in this study.
Thus, m -vendors ignoring the fact that users value trustworthiness of a system, risk
impairing the overall satisfaction severely. Trust may therefore reasonably be subject to
developing a com petitive advantage, if one m -vendor manages to convince users that
his/her m -website is more reliable and trustworthy than that of his/her competitors. In
addition, the findings suggest that confirmation of prior experiences on the other end
influences tru st significantly (B = .57**), which is also in line with several other studies
(Akter et al., 2013; Chong et al., 2012; Hung et al., 2012; Weisberg et al., 2011) . Indeed,
trust, in this study, is the construct influenced mostly by confirmation, indicating that trust

Chapter VI – Recapitulation
Page 94 of 155
is a highly viable area to focus on for m -vendors, as impact goes both ways. That is, having
initial expectations disconfirmed, could severely diminish users’ perception of trust in future
commercial activities.

Flow is not nearly as heavily investigated as the other constructs of this s tudy, but the
aspects hereof have been investigated prior to the development of the constructs.
However, flow is a more complex construct, as it is a compilation of different aspects that,
in this study, are attention focus, perceived control and perceived enjoyment. According to
the theory, to experience flow, users must fully focus their attention on the interaction to
a degree, where any disturbing elements and irrelevant thoughts are omitted in the process.
Secondly, the design of the system must strike a balance between challenging the user
sufficiently, and simultaneously designing the systems to meet the skillset of the user. It
thus stems the risks of users being bored due to the triviality of the website, and on the
other hand, being frustrated due to its level of difficulty, thereby enhancing the chances of
users experiencing flow, which in turn leads to a more satisfactory experience. This
correlation is evident in the context of mobile sites (Hsu et al., 2012; Novak et al., 2000;
Zhou, 2014a, 2013b) , e-learning (Cheng, 2014) , mobile social networking services (Gao and
Bai, 2014) and finally m -commerce (Zhou, 2011) . Within m -commerce, this study affirms
the correlation, since the findings suggest that flow influence satisfaction significantly (B =
.27**). Having such cohesiveness, it is suggestively important to allocate resources to
induce flow experience among users, thereby enhancing the chances of user satisfaction.
Finally, as what is also the case for the other cognitive constructs, flow as a post -usage
expectation is significantly influenced by confirmation (B = .56 **), which is also in line with
other research, e.g. in the study of Cheng (2014) , though studies involving this effect are
rather limited.

So, returning to the proposed mediation effect of the implemented model extensions,
this hypothesized phenomena is now assessable, since all cognitive variables induce
significant effects on satisfaction, and that these in turn are influenced by confirmation.
Recall that confirmation was highly related to satisfaction (B = .71**). However, solely

Chapter VI – Recapitulation
Page 95 of 155
focusing on this relationship, which is the recommendation in the basic ECT, m -vendors
might e rroneously assume that users who have previously had good experiences with m –
commerce, will most likely generate satisfaction with the concept, without assessing
possible obstacles, which, according to the findings of this study, are evidently present.
Thus, when introducing the four additional cognitive variables as mediators between
confirmation and satisfaction, the findings were that of the effect of .71**, only .24** is
directly attributed the confirmation of prior expectations. The remaining effect of .47* are
thus dependent on the fulfillment of the future needs relating to the four cognitive
variables, perceived ease of use (.08*), perceived usefulness (.11*), trust (.18*) and flow
(.10*). Consequently, assuming that an m -vendor is unable to inflict sufficient trust for a
given user, this particular m -vendor cannot expect to yield the mediation effect of .18*,
why, if all other needs are fulfilled, s/he can only calculate with .53**. Therefore, it is the
opinion of this study that focusing on post -usage expectations are of utmost importance,
since failure to do so might cause marketers to act on wrong premises. Similar findings
regarding the mediation effects in this construct are present for perceived ease of use and
perceived usefulness (Jabri, 2015) , trust (Carlander et al., 2011) and flow (Cheng, 2014) .

6.1.2. Assessing RQ2

By what extent does user satisfaction explain the effects of perceived ease of use,
perceived usefulness, trust and flow on continuance intention?

Many studies investigating the mechanisms of ECM -IS have found evidence supporting
the highly significant relationship between user satisfaction and continuance intention (e.g.
Bhattacherjee, 2001; Chong, 2013; Hong et al., 2006; Kim, 2010; Lee, 2014; Thong et al.,
2006) , which is also the case in this study, as satisfaction is found to significantly influence
continuance intention (B = .36**). However, as hypothesized in this thesis, the effect of
satisfaction does not solely involve this obvious direct effect. This will be elaborated
momentarily.

Chapter VI – Recapitulation
Page 96 of 155
Returning to the implemented four cognitive belief variables, both perceived ease of use
and perceived usefulness represent not only significant correlations with user satisfaction,
but do evidently also correlate with users’ continuance intentions. Thus, results from the
multiple regression reveal that PEOU (B = .22*) and PU (B = .41** ) both attribute to the
willingn ess of users’ intention to conduct m -commerce activities in the future, if
expectations towards these needs are fulfilled. These results are consistent with those of
previous studies for PEOU (Chong, 201 3; Hong et al., 2006; Lee, 2014 ; Thong et al., 2006)
and PU (Bhattacherjee, 2001; Chong, 2013; Hung et al., 2012; Kim, 2010) respectively. It
thus seems that if prior experiences with m -commerce prompts users to anticipate that
future use of the concept will be sufficiently useful and easy, the more likely it is that the
user will continue using m -commerce.

Furthermore, assuming that prior experiences cause users to perceive a high level of
trust in future m -commerce operations, research has it that this perception can result in a
significantly greater likelihood that users will continue using m -commerce, which has been
stressed in the context of mobile payment systems (Zhou, 2013a) , e-commerce (Liu et al.,
2005; Ziaullah et al., 2014) , mobile services providers (Alsajjan, 2014) and m -commerce
(Chong, 2013; H ung et al., 2012; Wang et al., 2006) . In this study, trust was initially identified
to have significantly positive influence on users’ continuance intention (B = .17*), why the
explaining abilities of trusting beliefs are verified as a viable intrinsic extension of the ECM
to explain both satisfaction and continuance intention.

Finally, flow was investigated and found to be significantly correlated with continuance
intention (B = .34 **), which is an affirmative for similar conceptual layouts within mobile
payment systems (Zhou, 2013c) , e-learning systems (Cheng, 2014) , e-commerce (Hsu et al.,
2013) , mobile internet services (Zhou, 2014b, 2011) and website loyalty (O’Cass and
Carlson, 2010; Zhou, 2014a, 2013b) . Note, however, that results customized to m –
commerce are rather limited, though several studies are comparable. The findings thus
confirm that users experiencing flow, hence are in control of in the process, feel sufficiently

Chapter VI – Recapitulation
Page 97 of 155
challenged and enjoys the process of shopping via the smartphone , are more willing to
engage in m -commerce activities prospectively.

In summation, all these processed cognitive beliefs seem to significantly influence users’
willingness to continue using m -commerce to a higher degree, why it is somewhat fair to
say that m -vendors could advantageou sly allocate resources to promote these aspects of
their operations to achieve more user loyalty. That is the initial proposition of this study
along with several other studies. However, as hypothesized, satisfaction is believed to pose
a significantly lar ger impact on continuance intention apart from its direct correlation.
These beliefs were subsequently demonstrated through a mediation analysis, were
satisfaction was implemented as a perceived mediator between the four cognitive belief
variables and cont inuance intention. The outcome was that two variables, PU (B = .29**)
and flow (B = .24**) were proved to be partial ly mediated by satisfaction, meaning that 27%
(B = .11*) and 29% (B = .10*) of the effect of PU and flow respectively were directly
attribut able, and thus dependent, on the users’ degree of satisfaction. It therefore means
that for m -vendors to yield the maximum effects of PU and flow, they must ensure users’
satisfaction with the system. It is important, however, to note that even though a si gnificant
portion of these effects were mediated by satisfaction, large effects are still directly
attributable to PU and flow. For PEOU (B = .15ns) and trust (B = .04 ns), on the other hand,
the results revealed that both constructs were fully mediated by satisfaction. Consequently,
this means that assuming that a user is fully satisfied with the system, and simultaneously
assumes that the system is trustworthy and easy to use, stakeholders could likely rely on
the proposed effects mentioned earlier. Conve rsely, if a user is not satisfied, these effects
will not be continued towards continuance intention, as satisfaction will not mediate them
if not present. In essence, satisfaction is a very important construct within this matter.
Because even though the i mplemented four cognitive variables seemingly impact
continuance intention to a higher degree, a large portion of these combined effects will
vanish if m -vendors fail to satisfy the users as well.

Chapter VI – Recapitulation
Page 98 of 155
6.1.3. Assessing RQ3

Assuming a mediational effect between flow and continuance intention through user
satisfaction, by what extent is this effect dependent on users’ tendency to buy
impulsively?

The outcome of a multiple regression analysis is based on an overall average of the
sample, and is thus meant to exhibit an overarching trend for the entire population. Thus,
what the results show are the most probable correlations between constructs for the most
ordinary users of m -commerce. These results do present plausible guidelines as to how
users will react when exposed to different stimulus, and when certain needs are fulfilled.
However, when using regression results, it’s important that stakeholders treat these with
caution, and acknowledge that all users are somehow different. Particularly, users in each
end of the scale might diverge severely from the average user, and thus also the regression
results, why these users must be treated differently for the m -vendors to yield the same
results.

Within various context, researchers have investigated different traits of peoples’
personality to figure out, whether these could in fact lead people to perceive things
differently. Within a general shopping context, a heavily investigated personal trait is
peoples’ level of impulsiveness. Several studies (e.g. Arens and Rust, 2012; Chih et al., 2012;
Parboteeah et al., 2009) have stressed an importance difference between impulsive and
non-impulsive buyers, which is that impulsive buyers are far more affected by sudden
positive stimuli, causing sudden ur ges to buy, as opposed non -impulsive buyers, who are
usually more thorough in their buying process. The outcome of these different approaches,
however, seems to be that non -impulsive buyers are more likely to feel satisfied with their
purchases, as they ha ve, to a higher degree, more thoroughly considered alternatives and
consequences of the actions, whereas impulsive people in certain situations act in a more
inadvertent manner.

Chapter VI – Recapitulation
Page 99 of 155
Therefore, given that satisfaction is considered a key antecedent of users’ c ontinuance
intention, and that several studies (e.g. Hsu et al., 2012) have identified flow as a likely
antecedent of impulsive behaviour , the proposition of this study is that the more impulsive
users are, the less likely users are to be satisfied and thus to use the system in the future.

