MORE-THAN-VOICE USE OF MOBILE AT THE BOTTOM OF THE PYRAMID:
ANALYSIS OF MOTIVATIONAL AND CONTEXTUAL DRIVERS
TO MOBILE USE AMONG LOW-INCOME USERS IN SOUTH ASIA

By
Juhee Kang

A DISSERTATION
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of
Media and Information Studies - Doctor of Philosophy
2014

ABSTRACT
MORE-THAN-VOICE USE OF MOBILE AT THE BOTTOM OF THE PYRAMID:
ANALYSIS OF MOTIVATIONAL AND CONTEXTUAL DRIVERS
TO MOBILE USE AMONG LOW-INCOME USERS IN SOUTH ASIA
By
Juhee Kang

Over the last decade, mobile communication has become increasingly available,
affordable and accessible even to the poor and disadvantaged at the Bottom of the Pyramid
(BOP) in developing countries. This unprecedented connectivity at the BOP introduces not only
an untapped group of media users for communication researchers, but also a new hope among
development practitioners of fostering social change through innovative mobile-based
intervention services. Despite the mounting interest, however, little is known about how
individuals at the BOP adopt and use mobile phones.
This dissertation investigates the factors influencing mobile use behaviors at the BOP in
South Asia, with a particular interest in their motivational drives behind the utilization of mobile
phones for services other than voice calls. Guided by the well-established technology adoption
theories, including the Theory of Planned Behavior (TPB) and Technology Acceptance Model
(TAM), the dissertation examines the cognitive process behind mobile use involving
technological utility (perceived usefulness), social influence (subjective norms) and contextual
conditions (perceived behavioral control). The study also extends the model with multiple
antecedents addressing ‘what makes a mobile useful’ and ‘what contributes to the enabling
conditions’ pertinent to the context of the BOP users. Furthermore, pointing out the limitations of

the adoption theories in their tendency to overlook socio-demographic effects, the dissertation
examines whether and where such demographic effects intervene in the motivational process.
Based on a large-scale multi-country random survey conducted in South Asia by
LIRNEasia (N=4,023), the dissertation empirically validates the proposed model using Structural
Equation Modeling (SEM). It also takes a novel approach to define a multi-dimensional effect of
socio-demographic factors using a two-step cluster analysis method and compares the path
differences between the sub-groups. In addition, it provides a supplementary analysis to explore
the moderating role of mobile efficacy in the actualization of the behavioral intention to mobile
use behavior.
The dissertation finds that, first, the western-oriented technology adoption theories
successfully explain the formation of behavioral intentions (BI), but they fail to explain the full
process of the actualization of behavioral intentions that lead to more-than-voice mobile use
among the BOP owners. Second, socio-demographic factors partially moderate the degree of
different effects of the motivational factors, indicating that the poor is not a homogenous mass
and there exist considerable differences in their cognitive evaluations relating to their sociodemographic conditions. Third, in exploring the additional factors affecting the behavioral
intentions to behavior path, the study suggests a tentative finding on the interaction effect of
mobile efficacy and behavioral intention on the actualization of BI among the BOP mobile
owners.

To my mother

iv

ACKNOWLEDGEMENTS

I would never have been able to complete this dissertation without the support from the
following people. Words cannot describe how much I am indebted to them.
I would like to express my deepest gratitude to my advisor, Dr. Johannes M. Bauer, for
his excellent guidance, unconditional support and generous encouragement. In many occasions,
his advice rescued me from confusion and frustration, both academically and personally, and
helped me establish a foundation to become a good researcher.
I would also like to thank my dissertation committee members, including Dr. Mark Levy
who inspired me much by his vigorous and unrelenting passion for communications and ICT4D
research, Dr. John Kerr who always offered me insightful and eye-opening advice on
international development research, Dr. Constantinos Cousaris who spent his valuable time
training me with research practices in Information Systems research, and Dr. Folu Ogundimu
who provided warm encouragement with advice on research in developing countries.
My sincere appreciation goes to Dr. Rohan Samarajiva and friends at LIRNEasia who
made this dissertation possible by allowing me to access their pioneering survey project and
helping me experience the practical issues beyond textbooks. I also want to thank Dr. Charles
Steinfield for enabling me to continue my study with financial support, Dr. Kurt DeMaagd for
offering me an invaluable experience of the ICT4D project in Tanzania, and Dr. Moutusy Maity
and Dr. Roger Calantone for training me with advanced statistical skills. I also thank ICT4D
practitioners – Dr. Kentaro Toyama, Mr. Sameer Sharma and Dr. Eun-Ju Kim at the ITU, whose
email conversations helped me balance academic rigor and practical relevance. Finally, I would
like to thank my family who always stood by my side through the good times and the bad.
Thank you all.
v