In both study one and study two, significant results presented itself in regards to the
proposed mediation effect between flow and continuance intention through satisfaction.
The aim of study two was th us to evaluate, if this mediation effect is in fact moderated by
users’ level of impulsiveness. So, first off, the proposed moderation effect of impulsiveness
was analyzed upon the correlation between satisfaction and continuance intention, as this
is a pr erequisite for testing the moderation of the mediation effect. What the results
exhibited were that there were significant ( p = .03*) different correlations for users with
low (.67*), medium (.54*) and high impulsiveness (.40*), thereby insinuating that pe ople
with low impulsive tendencies are far more likely to convert satisfactory experiences into
continuance usage intentions. Furthermore, evaluating the moderated mediation effect
between flow and continuance intention through satisfaction, the study show ed an
equivalent trend, accommodating the initial hypothesis that the mediated effect was
significant severely higher for less impulsive people (.25*) than for both medium (.19*) and
highly impulsive people (.14*). Thus, the less impulsive users are, the s tronger the
mediation effect caused by satisfaction between flow and continuance intention.

6.1.4. Assessing RQ4

Assuming a mediational effect between flow and con tinuance intention through user –
satisfaction, by what extent is this effect dependent on users’ degree of self -efficacy?
Regarding the moderating effect of self -efficacy, the initial theory was that individual
with higher confidence in their own abilities, would pose a moderation of the effect caused
by satisfaction upon continuance intention, as well the mediated effect caused by
satisfaction between flow and continuance intention. However, though the trend of
moderation between satisfaction and continuance intentions corresponded t o the stated

Chapter VI – Recapitulation
Page 100 of 155
hypothesis that people with a high degree of self -efficacy, would better transmit overall
satisfaction to continuance intention, these results were not significant and therefore not
sufficiently reliable. In addition, since this moderation eff ect is a prerequisite for testing the
moderated mediation, the proposed moderated mediation was not assessed. A possible
explanation might be that, as users gain more experience with m -commerce, as well as the
conditions become more user friendly , e.g. lar ger screens, better interfaces, quicker
payment features , the importance of one’s level of self -efficacy might diminish, as opposed
to earlier stages of m -commerce, where mobile webpages were more opaque, causing
more anxiety amongst users .

6.2. Implications

6.2.1. Managerial I mplications

M-commerce activities have rather rapidly increased during the past few years, and
though it is unlikely that these will exceed the amount of traditional commercial activities
any time soon, the concept of m -com merce are ceteris paribus acquiring more shares of the
market at the expense of alternative methods. Thus, m -vendors are therefore in direct
competition with vendors of more traditional manners, such as physical stores, e -commerce
etc., provided that a giv en vendor doesn’t operate flawlessly on every platform. So, what is
important from an m -commercial point of view is thus to figure out, which aspects of m –
commerce constitute a competitive advantage, vis -à-vis alternative shopping methods, for
shoppers to prefer shopping via their smartphones .

This study proposes four cognitive beliefs to contribute to this competitive advantage.
Thus, the results suggest that users’ perception of the easiness of using the system is of
significant importance. Therefore, an apparently important aspect of the shopping
experience, is that maneuvering m -websites and browsing for products are easy measures.
So, to avoid users being frustrated with the process of shopping through their smartphones ,
an initial step for m -vendors c ould be to transform their original webpage into a more
appropriate size to fit popular phones.

Chapter VI – Recapitulation
Page 101 of 155
In addition, the same users seem to value that shopping is co nvenient and useful . In fact,
according to the findings of this study, users’ perception of useful ness is by far one of the
most significant antecedents for users’ continuance intention. Therefore, m -vendors could
benefit from the fact that accessing a mobile website is quicker and more convenient for
many, as opposed to visiting physical shops and man ually browse through their products by
hand. A possible measure of increasing the usefulness could be by controlling the amount
of information available on the webpage, as large amount of information becomes
confusing on small screens. M -vendors could ther efore consider to only place information
that are relevant to the user, as this may increase users’ likelihood of perceiving the mobile
pages as useful (Zhou, 2014a) .

Trusting the process is another important factor. However, since t rust within m –
commerce not only involves users’ trust within m -vendors, but equally within the process
of sending personal information through a wireless network etc., this particular factor is not
entirely controlled by the m -vendors themselves. However, it seems that a weighting factor
is the isolated trust within the vendors, why vendors might yield benefit if inflicting higher
perceptions of trust by complying with product delivery dates, making sure that products
match the descriptions and ensure that product/service qualities are as promised. Also,
vendors might strive to express user -caring and act in their best interest. For example,
establishing clear politics addressing return and privacy concerns. Furthermore, the m –
vendor’s mobile webpage is where the m -vendor connects with the users, and users are
likely to continuously extract information based on the quality of the webpage that help
them rate the trustworthiness of the m -vendor. Thus, ensuring a more customized interface
design , indicating better knowledge of their users, may be vital for m -vendors, as it signals
whether or not the m -vendor is in fact capable of delivering what is expected from them
(Shao Yeh and Li, 2009) . Several companies are already utilizing cookies in the effort of
customizing pages to the individual user.

Finally, the findings suggest that enhancing the websites characteristics to become
interesting, so that users’ perceived enjoyment and sense of control will increase, and

Chapter VI – Recapitulation
Page 102 of 155
equally their desire to immerse and not let one be disturbed by external disturbances, could
significantly improve the likelihood of users con tinuing using m -commerce. For example, it
is possible that m -vendors could engender flow experience by applying exciting and
catching visual effects and/or implement features that are fun and enjoyable to use. One
way of doing this could be by offering use rs the opportunity to share products and offerings
with nearby friends through the location awareness feature. On the other hand, introducing
too many features may increase the workload of the page, which could compromise the
process time. M -vendors may th erefore carefully balance the vividness of the page, and
perhaps offer the possibility of a ‘lite” version that users can select in the case of low or
unstable 3G/4G signal. This might help ensure a more fluent and enjoyable interaction.

However, as alrea dy established, the far most important antecedent of users’
continuance intention to use m -commerce, is users’ overall degree of satisfaction. Not only
does satisfaction hold a direct significant influence, but is equally an important precondition
of effec ts caused by the examined cognitive beliefs. Thus, findings are that if users are not
satisfied with their involvement with m -commerce, both their perceptions of easiness and
trust become redundant aspects of directly influencing users’ intention to repurc hase
through their smartphones. Furthermore, the direct effects of users’ perception s of
usefulness and flow diminish significantly if m -vendors fail to satisfy their users, why large
portions of the effects from the cognitive beliefs will vanish, why this study stress that
ensuring user satisfaction is of utmost importance.

Satisfaction is a composition assessing prior m -commerce experiences, as well as
expectations for future use. Thus, the confirmation of users’ initial expectations before
conducting m -commerce activities are important when assessing satisfaction. However,
subsequent to the purchase, users may alter their perceptions of different beliefs towards
future usage; these beliefs, in turn, also influence satisfaction. This study found that users’
perception of easiness, usefulness, trust and flow all influence satisfaction positively. Also,
the results revealed that these perceptions in fact mediate a large portion of the effect
between confirmation of prior experiences and satisfaction. Thus, as what was the case of

Chapter VI – Recapitulation
Page 103 of 155
satisfaction, these four cognitive beliefs hold the majority of the effect s between prior
experiences and satisfaction. In essence, this means that provid ing a good m -commerce
experience for a certain user is not necessarily satisfactory. The users’ degree of satisfaction
is thus equally dependent on expectations towards future usage.

In regards of personal traits, this study found that highly impulsive pe ople are less likely
to use m -commerce prospectively based on their degree of satisfaction, as opposed to less
impulsive people. The results insinuate that creating satisfaction for less impulsive users is
more important than doing so for highly impulsive people. A plausible reason for this
phenomenon is that the buying processes are severely different between the two groups.
Whereas users without notable impulsive tendencies tend to thoroughly evaluate product
alternatives, delivery concerns and similar co nsequences through rational purchasing
patterns, highly impulsive people tend to act on more sudden affects, and therefore less on
an assessment of general satisfaction. The same trend prevailed as a significantly larger part
of the effect caused by flow t oward continuance intention, was dependent on the
fulfillment of users’ satisfaction. M -vendors may therefore consider establishing systems
that track users ’ profiles. For example, tracking the amount of time spend before a purchase
is made could allow m -vendors to decipher their users’ buying tendencies and separate
impulsive buyers from less impulsive ones. Acquiring this information would enable m –
vendors to recognize the shortcomings of selective investing in flow experience, as this will
have little va lue if their non -impulsive users are not satisfied with the overall experience.

6.2.2. Literature I mplications

From a theoretical perspective , this study contributes current m -commerce literature
by implementing and validating the ECM -IS. This research successfully extended the ECM –
IS with three additional variables: perceived ease of use, trust and flow. The results provided
a unique insight into the underlying relationship between predictor variables , and
illustrated that m -comme rce users include additional post -adoption beliefs beside
perceived usefulness in their decision to continue using m -commerce. Moreover, this study
demonstrated that the condition s causing users ’ intentions to re -engage in m -commerce

Chapter VI – Recapitulation
Page 104 of 155
activities differ acco rding to personal traits. I t is therefore important to recognize m –
commerce users as heterogeneous , who have different needs and perceptions .

6.3. Conclusive R emarks

M-commerce is arguably a shopping channel in rapid growth. However, though more
and more people consid er shopping via a smartphone a viable option, m -commerce, as
defined in this thesi s, struggles with several issues. For instance, one third of all Danish m –
commerce users have cancelled an m -commerce process in the past six months , since m –
commerce somehow insufficiently managed to satisfy them. Furthermore, it seems that
several users are reluctant to continue using m -commerce after the first use. It was
therefore the purpose of this thesis to investigate the factors influencing th e continuance
usage intention to determine, how m -vendors can retain their users and thus establish a
competitive advantage towards other shopping channels. The analysis was based on the
theories of the ECM in conjunction with the TAM, as literature has th oroughly proved that
aspects effecting the intention to adopt a technology , also pose effects in a continuance
context. In addition, this hybrid was extended with the constructs of trust and flow, since
several studies have stressed the effects of these in this context.

Not only did the study investigate these direct effects. Acknowledging the fact that m –
commerce is a process, why some effects might be dependent other aspects , the analysis
was conducted using a mediational approach to find out, which meas ures were prerequisite
for m -vendors to actually yield the proclaimed effects, and how they could eventually
enhance the different aspects of each construct . Finally, this thesis set out to investigate, if
individual personal traits could in fact be of sig nificant importance when evaluating the
different beliefs assessed in this study. Therefore, the moderation s of users’ degree of self –
efficacy and impulsiveness were investigated for some of the regression results.

What are the underlying mechanisms causi ng users to continue using m -commerce, and
are these dependent upon users’ personal traits?

Chapter VI – Recapitulation
Page 105 of 155

These propositions were divided into two separate studies: Study one investigated the
proposed conceptual model to test the direct as well as the mediational effec ts. The model
was tested empirically with 187 respondents, and managed to explain 71 percent of users
overall satisfaction with m -commerce and 66 percent of their continuance usage intention.
All proposed paths were significant. Study two investigated the proposed moderation
effects caused by users’ degree of self -efficacy and impulsiveness, and was tested
empirically with 125 respondents. However, only one of two hypotheses was significant.

The results were that prior experiences with m -commerce directly influence not only
the perceptions of cognitive beliefs regarding future usage, but also the overall satisfaction,
though much of this effect was mediated by the cognitive beliefs. These cognitive beliefs –
percei ved ease of use, perceived usefulness, trust and flow – all have significant influence
on the overall satisfaction and continuance intention, though the effect of perceived ease
of use and trust were fully mediated by satisfaction, why establishing user sa tisfaction is a
necessary measure for m -vendors to yield these effects. Only partial mediation were found
for perceived usefulness and flow, indicating that these beliefs pose direct influence on
users’ continuance intention regardless of the level of sati sfaction, though the effect
diminish if users are not satisfied. Furthermore, the findings suggest that the more
impulsive a user is, the less dependent s/he is of being satisfied to use the service
prospectively. This is also evident, as the mediational e ffect between flow experience and
continuance intention is severely lower for highly impulsive people.

6.4. Limitations

This study suffers from notable limitations that need to be discussed. First off, caution
should be taken in relation to generalization of the results. The non -probability method
used to collect the data is a natural barrier for generalization. Subjects within the population
were not offered the same probabilities of being selected , and responses were driven
mostly on self -selected members, which did not allow for any real control over the

Chapter VI – Recapitulation
Page 106 of 155
distribution or inclusion of subject s. There were also no measures preventing subjects from
retaking the survey , and the lack of geographic documenta tion also impede s the
generalizability of the results. Thus, the true relationship between the sample obtained and
the general population is , as a consequence of the non -probability method , uncertain .

Furthermore, non -response bias may have been a problem in study two , since the data
collection phase was much shorter. Also, one of the goals were to investigate users with
different levels of technological self -efficacy . This, however, does not correspon d well with
the fact that users only had the opportunity to respond through an online survey. Moreover,
it is plausible that the use of self -reporting method can have induced some issues in relation
to the accuracy of the results. Users may not have been able to perfectly recall an m –
commerce experience performed six months a go. A t the same time, questions were not
mixed or reversed coded, which could have encouraged senseless clicking , and increased
the chance s of participants being able to “gu ess” the pur pose s of the study . Furthermore,
mediation analyse s were conducted using cross -sectional data , making it impossible to
unambiguously claim a pattern of causation , given the probability that mediation effects
demonstrated on cross -sectional data becomes ins ignificant on longitudinal data (Maxwell
et al., 2011) . Finally, users’ continuance usage behaviours were measured as intentions and
thus, no certainty about their actual behaviours can be proclaimed.

6.5. Further R esearch

In the future, researchers could accommodate some of the aforementioned
shortcomings of this study by concluding a longitudinal study, which could possibly provide
useful insights regarding behavioural development , and thereby confirm or disconfirm the
proposed causalities provided in this study. Additionally, the process leading users to
continue using m -commerce services, is far more complex than initially anticipated. Thus, it
stresses a flagrant necessity to inv estigate other aspects of importance. E.g., this study is
conducted in Denmark, where the primary m -commerce shopping activities involve
purchasing of low -priced and/or low -risk products or services. An idea would therefore be

Chapter VI – Recapitulation
Page 107 of 155
to investigate the possible d ifferences of perceptions according to product type as well as
the amount of risks associated with the purchase of different products. Also, this study
successfully proved that individuals’ degree of impulsiveness have severe impact on the
proposed relatio nships between selected perceptions. Thus, it seems reasonable that other
personal traits might equally impact this and other relationships, why future studies could
focus on other significant personal traits. Moreover, though this study identified constru cts
inducing large impacts on users’ satisfaction, these impacts had less aggregated impact on
users’ continuance intention. Evidently, though 66 percent of the variance is explained by
this framework, other salient beliefs could advantageously be investig ated to better explain
this phenomena.

Finally, past research have intensively focused on young adults , which is no exception
for this study, and is also aligned with the actual share of m -commerce users in the general
population. However, recently published statistics indicate a rising acceptance of m –
commerce among older generation (DIBS, 2015) , why researchers might focus more on
these groups in the future.

References
Page 108 of 155
References

ADAMS, J., KHAN, H.T .A., REASIDE, R ., 2007. "Research methods for graduate business and
social science students", Sage, India.

AJZEN, I., 1985. "From intention to action: A theory of planned behavior". International
Journal of Kuhl & Beckman J. 11–39.

AKTER, S., RAY, P., AMBRA, J. D ’, 2013. "Continuance of mHealth services at the bottom of
the pyramid: the roles of service quality and trust". Electron Markets. 23, 29 –47.

ALSAJJAN, B.A ., 2014. "Satisfacti on-Trust Model: Developing customer satisfaction and trust
indices for mobile service providers in the UK". International Review of
Management. 3 (1), 1088 -1104

ANDERSON, E.W., SULL IVAN, M.W., 1993. "The antecedents and consequences of customer
satisfacti on for firms". Marketing Science. 12 (2), 125 –143.

ARENS, Z.G., RUST, R .T., 2012. "The duality of decisions and the case for impulsiveness
metrics". Journal of the Academy of Marketing Science . 40 (3), 468 –479.

BANDURA, A., 20 01. "Social cognitive theory of mass communication". Media Psychology .
3 (3), 265 –299.

BANDURA, A ., 1999. "Social cognitive theory: An agentic perspective". Asian Journal of
Social Psychology . 2, 21 -41.

BANDURA, A ., 1997. "Self -efficacy: "The exercise of control". Freeman: New York.

BANDURA, A ., 1994. "Self -efficacy. In V.S. Ramachaudran (Ed.)". Encyclopedia of Human
Behaviour . 4, 71 -81.

BANDURA, A., 1993. "Perceived self -efficacy in cognitive development and functioning".
Educational Psychology . 28 (2), 117 -148.

BANDURA, A., 1991. "Self -Regulation of motivation through anticipatory and self -reactive
mechanisms" In R. A. Dienstbier, "Perspectives on motivation: Nebraska symposium
on motivation". University of Nebraska: Lincoln . 38, 69 -164.

BANDURA , A., 1989. "Human agency in social cognitive theory". American Psychologist . 44
(9), 1175.

BANDURA, A., 1986. "Social foundations of thought and action: A social cognitive theory".
Englewood Cliffs, NJ: Prentice -Hall.

BANDURA, A. , 1977. "Self -efficacy: toward a unifying theory of behavioral change".
Psychological Review . 84 (2), 191 -215.

BANDURA, A., ROSS, D ., ROSS, S.A., 1961. "Transmission of aggression through imitation of
aggressive models". 84 (3), 191 -215. Journal of Abnormal and Social Psychology . 63,
575-582.

BARON, R.M., KENNY, D.A. , 1986. "The moderator –mediator variable distinction in social
psychological research: Conceptual, strategic, and statistical considerations".
Journal of Personality and Social Psychology . 51 (6), 1173.

References
Page 109 of 155

BHATTACHERJEE, A., 2 001. "Understanding information systems continuance: an
expectation -confirmation model". Management Information System Quarterly . 25
(3), 351 –370.

BIGNÉ, E., RUIZ -MAFÉ, C., SANZ -BALZ, S., 2007 . "Key drivers of mobile commerce adoption.
an exploratory study of Spanish mobile users". Journal of Theoretical and Applied
Electronic Commerce Research. 2 (2), 48 –60.

BILGIHAN, A., OKUMUS , F., NUSAIR, K., BU JISIC, M ., 2014. "Online experiences: flow theory,
measuring online customer experience in e-commerce and managerial implications
for the lodging industry". Information Technology & Tourism . 14 (1), 49 –71.

BOUNAGUI, M., NEL, J ., 2009 . "Towards understanding intention to purchase online music
downloads". Management Dynamics: Journal of the S outhern African Institute for
Management Scientists . 18 (1), 15 –26.

BREUSCH, T.S., PAGAN , A.R., 1979. "A simple test for heteroscedasticity and random
coefficient variation". Journal of the Econometric Society. 47 (5), 1287 -1294

BROWN, S.A., VENKATE SH, V ., GOYAL, S. , 2012. "Expectation confirmation in technology
use". Information Systems Research. 23 (2), 474 –487.

BURTON -JONES, A., HUBONA, G .S., 2006. "The mediation of external variables in the
technology acceptance model". Information & Management . 43 ( 6), 706 –717.

CARLANDER, A., GAMBL E, A., GÄRLING, T., HAUFF, J., JOHANSSON , L.O., HOLMEN, M. , 2011.
"Trust -mediated effects of personal services on satisfaction with Banks". 1 -16.

CARUANA, A. , 2002. "Service loyalty: The effects of service quality and the mediating role
of customer satisfaction". European Journal of Marketing. 36 (7/8), 811 –828.

CHANG, S., CHOU, C. , 2011. "Factors affecting user’s online shopping behavior: Integrating
the con straint -based and dedication -based relationship perspectives". African
Journal of Business Management . 5 (2), 370 –382.

CHENG, Y., 2014 . "Extending the expectation -confirmation model with quality and flow to
explore nurses’ continued blended e -learning int ention". Information Technology &
People . 27 (3), 230 –258.

CHEN, Y., CHOU, T., 2012. "Exploring the continuance intentions of consumers for B2C
online shopping: Perspectives of fairness and trust". Online Information Review . 36
(1), 104 –125.

CHIH, W., WU, C.H., LI, H., 20 12. "The antecedents of consumer online buying impulsiveness
on a travel website: Individual internal factor perspectives". Journal of Travel &
Tourism Marketing . 29 (5), 430 –443.

CHII, J., BRAUN, L., 1995 . "Bandura´s self -efficacy th eory: A guide for cardiac rehabilitation".
Journal of holistic nursing . 12 (4), 425 -436

CHIN, W., GOPAL, A., SALISBURY, D., 1997. "Advancing the theory of adaptive structuration:
The development of a scale to measure faithfulness of appropriation". Inform ation
Systems Research . 8 (4), 342 –367.

CHOI, J., SEOL, H., LEE, S., CHO, H., PA RK, Y., 2008 . "Customer satisfaction factors of mobile
commerce in Korea". Internet Research . 18 (3), 313 –335.

References
Page 110 of 155

CHONG, A.Y., 2013. "Understanding mobile commerce continuance intentions: an empirical
analysis of Chinese consumers". The Journal of Computer Information Systems . 53
(4), 22 –30.