TABLE OF CONTENTS

LIST OF TABLES ....................................................................................................................... viii
LIST OF FIGURES ....................................................................................................................... xi
INTRODUCTION ...........................................................................................................................1
CHAPTER 1: RESEARCH CONTEXTS .......................................................................................6
1.1. Mobile Explosion in Developing Countries ..........................................................................6
1.2. ICTs and Mobile for Development ........................................................................................9
1.3. Understanding Mobile Use at the BOP ................................................................................13
1.4. Research Gaps in the Current M4D Studies ........................................................................16
1.4.1. Need for Quantitative Social Science Approaches ........................................................16
1.4.2. Conceptual Shift on the Poor .........................................................................................17
1.4.3. Exploring Motivations within Socio-demographic Constraints ....................................18
1.4.4. More-than-voice Use of Mobile.....................................................................................20
CHAPTER 2: THEORETICAL FRAMEWORK ..........................................................................23
2.1. Review of Technology Adoption Theories ..........................................................................23
2.1.1. Brief Summary of Technology Adoption Theories .......................................................24
2.1.2. Comparison of the Adoption Theories ...........................................................................28
2.2. Theoretical Strengths of the TPB .........................................................................................30
2.3. Proposed Model Extension and Hypotheses ........................................................................33
2.3.1. TPB Framework with Perceived Usefulness .................................................................34
2.3.2. Exploring the Antecedents to PU and PBC ...................................................................37
2.3.3. The Effect of Socio-demographic Factors .....................................................................43
CHAPTER 3: RESEARCH METHOD .........................................................................................46
3.1. Data ......................................................................................................................................46
3.2. Measures ..............................................................................................................................54
3.3. Data Preparation...................................................................................................................57
3.4. Reliability and Validity ........................................................................................................58
CHAPTER 4: ANALYSES ...........................................................................................................61
4.1. Path Analysis ......................................................................................................................61
4.2. Multi-Group Comparison....................................................................................................63
4.3. Supplementary Analysis: Effects of Behavioral Intention on Behavior .............................68
CHAPTER 5: DISCUSSION .........................................................................................................76
5.1. Key Findings ........................................................................................................................76
5.2. Research Implications ..........................................................................................................80
5.3. Limitations and Further Research ........................................................................................84
5.4. Concluding Remarks ............................................................................................................85

vi

APPENDIX ....................................................................................................................................87
BIBLIOGRAPHY ........................................................................................................................103

vii

LIST OF TABLES

Table 1: Country Profiles ............................................................................................................ 47
Table 2: Sample SEC Criteria for India ...................................................................................... 48
Table 3: Sample Characteristics.................................................................................................. 51
Table 4: Ranking of the Most Used More-than-Voice Mobile Features .................................... 53
Table 5: Measurement Items ....................................................................................................... 55
Table 6: Exploratory Factor Analysis on Perceived Benefits of Mobile .................................... 56
Table 7: Standardized Factor Loadings ...................................................................................... 59
Table 8: Construct Statistics ....................................................................................................... 59
Table 9: Factor Correlation Matrix and Discriminant Validity Assessment .............................. 60
Table 10: The Path Coefficients and R2 Difference among Sub-groups ..................................... 66
Table 11: Summary of Dependent Variables .............................................................................. 70
Table 12: Pearson’s Product Moment Correlation among Dependent Variables ....................... 72
Table 13: Model Testing with Multiple Mobile Use Measures .................................................. 72
Table 14: Mobile Efficacy Measures and Item Statistics .......................................................... 74
Table 15: Effects of Intention, Mobile Efficacy and Interaction on Mobile Use ....................... 75
Table 16: Summary of Hypotheses Testing ................................................................................ 77