CHONG, A.Y., CHAN, F .T.S., OOI, K., 2012 . "Predicting consumer decisions to adopt mobile
comme rce: Cross country empirical examination between China and Malaysia".
Decision Support Systems . 53 (1), 34 –43.

CHUNG, J., TAN, F.B. , 2004. "Antecedents of perceived playfulness: an exploratory study on
user acceptance of general information -searching websites". Information &
Management . 41 (7), 869 –881.

CHUTTUR, M., 2009. "Overview of the technology acceptance model: Origins,
developments a nd future directions". Working paper on Information Systems . 1-21.

COBB, C.J., HOYER, W .D., 1986 . "Planned versus impulse purchase behavior". Journal of
Retailing . 62 (4), 384 –409.

COMPEAU, D.R., HIGGI NS, C.A., 1995 . "Computer self -efficacy: Development of a measure
and initial test". Management Information System Quarterly . 189 –211.

CONRAD, F., SCHOBER, M., 2000. "Clarifying question meaning in a household telephone
survey". The public opinion quarterly . 64 (1), 1 –28.

COOK, R.D., WEISBERG , S., 1982 . "Residuals and influence in regression, Monographs on
statistics and applied probability". Chapman and Hall, New York .

CSIKSZENTMIHALYI, M. , 1997 . "Finding flow". Psychology Today . July/August. 46.

CSIKSZENTMIHALYI, M. , 1975 . "Play and Intrinsic Rewards". Journal of Humanistic
Psychology . 15 (3), 41 -63

CSIKSZENTMIHALYI, M. , LEFEVRE, J., 1989. "Optimal experience in work and leisure". Journal
of Personality and Social Psychology . 56 (5), 815 -822.

CYR, D., HEAD, M., I VANOV, A., 2006 . "Design aesthetics leading to m -loyalty in mobile
commerce". Information & Management . 43 (8), 950 –963.

DAVIS, F.D., 1989 . "Perceived usefulness, perceived ease of use, and user acceptance of
information technology". Management Information System Quarterly . 13 (1), 139 –
339.

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.

DE RUYTER, K., WETZE LS, M., KLEIJNEN, M. , 2001 . "Customer adoption of e -service: an
experimental study". International Journal of Service Industry Management . 12 (2),
184–207.

DIBS, 2015. "Dansk E -handel 2014: DIBS årlig rapport om E -handel, mobil handel og online
betaling". Copenhagen.

DIGI -CAPITAL , 2014. "Mobile internet to grow 300% to $700B in 4 years" [Online]. Available:
Link [Accessed July 2015].

References
Page 111 of 155
DROSSOS, D.A., KOKKI NAKI , F., GIAGLIS, G.M., FOUSKAS, K.G., 2014 . "The effects of product
involvement and impulse buying on purchase intentions in mobile text advertising".
Electronic Commerce Research and Applications . 13 (6), 423 –430.

DST, 2015. "Elektronik i hjemmet". [Online].
Available : Link. [Accessed April 2015].

EMARKETER, 2013 . "Tablets, smartphones drive mobile commerce to record heights
eMarketer " [Online].
Available: Link [Accessed July 2015].

EMV, 2015 . "Agreement on the growth plan for digitization in Denmark" [Online]. Available:
Link [Accessed April 2015]

ETZIONI, A., 1986. "Rationality is anti -entropic". Journal of Economic Psychology . 7, 17 -36.

FIELD, A.P., 2013 . "Discovering statistics using IBM SPSS statistics: and sex and drugs and
rock “n” roll" . 4th edition . Sage, Los Angeles .

FIELD, A.P., 2009 . "Discovering statistics using SPSS: (and sex, drugs and rock “n” roll)". 3rd.
Edition . Sage, Los Angeles .

FISHBEIN, M., AJZEN, I., 1975. "Belief, attitude, intention and behavior: An introduction to
theory and research". Addison -Wesley Publishing Company .

FLIGHT, R.L., ROUNTR EE, M.M., BEATTY , S.E., 2012. "Feeling the urge: Affect in impulsive
and compulsive buying". The Journal of Marketing Theory and Practice . 20 (4), 453 –
466.

FORNELL, C., LARCKER , D., 1981 . "Evaluating structural equation models with unobservable
variables and measurement error. Journal of Marketing Research . 18, 39 –50.

FRAZIER, P.A., TIX, A.P., BARRON, K.E., 2004. "Testing moderator and mediator effects in
counseling psychology re search. Journal of Counseling Psychology . 51 (1), 115 -134.

FRITZ, M.S., MACKINN ON, D.P., 2007. "Required sample size to detect the mediated effect".
Psychological Science . 18 (3), 233 –239.

GAO, L., BAI, X., 20 14. "An empirical study on continuance intent ion of mobile social
networking services: Integrating the IS success model, network externalities and
flow theory". Asia Pacific Journal of Marketing and Logistics . 26 (2), 168 –189.

GEFEN, D., 2002 . "Reflections on the dimensions of trust and trustworthi ness among online
consumers". ACM Sigmis Database 33 (3), 38 –53.

GEFEN, D., DETMAR W, S., ELENA, K., 2003A . "Trust and TAM in online shopping: An
integrated model". Management Information Quarterly . 27 (1), 51 -90.

GEFEN, D., KARAHANNA , E., STRAUB, D.W., 2003B. "Potential and repeat e -consumers: the
role of and trust vis -a-vis TAM". IEEE Transactions on Engineering Management . 50
(3), 307 -321.

GEFEN, D., STRAUB, D .W., 2004 . "Consumer trust in B2C e -Commerce and the importance
of social presence: experimen ts in e -Products and e -Services". The International
Journal of Management Science . 32, 407 –424.

References
Page 112 of 155
GEORGE, J.F., 2004 . "The theory of planned behavior and Internet purchasing". Internet
Research . 14 (3), 198 –212.

GHOSH, A.K., SWAMINA THA, T.M., 2001. "Software security and privacy risks in mobile e –
commerce". Communications of the ACM . 44 (2), 51 –57.

GITAU, L., NZUKI, D. , 2014 . Analysis of Determinants of M -Commerce Adoption by Online
Consumers. International Journal of Business, Humanities and Techn ology . 4 (3) 88 –
94.

GROß, M., 2015. "Mobile shopping: a classification framework and literature review".
International Journal of Retail & Distribution Management . 43 (3), 221 –241.

HALILOVIC, S., CICIC , M., 2013 . "Antecedents of information systems user behaviour –
extended expectation -confirmation model". Behaviour & Information Technology .
32 (4), 359 –370.

HAYERS, A.F., CAI, L ., 2007. "Using heteroskedasticity -consistent standard error estimators
in OLS regression: An introduction and software implementation". Behavioral
Research Methods . 39 (4), 709 -722.

HAYES, A.F., PREACHE R, K.J., 2014 . "Statistical mediation analysis with a multi categorical
independent variable". British Journal of Mathematical and Statistical Psychology .
67 (3), 451 –470.

HERNÁNDEZ, B., JIMÉN EZ, J., MARTÍN, M.J. , 2010 . "Customer behavior in electronic
commerce: The moderating effect of e -purchasing experience". Journal of Business
Research . 63 (9/10), 964 –971.

HMIELOWSKI, J.D., HU TCHENS, M.J., CICCHI RILLO, V.J., 2014 . "Living in an age of online
incivility: examining the conditional indirect effects of online discussion on political
flaming". Information, Communi cation & Society . 17 (10), 1196 –1211.

HOFFMAN, D.L., NOVAK , T.P., 2009 . "Flow Online: Lessons Learned and Future Prospects".
Journal of Interactive Marketin g. 23 (1), 23 –34.

HOFFMAN, D.L., NOVAK , T.P., 1996. "Marketing in hypermedia computer -mediated
environments: conceptual foundations". The Journal of Marketing . 16, 50 –68.

HONG, S., THONG, J.Y .L., TAM, K.Y., 2006 . "Understanding continued information
technology usage behavior: A comparison of three models in the context of mobile
internet". Decision Support Systems . 42 (3), 1819 –1834.

HSU, C., CHANG, K., CHEN, M. -C., 2012. "The impact of website quality on customer
satisfaction and purchase intention: perceived playfulness and perceived flow as
mediators". Information Systems and e -Business Managemen t. 10 (4), 549 –570.

HSU, C., WU, C., CHE N, M., 2013. "An empirical analysis of the antecedents of e -satisfaction
and e -loyalty: focusing on the role of flow and its antecedents". Information Systems
and e -Business Management. 11 (2), 287 –311.

HUNG, M., YANG, S., HSIEH, T., 2012 . "An examination of the determinants of mobile
shopping continuance". International Journal of Electronic Business Management .
10 (1), 29 -37.

IPSOS, 2013 . "Our mobile planet: Danmark: Lær mobilforbrugeren at kende". Denmark.

References
Page 113 of 155
IRIANTO, H., 2015 . "Consumers’ attitude and intention towards orgnic food purchase: An
extension of theory of planned behavior in gender perspective". International
Journal of Management, Economies and Social Sciences . 4 (1), 17 –31.

AL-JABRI, I.M., 2015 . "An tecedents of user satisfaction with ERP systems: mediation
analyses". Kybernetes . 44 (1), 107 –123.

JARADAT, M. -I.R.M., FAQIH, K.M.S ., 2014 . "Investigating the moderating effects of gender
and self -efficacy in the context of mobile payment adoption: A developing country
perspective". International Journal of Business and Management . 9 (11), 147 -169.

JARVENPAA, S., TRACT INSKY, N., VITALE, M ., 2000 . "Consumer trust in an Internet store".
Information Technology and Management . 1 (1), 45 -71.

JOHNSON, D., GRAYSON , K., 2005 . "Cognitive and affective trust in service relationships".
Journal of Business Research . 58 (4), 500 –507.

JOUBERT, J., VAN BELLE, J., 2013 . "The role of trust and risk in mobile commerce adoption
within South Africa". International Journal of Business, Humanities and Technology .
3 (2), 27 –38.

KENNY, D.A., KASHY, D., BOLGER, N., 1998. "Data Analysis in Social Psychology". In D.
Gilbert, S. Fiske & G. Lindsey (Eds.), "Handbook of social psychology (4th edition,
233-265). McGraw Hill: New York.

KHEDHAOURIA, A., BEL DI, A., BELBALY, N., 2014. "The moderating effect of gender on
continuance intention for mobile Internet services (MIS). System Information and
Management . 18 (3), 117 –137.

KIM, B., 2010. "An empirical investigation of mobile data service continuance: Incorporating
the theory of planned behavior into the expectation –confirmatio n model". Expert
Systems with Applications . 37 (10), 7033 –7039.

KIM, D.J., FERRIN, D .L., RAO, H.R., 2009 . "Trust and satisfaction, two stepping stones for
successful e -commerce relationships: A longitudinal exploration". Information
Systems Research . 20 ( 2), 237 –257.

KIM, D.J., FERRIN, D .L., RAO, H.R., 2008 . "A trust -based consumer decision -making model in
electronic commerce: The role of trust, perceived risk, and their antecedents".
Decision Support Systems . 44 (2), 544 –564.

KNAPP, T.R., 1991 . "Focus o n psychometrics coefficient alpha: Conceptualizations and
anomalies. Research in Nursing & Health . 14 (6), 457 –460.

KORZAAN, M., 2003. "Going with the flow: Predicting online purchase intention". Journal of
Computer Information Systems . 25-31.

KOUFARIS, M., 2002 . "Applying the technology acceptance model and flow theory to online
consumer behavior". Information systems research . 13 (2), 205 –223.

LANCE, C.E., 2006 . "The Sources of four commonly reported cutoff criteria: What did they
Really say? Organizational Research Methods . 9 (2), 202 –220.

LEE, H., 2014 . "Perceived quality as a key antecedent in continuance intention on mobile
commerce". International Journal of Electronic Commerce Studies . 5(2), 123 –142.

References
Page 114 of 155
LEE, K.C., CHUNG, N. , 2009 . "Understanding factors affecting trust in and satisfaction with
mobile banking in Korea: A modified DeLone and McLean’s model perspective".
Interacting with Computers . 21 (5/6), 385 –392.

LEE, K.C., KANG, I., MCKNIGHT, D.H ., 2007. "Transfer from offline trust to key online
perceptions: An empirical study". IEEE Transactions on Engineering Management .
54 (4), 729 –741.

LEGRIS, P., INGHAM, J., COLLERETTE, P., 2003. "Why do people use information technology?
A critical review of the technology acceptance mo del". Information & managemen t.
40 (3), 191 –204.

LIM, H., DUBINSKY, A .J., 2005 . "The theory of planned behavior in e -commerce: Making a
case for interdependencies between salient beliefs". Psychology and Marketing . 22
(10), 833 –855.

LIM, H., WIDDOWS, R. , PARK, J., 2006 . "M ‐loyalty: winning strategies for mobile carriers".
Journal of Consumer Marketing . 23 (4), 208 –218.

LIM, N., 2003. " Consumers’ perceived risk: sources versus consequences". Electronic
Commerce Research and Applications . 2 (3), 216 –228.

LIN, H., WANG, Y., 2 006. "An examination of the determinants of customer loyalty in mobile
commerce contexts". Information and Management . 43 (3), 271 –282.

LIN, J., WANG, B., W ANG, N., LU, Y., 201 4. "Understanding the evolution of consumer trust
in mobil e commerce: a longitudinal study". Information Technology and
Management . 15 (1), 37 –49.

LIU, C., MARCHEWKA, J.T., LU, J., YU, C. , 2005 . "Beyond concern —a privacy -trust -behavioral
intention model of electronic commerce". Information and Management . 42 (2 ),
289–304.

LONGE, O., BOATENG, R., LONGE, F., OLATU BOSUN, K., 2010 . "Information &
communication technology adoption among adults in South Western Nigeria: An
assessment of usage -phobia factors". Journal of Information Technology Impact . 10
(1), 65 -86.

LU, J., 2014. "Are personal innovativeness and social influence critical to continue with
mobile commerce?" Internet Research . 24 (2), 134 –159.

LU, Y., ZHOU, T., WA NG, B., 2009 . "Exploring Chinese users’ acceptance of instant messaging
using the theory of planned behavior, the technology acceptance model, and the
flow theory". Computers in Human Behavior . 25 (1), 29 –39.

LUQMAN, A., RAZAK, R .C., ISMAIL, M., 201 4. "A Review on mobile commerce continuance
intention studies and its underlying model. Journal o f Basic and Applied Scientific
Research . 4 (3), 1 -9.

MACKINNON, D.P., LOC KWOOD, C.M., HOFFMAN , J.M., WEST, S.G., SHEETS, V., 2002 . "A
comparison of methods to test mediation and other intervening variable effects".
Psychological Methods . 7 (1), 83 –104.

MACKINNON, D.P., LOC KWOOD, C.M., WILLIAM S, J., 2004 . "Confidence limits for the
indirect effect: Distribution of the product and resampling methods. Multivariate
Behavioral Research . 39 (1), 99 –128.

References
Page 115 of 155
MAHATANANKOON, P., W EN, H.J., LIM, B., 2 005. "Consumer -based m -commerce: exploring
consumer perception of mobile applications". Computer Standards and Interfaces .
27 (4), 347 –357.

MALHOTRA, N., BIRKS, D.F., 2011 . "Marketing Research: An applied approach". 3rd. Edition.
Pearson Education Limited .

MALHOTRA, N.K., BIRKS, D.F., 2 006. "Marketing research: an applied approach". Updated
2nd. Edition. Pearson Education, Harlow .

MARTIN, E., POLIVKA, A.E., 1995 . "Diagnostics for redesigning survey questionnaires
measuring work in the current population survey". Publ ic Opinion Quarterly . 59 (4),
547–567.

MAXWELL, S.E., COLE, D.A., MITCHELL, M.A. , 2011 . "Bias in cross -sectional analyses of
longitudinal mediation: Partial and complete mediation under an autoregressive
model". Multivariate Behavioral Researc h. 46 (5), 8 16–841.

MAYER, R.C., DAVIS, J.H., SCHOORMAN, F.D ., 1995 . "An integrative model of organizational
trust". Academy of Management Review . 20 (3), 709 –734.

MCKNIGHT, D.H., CHOU DHURY, V., KACMAR, C ., 2002 . "Developing and validating trust
measures for e -comme rce: An integrative typology". Information systems research .
13 (3), 334 –359.

MCKNIGHT, D.H., CUMM INGS, L.L., CHERVANY , N.L., 1998 . "Initial trust formation in new
organizational relationships". Academy of Management review . 23 (3), 473 –490.

MOHAMED, N., HUSSEIN, R., HIDAYAH AHMAD ZAMZURI, N., H AGHSHENAS, H., 2014 .
"Insights into individual’s online shopping continuance intention". Industrial
Management and Data Systems . 114 (9), 1453 –1476.

MOLINA -CASTILLO, F., LÓPEZ -NICOLÁS, C., BOUWMAN , H., 2008 . "Expl aining mobile
commerce services adoption by different type of customers". Systemics, Cybernetics
and Informatics . 6 (6), 73 –79.

NEL, D., VAN NIEKERK , R., BERTHON, J. -P., DAVIES, T., 1999 . "Going with the flow: Web sites
and customer involvement". Internet Research. 9 (2), 109 –116.

NOVAK, T.P., HOFFMAN , D.L., YUNG, Y., 20 00. "Measuring the customer experience in
online environments: A structural modeling approach". Marketing Science . 19 (1),
22–42.

OBADĂ, D.R., 2013 . "Flow theory and online marketing outcomes: A critical literature
review". Procedia Economics and Finance . 6, 550 –561.

O’CASS, A., CARLSON, J., 2010 . Examining the effects of website ‐induced flow in
professional sporting team websites. Internet Research . 20 (2), 115 –134.

OLIVER, R.L., 1980 . "A cognitive model of the antecedents and consequences of satisfaction
decisions". Journal of Marketing Research . 27, 460 –469

OLIVER, R.L., BEARDE N, W.O., 1985 . "Disconfirmation processes and consu mer evaluations
in product usage". Journal of Business Research . 13, 235 -246.

OLIVER, R.L., DESARB O, W.S., 1988 . "Response determinants in satisfaction judgments".
Journal of Consumer Research . 14 (4) 495 –507.

References
Page 116 of 155

OSBORNE, J., WATERS, E., 2002 . "Four assumpt ions of multiple regression that researchers
should always test". Practical Assessment, Research and Evaluation . 8 (2), 1 –9.

PALLANT, J., 2007 . "SPSS Survival Manual: A step by step guide to data analysis using SPSS
for windows". 3rd. Edition. McGraw Hill : Berkshire.

PAPPAS, I., G. PATEL I, A., N. GIANNAKOS, M., CHRISSIKOPOULOS, V., 2014 . "Moderating
effects of online shopping experience on customer satisfaction and repurchase
intentions". International Journal of Retail & Distribution Management . 42 (3), 187–
204.

PARBOTEEAH, D.V., VA LACICH, J.S., WELLS, J.D., 2009 . "The Influence of website
characteristics on a consumer’s urge to buy impulsively. Information Systems
Research . 20 (1), 60 –78.

PARK, J.E., CHOI, E. J., 2013 . "Consequences of impulse buying cr oss-culturally: a qualitative
study". International Journal of Software Engineering and its Applications . 7 (1),
240–60.

PAVLOU, P.A., FYGENS ON, M., 2006 . "Understanding and predicting electronic commerce
adoption: An extension of the theory of planned be havior". Management
Information System Research Quarterly . 30 (1), 115 –143.

PREACHER, K.J., HAYE S, A.F., 2008 . "Asymptotic and resampling strategies for assessing and
comparing indirect effects in multiple mediator models". Behavior Research
Methods . 40 ( 3), 879 –891.

PREACHER, K.J., HAYE S, A.F., 2004. "SPSS and SAS procedures for estimating indirect effects
in simple mediation models". Behavior Research Methods, Instruments, and
Computers . 36 (4), 717 –731.

PREACHER, K.J., RUCK ER, D.D., HAYES, A.F ., 2007 . "Addressing moderated mediation
hypotheses: Theory, methods, and prescriptions". Multivariate behavioral research .
42 (1), 185 –227.

RAYKOV, T., 1997 . "Estimation of composite reliability for congeneric measures". Applied
Psychological Measurement . 21 (2 ), 173 –184.

ROBINS, R.W., FRALEY , C.R., KRUEGER, R.F ., 2007 . "Handbook of Research Methods in
Personality Psychology". 224 -239. Guilford: New York.

ROGELBERG, S.G., CON WAY, J.M., SEDERBURG , M.E., SPITZMÜLLER, C., AZIZ, S., KNIGHT ,
W.E., 200 3. "Profiling active and passive nonrespondents to an organizational
survey. Journal of Applied Psychology. 88 (6), 1104 –1114.

ROOK, D.W., 1987 . "The buying impulse". Journal of Consumer Research . 14 (2), 189 –199.

ROOK, D.W., FISHER, R.J., 1995. "Normat ive influences on impulsive buying behavior".
Journal of Consumer Research . 22 (3), 305 –313.

ROSE, B.M., HOLMBECK , G.N., COAKLEY, R.M ., FRANKS, E.A., 200 4. "Mediator and
moderator effects in developmental and behavioral pediatric research". Journal of
Developmental & Behavioral Pediatrics . 25 (1), 58 –67.

References
Page 117 of 155
ROSE, S., CLARK, M., SAMOUEL, P., HAIR, N ., 2012 . "Online customer experience in e –
retailing: An empirical model of antecedents and outcomes". Journal of Retailing . 88
(2), 308 –322.

SADI, A.H.M., NOORDI N, M.F., 2011 . "Factors influencing the adoption of m -commerce: An
exploratory analysis". International Conference on Industrial Engineering and
Operations Management. 492-499.

SAN‐MARTIN, S., LÓPEZ ‐CATALÁN, B., 2013 . "How can a mobile vendor get satisfied
customers?" Industrial Management and Data Systems . 113 (2), 156 –170.

SCHEFTER, P., REICHH ELD, F., 2000. "E-loyalty: your secret weapon on the web". Harvard
Business Review . 78 (4), 105.

SHAO YEH, Y., LI, Y. , 2009 . "Building trust in m -commerce: contributions from quality and
satisfaction". Online Information Review . 33 (6), 1066 –1086.

SHARKEY, U., ACTON, T., CONBOY, K., 2012 . "Optimal experience in online shopping: the
influence of flow". 1 -12.

SHROUT, P.E., BOL GER, N., 2002 . "Mediation in experimental and nonexperimental studies:
New procedures and recommendations". Psychological Methods. 7(4), 422 –445.

SIEGRIST, M., GUTSCH ER, H., EARLE, T., 2 005. "Perception of risk: the influence of general
trust, and general confidence". Journal of Risk Research . 8 (2), 145 –156.

SIEKPE, J.S., 2005. "An examination of the multidimensionality of flow construct in a
computer -mediated environment". Journal of Elec tronic Commerce Research . 6 (1),
31–43.

SKADBERG, Y.X., KIMM EL, J.R., 2004. "Visitors’ flow experience while browsing a Web site:
its measurement, contributing factors and consequences". Computers in Human
Behaviour . 20 (3), 403 –422.

SØREBØ, Ø., HALVARI , H., GULLI, V.F., K RISTIANSEN, R., 2009 . "The role of self –
determination theory in explaining teachers’ motivation to continue to use e –
learning technology". Computers and Education . 53 (4), 1177 –1187.

SPRENG, R.A., OLSHAV SKY, R.W., 1993 . "A desires cong ruency model of consumer
satisfaction". Journal of the Academy of Marketing Science . 21 (3), 169 –177.

STEFL -MABRY, J., 2003. "A social judgment analysis of information source preference
profiles: An exploratory study to empirically represent media selecti on patterns".
Journal of the American Society for Information Science and Technology . 54 (9), 879 –
904.

STREINER , D.L., 2003. "Starting at the beginning: an introduction to coefficient alpha and
internal consistency". Journal of personality assessment . 80 (1), 99 –103.

SWAN, J.E., TRAWICK , F., 1981. "Disconfirmation of expectations and satisfaction with a
retail service". Journal of Retailing. 57 (3), 49 -67.

TABACHNICK, B.G., FI DELL, L.S., 2007. "Using multivariate statistics". 5th. Edition.
Pearson/Allyn & Bacon: Boston.

ERHVERVSSTYRELSEN, 2 014. "Telestatistik: Første halvår". Denmark .

References
Page 118 of 155
ERHVERVSSTYRELSEN, 2013. "Økonomiske nøgletal for telebranchen 2013". Denmark.

THONG, J.Y.L., HONG, S., TAM, K.Y., 2006. "The effects of post -adoption beliefs on the
expectation -confirmation model for information technology continuance".
International Journal of Human -Computer Studies . 64 (9), 799 –810.

TIWARI, R., BUSE, S. , 2007. "The mobile commerce prospects: A st rategic analysis of
opportunities in the banking sector". Hamburg University Press: Hamburg .

TRIVEDI, J.P., KUMAR , S., 2014. "Determinants of mobile commerce acceptance amongst
gen Y". Journal of Marketing Management . 2 (2), 145 –163.

TSU WEI, T., MARTHANDAN, G., YEE ‐LOONG CHONG, A., OOI , K., ARUMUGAM, S ., 2009.
"What drives Malaysian m -commerce adoption? An empirical analysis". Industrial
Management and Data Systems . 109 (3), 370 –388.

VASILEIADIS, A., 2014. "Security concerns and trust in the adop tion of m -commerce". Social
Technologies . 4 (1), 179 –191.

VENKATESH, V., MORRI S, M.G., DAVIS, G.B. , DAVIS, F.D., 2003. "User acceptance of
information technology: Toward a unified view". Management Information System
Quarterly. 27 (3), 425 –478.

VIJAYASAR ATHY, L.R., 2004. "Predicting consumer intentions to use on -line shopping: the
case for an augmented technology acceptance model". Information and
Management . 41 (6), 747 –762.

WANGPIPATWONG, S., C HUTIMASKUL, W., PAPA SRATORN, B., 2008 . "Understanding
citizen’s continuance intention to use e -government website: A composite view of
technology acceptance model and computer self -efficacy". The Electronic Journal of
E-government . 6 (1), 55 –64.

WANG, Y. -S., LIN, H. -H., LUARN, P., 2006 . "Predicting consumer inte ntion to use mobile
service". Information Systems Journal . 16 (2), 157 –179.

WEISBERG, J., TE’ENI , D., ARMAN, L., 201 1. "Past purchase and intention to purchase in e ‐
commerce: The mediation of social presence and trust". Internet Research. 21 (1),
82–96.

WONG, C.H., LEE, H.S ., LIM, Y.H., CHUA, B.H., TAN, G.W.H. , 2012. "Predicting the consumers’
intention to adopt mobile shopping: an emerging market perspective". International
Journal of Network and Mobile Technologies . 3 (3), 24 –39.

WOOD, R., BANDURA, A ., 1989. "Social cognitive theory of organizational management".
Academy of Management Review . 14 (3), 361 –384.

WU, A.D., ZUMBO, B.D ., 2008 . "Understanding and using mediators and moderators". Social
Indicators Research . 87 (3), 367 –392.

XIN DING, D., HU, P.J.-H., VERMA, R., WARDE LL, D.G., 2010 . "The Impact of service system
design and flow experience on customer satisfaction in online financial services".
Journal of Service Research . 13 (1), 96 –110.

YANG, H., YOO, Y., 2 004. "It’s all about attitude: revi siting the technology acceptance
model". Decision Support Systems . 38 (1), 19 –31.

References
Page 119 of 155
YANG, K., 2012 . "Consumer technology traits in determining mobile shopping adoption: An
application of the extended theory of planned behavior". Journal of Retailing and
Consumer Service s. 19 (5), 484 –491.

YANG, K., 2010 . "The effects of technology self -efficacy and innovativeness on consumer
mobile data service Adoption between American and Korean consumers". Journal
of Consumer Marketing . 22 (3), 117 –127.

ZHANG, L., ZHU, J., LIU, Q., 201 2. "A meta -analysis of mobile commerce adoption and the
moderating effect of culture". Computers in Human Behavior . 28 (5), 1902 –1911.

ZHOU, T., 2014A . "Understanding continuance usage intention of mobile internet sites".
Universal Access in the Information Society . 13, 329 –337.

ZHOU, T., 2014 B. "Examining continuance usage of mobile Internet services from the
perspective of resistance to change". Information Development . 30 (1), 22 –31.

ZHOU, T., 2013A . "An empirical examination of conti nuance intention of mobile payment
services". Decision Support Systems . 54 (2), 1085 –1091.

ZHOU, T., 2013B. "Understanding continuance usage of mobile sites". Industrial
Management and Data Systems . 113 (9), 1286 –1299.

ZHOU, T., 2013C . "The effect of fl ow experience on user adoption of mobile TV". Behaviour
and Information Technology . 32 (3), 263 –272.

ZHOU, T., 2011 . "Understanding mobile Internet continuance usage from the perspectives
of UTAUT and flow". Information Development . 27 (3), 207 –218.

ZHO U, T., LI, H., LIU, Y., 2010 . "The effect of flow experience on mobile SNS users’ loyalty".
Industrial Management and Data Systems . 110 (6), 930 –946.

ZIAULLAH, M., FENG, Y., AKHTER, S.N., KH AN, M.F ., "University of electronic science and
technology of ch ina, 2014. An empirical study on exploring relationship among
information quality, e -satisfaction, e -trust and young generation’s commitment to
Chinese online retailing". Journal of Competitiveness . 6 (4), 3 –18.

Appendices
Page 120 of 155

APPENDICES

Appendices
Page 121 of 155
Appendix 1 – Questionnaire Items:

Constructs Items Measure Sources
Confirmation CON1 My experience with using the mobile commerce services was better than what I expect
Bhattacherjee
(2001) CON2 The service level provide by the mobile commerce services was better than what I expected
CON3 Overall, most of my expectation from using mobile commerce services were confirmed
Perceived
Ease of Use PEOU1 Learning how to use m -commerce services is easy for me
Thong et al. (2006) ,
adopted from Davis
et al. (1989) PEOU2 My interaction with m -commerce services is clear and understandable
PEOU3 I find m -commerce services easy to use
PEOU4 It is easy for me to become skillful at using m -commerce services
Perceived
Usefulness PU1 I find m -commerce useful in my daily life (such as web browsing, mobile shopping, etc.)
Thong et al. (2006) ,
adopted from Davis
et al. (1989) PU2 Using m -commerce enables me to accomplish tasks more quickly
PU3 Using m -commerce services helps me perform many things more conveniently
PU4 Using m -commerce increases my productivity (for example, makes my life easier)
Trust
TR1 This service provider is trustworthy Zhou (2013a) ,
adopted from
(Gefen et al.,
2003a) TR2 This service provider keeps its promise
TR3 This service provider keeps customers' interests in mind
TR4 Based on my experience with the m -commerce vendor in the past, I know it is predictable
Flow FL1 When using m -commerce, my attention was focused on the activity Zhou (2014b) ,
adopted from Lee
et al. (2007 ) FL2 When using m -commerce, I felt in control
FL3 When using m -commerce, I found a lot of pleasure
Satisfaction SAT1 I am satisfied with decision on the use of mobile commerce services Lee (2014) ,
adopted from
Spreng and
Olshavsky (1993) SAT2 My choice to use the mobile commerce services was a wise one
SAT3 I think I did the right thing by deciding to use my mobile commerce services
Continuance
Intention CI1 I intend to continue using m -commerce services in the future Thong et al. (2006) ,
adapted from
Bhattacherjee
(2001) CI2 I will always try to use m -commerce services in my daily life
CI3 I will keep using m -commerce services as regularly as I do now
Self-Efficacy SE1 I am satisfied with decision on the use of mobile commerce services
Yang (2010) SE2 My choice to use the mobile commerce services was a wise one
SE3 I think I did the right thing by deciding to use my mobile commerce services
Impulsiveness IMP1 I am satisfied with decision on the use of mobile commerce services
Rook and Fisher
(1995) IMP2 My choice to use the mobile commerce services was a wise one
IMP3 I think I did the right thing by deciding to use my mobile commerce services

Appendices
Page 122 of 155
Appendix 2 – Independent t-test for Online and Offline Respondents:

Appendix 2 A – Satisfaction

DV Site of survey n Mean Std.
deviation Std. error
mean
SAT Online 148 5.82 1.11 .09
Offline (Bus stop) 39 5.98 1.14 .18

Levene’s test for
equality of
variances t-test for equality of means
DV Hypotheses F Sig. t df Sig (2 –
tailed) Mean
difference s.e.
difference
SAT Equal variances
assumed .07 .80 -.80 185 .42 -.16 .20
Equal variances
not assumed -.79 58.078 .44 -.16 .20

Appendices
Page 123 of 155
Appendix 2B – Continuance Intention

DV Site of survey n Mean Std.
deviation Std. error
mean
CI Online 148 5.74 1.16 0.10
Offline (Bus stop) 39 6.00 1.07 0.17

Levene’s test for
equality of
variances t-test for equality of means
DV Hypotheses F Sig. t df Sig (2 –
tailed) Mean
difference s.e.
difference
CI Equal variances
assumed 1.59 .21 -1.26 185 .21 -.26 .21
Equal variances
not assumed -1.32 63.520 .19 -.26 .20

Appendices
Page 124 of 155
Appendix 3 – Independent t-test for Early and Late Respondents:

Appendix 3 A – Satisfaction

DV Period of
response n Mean Std.
deviation Std. error
mean
SAT Early 15 5.98 1.06 .27
Late 15 5.71 1.21 .31

Levene’s test for
equality of
variances t-test for equality of means
DV Hypotheses F Sig. t df Sig (2 –
tailed) Mean
difference s.e.
difference
SAT Equal variances
assumed .69 .41 .641 28 .526 .27 .42
Equal variances
not assumed .641 27.50 .527 .27 .42

Appendices
Page 125 of 155
Appendix 3B – Continuance Intention

DV Period of
response n Mean Std.
deviation Std. error
mean
CI Early 15 6.11 .87 .22
Late 15 5.93 1.16 .30

Levene’s test for
equality of
variances t-test for equality of means
DV Hypotheses F Sig. t df Sig (2 –
tailed) Mean
difference s.e.
difference
CI Equal variances
assumed 1.70 .20 .47 28 .64 .18 .37
Equal variances
not assumed .47 25.29 .64 .18 .37

Appendices
Page 126 of 155
Appendix 4 – Demography Analysis – Study One:

84
103
M A LE FEM A LEGENDER14
141
17
6
97,49%
75,40%
9,09%
3,21%
4,81%
18-20 21-30 31-40 41-50 > 50AGE

Appendices
Page 127 of 155
9
16
47
28
48
25
0
144,81%
8,56%
25,13%
14,97%
25,67%
13,37%
0,00%
7,49%HIGHEST COMPLETED EDUCATION7
4
77
89
4
63,74%
2,14%
41,18%
47,59%
2,14%
3,21%
UEN M PLOYED PA RT TI M E
JOBTEN URE STUDEN T SELF –
EM PLOYEDOTH ERPRIMARY OCCUPATION23
76
12
18
22
17
1912,30%
40,64%
6,42%
9,63%
11,76%
9,09%
10,16%
< 6000 6. 001 –
15. 00015. 001 –
20. 00020. 001 –
25. 00025. 001 –
30. 00030. 001 –
35. 000> 35. 000INCOME LEVEL

Appendices
Page 128 of 155

20
108
44
12
310,70%
57,75%
23,53%
6,42%
1,60%
< 1 1 -3 3 -5 5 -7 > 7SMARTPHONE USAGE (HOURS PER DAY)
43
143
40
218
142
42
5M-COMMERCE TRANSACTIONS (PAST 6
MONTHS)13
44
48
53
10
7
126,95%
23,53%
25,67%
28,34%
5,35%
3,74%
6,42%
< 1 1 -2 2 -3 3 -4 4 -5 5 -6 > 6M-COMMERCE EXPERIENCE (YEARS)

Appendices
Page 129 of 155
Appendix 5 – Demography Analysis – Study Two:

59
66
M A LE FEM A LEGENDER7
89
14
6
9
18 –20 21-30 31-40 41-50 > 50AGE

Appendices
Page 130 of 155

11
12
25
29
23
14
0
11HIGHEST COMPLETED EDUCATION4
9
56
45
3
8
UN EM PLOYED PA RT TI M E
JOBTEN URE STUDEN T SELF –
EM PLOYEDOTH ERPRIMARY OCCUPATION

Appendices
Page 131 of 155
Appendix 6 – Scale R eliabilities – Study One:

Construct Item Mean Std. deviation Loading s CR AVE
Confirmation CON1 5.29 1.24 .88
.90 .75 CON2 4.91 1.33 .90
CON3 5.60 1.14 .81
Perceived Ease of
Use PEOU1 6.16 1.04 .90
.95 .83 PEOU2 6.02 1.08 .92
PEOU3 6.21 1.01 .94
PEOU4 6.43 0.89 .89
Perceived
Usefulness PU1 6.16 1.20 .80
.89 .66 PU2 6.44 0.89 .83
PU3 6.11 1.08 .84
PU4 5.39 1.50 .77
Trust TR1 5.74 1.19 .82
.92 .75 TR2 5.78 1.19 .86
TR3 5.28 1.35 .87
TR4 5.49 1.22 .90
Flow FL1 4.51 1.56 .63
.82 .61 FL2 5.48 1.17 .88
FL3 5.72 1.14 .82
Satisfaction SAT1 5.98 1.14 .93
.96 .89 SAT2 5.77 1.20 .95
SAT3 5.82 1.19 .95
Continuance
Intention CI1 6.38 1.09 .86
.89 .74 CI2 5.04 1.68 .83
CI3 5.97 1.24 .88

Appendices
Page 132 of 155
Appendix 7 – Scale R eliabilities – Study Two:

Construct Item Mean Std. deviation Loading s CR AVE
Flow FL1 4.80 1.59 .63
.85 .66 FL2 5.32 1.22 .91
FL3 5.51 1.25 .89
Impulsiveness IMP1 3.29 1.70 .91
.88 .71 IMP2 3.21 1.83 .87
IMP3 3.86 1.96 75
Self-Efficacy SE1 6.20 1.13 .77
.86 .66 SE2 5.92 1.20 .82
SE3 5.85 1.13 .85
Satisfaction SAT1 6.11 1.00 .81
.91 .78 SAT2 5.97 1.05 .90
SAT3 6.03 .99 .93
Continuance
Intention CI1 6.54 .63 .74
.81 .60 CI2 4.90 1.56 .69
CI3 6.00 1.10 .88

Appendices
Page 133 of 155
Appendix 8 – Normality of Residuals for Satisfaction – Study One:

Appendix 8A – Histogram

Appendices
Page 134 of 155
Appendix 8B – Q-Q Plot

Appendices
Page 135 of 155
Appendix 8C – Skewness and Kurtosis

Statistic Std. error
Unstandardized
residuals Mean -.00 .04
95% Conf. int. Lower bound
Upper bound -.09
.08
5% Trimmed mean .01
Median -.02
Variance .36
Std. deviation .60
Minimum -1.52
Maximum 1.50
Range 3.02
Interquartile Range .74
Skewness -.25 .18
Kurtosis .13 .35

Appendices
Page 136 of 155
Appendix 9 – Normality of Residuals for Continuance Intention – Study One:

Appendix 9 A – Histogram

Appendices
Page 137 of 155
Appendix 9B – Q-Q Plot

Appendices
Page 138 of 155
Appendix 9 C – Skewness and Kurtosis

Statistic Std. error
Unstandardized
residuals Mean -.00 .05
95% Conf . int. Lower bound
Upper bound -.1
.01
5% Trimmed mean -.00
Median .02
Variance .45
Std. deviation .67
Minimum -1.64
Maximum 2.01
Range 3.65
Interquartile Range .86
Skewness -.04 .18
Kurtosis .11 .35

Appendices
Page 139 of 155
Appendix 10 – Normality of Residuals for Continuance Intention – Study Two:

Appendix 10 A – Histogram

Appendices
Page 140 of 155
Appen dix 10B – Q-Q Plot

Appendices
Page 141 of 155
Appendix 10 C – Skewness and Kurtosis

Statistic Std. error
Unstandardized
residuals Mean .00 .06
95% Conf. int. Lower bound
Upper bound -.12
.12
5% Trimmed mean .01
Median .00
Variance .46
Std. deviation .68
Minimum -1.67
Maximum 1.85
Range 3.52
Interquartile Range 1.00
Skewness -.14 .22
Kurtosis -.35 .43

Appendix 11 – Homoscedasticity Analysis:

Satisfaction1 Continuance
Intention1 Continuance
Intention2
Sample size 187 187 125
Number of IVs 5 5 4
Breusch -Pagan test 11.50 25.70 7.02
Sig. level .04 .00 .14
Koenker test 10.84 24.72 4.90
Sig. level .06 .00 .29
1 Study one
2 Study two

Appendices
Page 142 of 155
Appendix 12 – Homoscedasticity -Consistent Regression Results:

Satisfaction Continuance Intention
Model fit .71 .66
Sig. level .00 .00

Homoscedasticity -Consistent Regression Results
B s.e.(hc) p B s.e.(hc) p
Confirmation .24 .06 .02 – – –
Perceived Ease of Use .20 .08 .00 .15 .10 .11
Perceived Usefulness .23 .07 .01 .29 .10 .00
Trust .31 .08 .00 .04 .08 .64
Flow .18 .08 .02 .24 .08 .00
Satisfaction – – – .36 .10 .00

Appendix 13 – Independence of Errors – Study One:

Satisfaction
Model R R² Adjusted R² Std. error of the
estimate Durbin -Watson
1 .85a .71 .71 .60 2.03
a. IVs: (Constant), CON, PEOU, PU, TR, FL
b. DV: SAT

Continuance Intention
Model R R² Adjusted R² Std. error of the
estimate Durbin -Watson
2 .81a .66 .65 .68 2.05
a. IVs: (Constant), SAT, PEOU, PU, TR, FL
b. DV: CI

Appendices
Page 143 of 155
Appendix 14 – Independence of Errors – Study Two:

Continuance Intention
Model R R² Adjusted R² s.e. Durbin -Watson
1 .62a .39 .37 .69 1.77
a. IVs: (Consta nt), SAT, FL, IMP, SE
b. DV: CI

Appendices
Page 144 of 155
Appendix 15 – Questionnaire – Study One:

Hello,

We are two master’s students enrolled at the cand.merc.mar program at Aarhus University, who
needs Your help to complete the following survey. The questionnaire is a part of our final thesis,
in which we wish to investigate people’s approach to M-Commerce*

*M-Commerce : Purchase of goods and services via a smartphone .

To get as accurately a result as possible, we ask You to answer the questionnaire as truthfully as
possible. Your answers will of course be confidential.

For every completed questionnaire received, an amount of 2 DKK will be donated to a good
cause. You will, at the end of the survey, get an opportunity to vote for the charity organization,
You think should receive the accumulated amount.

Thank You in advance,

Morten Riise Jensen &
Kasper Urbrand Nielsen

Appendices
Page 145 of 155
(1) Do you have a smartphone
(1)  Yes
(2)  No

2) Have you previously used your smartphone to purchase
goods or services?
(1)  Yes
(2)  No

(3) Gender
(1)  Male
(2)  Female

(4) Age
(1)  20 or younger
(2)  21-30
(3)  31-40
(4)  41-50
(5)  51 or older

(5) Highest completed education
(1)  Elementary school
(2)  Craftsman
(3)  High school
(4)  Professions bachelor
(5)  Bachelor
(7)  Master
(8)  Phd.
(9)  Other _____

(6) Primary occupation
(1)  Uemployed
(2)  Part time job
(3)  Tenure
(4)  Student
(5)  Self-employed
(7)  Other _____

(7) Approximately how many hours do you use your
smartphone each day?
(1)  Less than 1 hour
(2)  1 – 3 hours
(3)  3 – 5 hours
(4)  5 – 7 hours
(5)  More than 7 hours

Appendices
Page 146 of 155
8) Approximately for how long have you had experience with
M-Commerce?
(1)  Less than 1 year
(2)  1 – 2 years
(3)  2 – 3 years
(4)  3 – 4 years
(5)  4 – 5 years
(7)  5 – 6 years
(8)  More than 6 years

9) In the past six months, what have you purchased via M –
Commerce (You are able to select more answers)
(1)  Betting / casino
(2)  Mobile ticketing (train tickets, cinema tickets etc.)
(3)  Mobile travel ordering
(4)  Physical products (electronics, groceries, clothes etc.)
(5)  Software downloads (paid apps, paid software etc.)
(7)  Subscriptions (netflix, viaplay, magazines etc.)
(8)  Other _____

In the following section , please refer to your most recent M-Commerce experience when answering.

10) I find M -Commerce useful in my daily life
Highly disagree Highly agree
1  2  3  4  5  6  7 

11) Using my smartphone, I can manage transactions of goods and services quicker
Highly disagree Highly agree
1  2  3  4  5  6  7 

12) The options of M -Commerce enables me to accomplish things more conveniently
Highly disagree Highly agree
1  2  3  4  5  6  7 

13) The use of M -Commerce makes me more productive (Makes my everyday life easier)
Highly disagree Highly agree
1  2  3  4  5  6  7 

Appendices
Page 147 of 155
14) It was easy to me to learn M -Commerce
Highly disagree Highly agree
1  2  3  4  5  6  7 

15) M -Commerce is clear and understandable
Highly disagree Highly agree
1  2  3  4  5  6  7 

16) M -Commerce is easy to me
Highly disagree Highly agree
1  2  3  4  5  6  7 

17) I believe that becoming skillful at M -Commerce comes easy to me
Highly disagree Highly agree
1  2  3  4  5  6  7 

18) Based on my experiences with M -Commerce, I feel it’s reliable
Highly disagree Highly agree
1  2  3  4  5  6  7 

19) Based on my experiences with M -Commerce, I’m convinced that I’ll get what I’ve been promised
Highly disagree Highly agree
1  2  3  4  5  6  7 

20) Based on my experiences with M -Commerce, I feel that the vendors have the consumer’s interest at
heart
Highly disagree Highly agree
1  2  3  4  5  6  7 

21) Based on my experiences with M -Commerce, I feel it’s predictable (The vendors act as expected of
them)
Highly disagree Highly agree
1  2  3  4  5  6  7 

22) When using M -Commerce, my attention was focused on the activity
Highly disagree Highly agree
1  2  3  4  5  6  7 

23) When using M-Commerce, I felt in control
Highly disagree Highly agree
1  2  3  4  5  6  7 

Appendices
Page 148 of 155

24) When using M -Commerce, I found a lot of pleasure
Highly disagree Highly agree
1  2  3  4  5  6  7 

25) My experiences with M -Commerce was better than expected
Highly disagree Highly agree
1  2  3  4  5  6  7 

26) The degree of service when using M -Commerce was better than expected
Highly disagree Highly agree
1  2  3  4  5  6  7 

27) In general, my expectations with M-Commerce were confirmed
Highly disagree Highly agree
1  2  3  4  5  6  7 

28) I’m satisfied with my decision to use M -Commerce
Highly disagree Highly agree
1  2  3  4  5  6  7 

29) My decision to use M -Commerce was a wise one
Highly disagree Highly agree
1  2  3  4  5  6  7 

30) I feel, I made the right call when choosing to use M -Commerce
Highly disagree Highly agree
1  2  3  4  5  6  7 

31) I intend to use M -Commerce in the future
Highly disagree Highly agree
1  2  3  4  5  6  7 

32) I will, in the extent possible, always try to use M -Commerce in my everyday life
Highly disagree Highly agree
1  2  3  4  5  6  7 

33) I will continue using M -Commerce as regularly, as I do now
Highly disagree Highly agree
1  2  3  4  5  6  7 

Appendices
Page 149 of 155
34) Which charity organization do you think, should receive the
donation?
(1) Børnefonden
(2) Folkekirkens nødhjælp
(3) Hjerteforeningen
(4) Kræftens bekæmpelse
(5) PlanDanmark
(6) Red Barnet
(7) Røde Kors
(8) SOS Børnebyerne
(9) UNICEF
(10) WSPA – World Society of Protection of Animals
(11) WWF – Verdensnaturfonden
(12) Other _____

Thank you for your contribution
Best,
Morten & Kasper

Appendices
Page 150 of 155
Appendix 16 – Questionnaire – Study Two:

Hello,

We are two master’s students enrolled at the cand.merc.mar program at Aarhus
University, who needs Your help to complete the following survey.
The questionnaire is a part of our final thesis, in which we wish to i nves tigate people’s
approach to M-Commerce*

*M-Commerce : Purchase of goods and services via a smartphone.

To get as accurately a result as possible, we ask You to answer the questionnaire as
truthfully as possible. Your answers will of course be confidential.

For every completed questionnaire received, an amount of 2 DKK will be donated to a
good cause. You will, at the end of the survey, get an opportunity to vote for the charity
organization, You think should receive the accumulated amount.

Thank You in advance,

Morten Riise Jensen &
Kasper Urbrand Nielsen

Appendices
Page 151 of 155
(1) Do you have a smartphone
(1)  Yes
(2)  No

2) Within the past 6 months, have you used your smartphone
to purchase goods or services?
(1)  Yes
(2)  No

(3) Gender
(1)  Male
(2)  Female

(4) Age
(1)  20 or younger
(2)  21-30
(3)  31-40
(4)  41-50
(5)  51 or older

(5) Highest completed education
(1)  Elementary school
(2)  Craftsman
(3)  High school
(4)  Professions bachelor
(5)  Bachelor
(7)  Master
(8)  Phd.
(9)  Other _____

(6) Primary occupation
(1)  Uemployed
(2)  Part time job
(3)  Tenure
(4)  Student
(5)  Self-employed
(7)  Other _____

7) In the past six months, what have you purchased via M –
Commerce (You are able to select more answers)
(1)  Betting / casino
(2)  Mobile ticketing (train tickets, cinema tickets etc.)
(3)  Mobile travel ordering
(4)  Physical products (electronics, groceries, clothes etc.)
(5)  Software downloads (paid apps, paid software etc.)
(7)  Subscriptions (netflix, viaplay, magazines etc.)
(8)  Other _____

Appendices
Page 152 of 155
In the following section, please refer to your most recent M-Commerce experience when answering.

8) When using M -Commerce, my attention was focused on the activity
Highly disagree Highly agree
1  2  3  4  5  6  7 

9) When using M -Commerce, I felt in control
Highly disagree Highly agree
1  2  3  4  5  6  7 

10) When using M -Commerce, I found a lot of pleasure
Highly disagree Highly agree
1  2  3  4  5  6  7 

11) I often buy things without thinking
Highly disagree Highly agree
1  2  3  4  5  6  7 

12) ‘I see it, I buy it’ describes me
Highly disagree Highly agree
1  2  3  4  5  6  7 

13) Sometimes I feel like buying things on the spur -of-the-moment
Highly disagree Highly agree
1  2  3  4  5  6  7 

14) I would feel confident that I can use m -commerce.
Highly disagree Highly agree
1  2  3  4  5  6  7 

15) I expect to become proficient in using m -commerce
Highly disagree Highly agree
1  2  3  4  5  6  7 

16) I would be able to use m -commerce even if there was no one around to show me how to use it
Highly disagree Highly agree
1  2  3  4  5  6  7 

Appendices
Page 153 of 155
17) I’m satisfied with my decision to use M -Commerce
Highly disagree Highly agree
1  2  3  4  5  6  7 

18) My decision to use M -Commerce was a wise one
Highly disagree Highly agree
1  2  3  4  5  6  7 

19) I feel, I made the right call when choosing to use M -Commerce
Highly disagree Highly agree
1  2  3  4  5  6  7 

20) I intend to use M -Commerce in the future
Highly disagree Highly agree
1  2  3  4  5  6  7 

21) I will, in the extent possible, always try to use M -Commerce in my everyday life
Highly disagree Highly agree
1  2  3  4  5  6  7 

22) I will continue using M -Commerce as regularly, as I do now
Highly disagree Highly agree
1  2  3  4  5  6  7 

23) Which charity organization do you think, should receive the
donation?
(1) Børnefonden
(2) Folkekirkens nødhjælp
(3) Hjerteforeningen
(4) Kræftens bekæmpelse
(5) PlanDanmark
(6) Red Barnet
(7) Røde Kors
(8) SOS Børnebyerne
(9) UNICEF
(10) WSPA – World Society of Protection of Animals
(11) WWF – Verdensnaturfonden
(12) Other _____

Thank you for your contribution
Best,
Morten & Kasper

Appendices
Page 154 of 155
Appendix 16 – Donation to The Danish Cancer Society:

Appendix 16a: Charity vote

020406080100120140160180200Charity Donation Votes

Appendices
Page 155 of 155

Appendix 16b: Kasper’s donation

Appendix 16c: Morten’s donation