viii

LIST OF FIGURES

Figure 1: Structural Changes behind Mobile Adoption in Developing Countries...........................7
Figure 2: Multi-Stage Process of Mobile for Development (M4D) ..............................................14
Figure 3: The Evolution of Technology Adoption Theories .........................................................29
Figure 4: Theory of Planned Behavior...........................................................................................31
Figure 5: Proposed Research Model ..............................................................................................34
Figure 6: Structural Model Testing ................................................................................................62
Figure 7: Four Sub-groups from the Cluster Analysis ...................................................................65
Figure 8: Path Comparison 1 - Husband vs. Housewife Group .....................................................67
Figure 9: Path comparison 2 - Husband vs. the Poorest Group .....................................................68
Figure 10: The Attributes of Innovation ........................................................................................89
Figure 11: The TRA and the TPB ..................................................................................................92
Figure 12: The Original TAM........................................................................................................95
Figure 13: TAM2 Extension ..........................................................................................................96
Figure 14: The Unified Theory of Acceptance and Use of Technology (UTAUT) ......................97
Figure 15: The UTAUT2 ...............................................................................................................98

ix

INTRODUCTION

By 2013, the number of global mobile subscriptions exceeded 6.8 billion driven primarily
by double-digit annual growth in developing countries (ITU, 2013). Due to increasing network
coverage, low-cost handsets and falling service prices, mobile telephony has become widely
available, affordable and accessible even to the poor and disadvantaged at the Bottom of the
Pyramid1 (BOP) (ITU, 2011). Nowadays, it is no longer surprising to find a mobile phone in the
hand of rural subsistence farmers in India or slum dwellers in African cities, who were
previously unable or unwilling to join other communication services (Aileen et al., 2011).
The BOP, as proposed by Prahalad (2004), refers to the poorest socioeconomic group in
the global economy which includes the 4 billion people living on less than USD 2.50 per day,
mostly from low- and middle-income nations in Africa, South Asia and Latin America. The
concept of the BOP is not just a new reference term for populations below the poverty line, but
rather it calls for a shift in our understanding of the poor. Prahalad (2004) perceives the poor as
active and sense-making consumers within their economic and social constraints, who have been
largely neglected by the private sector. Indeed, mobile phones for the poor were never offered by
international aids. Rather, the high penetration of mobile ownership among the poor was enabled
by the private sector, that is, mobile operators who acknowledged the business potential at the
BOP together with the poor’s readiness to own a mobile phone as individual consumers. More
and more people at the BOP consider having their own mobile phones, evaluate the cost and
benefits, and make their own decisions to purchase their first personal electronic communication
device. They now constitute a new group of consumers whose preferences are yet to be defined

1

Recently the term the ‘base’ of the pyramid is preferred by some but, as Prahalad notes, the ‘bottom’ is a more
accurate description of their destitute reality (Prahalad, 2010).

1

as well as new members of media users whose behaviors and related social changes are to be
explored by communication researchers.
At the same time, the increasing connectivity among the BOP also invites new hope
among the international development communities that a variety of innovative services can be
delivered directly to the poor via mobile phones. Encouraged by the prevalence of mobile phones,
more and more researchers and development practitioners are interested in finding out the
potential of mobile phones for social changes in developing countries. Consequently, initiatives
using mobile for development (M4D) are proliferating across the world (InfoDev, 2003, 2013;
UNDP, 2012; GSMA, 2012; Donner, 2010). Indeed, the issues relating to mobile
communications in developing countries are now shifting from how to provide access to what
potential services can be delivered over the newly introduced mobile connectivity and how the
BOP users adopt and utilize the potentials of mobile phones.
Nevertheless, little is known about how individuals at the BOP adopt and use mobile
phones and how such behaviors yield developmental outcomes. Several critics point out that the
access to information communications technologies (ICT) does not automatically lead to
socioeconomic development and the evidence documenting tangible impacts are scarce
(Toyama, 2011; Chaudhuri, 2012; Heeks, 2002; 2012; Unwin, 2009; Blattman et al., 2003;
Thomas & Prayil, 2008; Kleine, 2010). As discussed in chapter 2, the past approach to ICTs for
Development (ICTD) was concerned mostly with providing access to technologies, often
assuming without critical exploration a two-step process where the provision of technologies
(input) leads to developmental changes (output). Such an assumption may no longer be
sustainable to explain the growing number of observations documenting unsuccessful,
unintended and unclear outcomes of ICTD projects. The field seems to be in need of a new

2

framework to explain the complex nature of the development process via the use of ICTs. Hence,
the dissertation suggests that it is necessary to deconstruct the ICTD and M4D process into
multiple stages and take a step-by-step approach to examine each stage. In doing so, it argues
that it is important to fully understand the poor’s mobile use behaviors and what factors
influence or hinder such behaviors prior to measuring the impacts of ICTD.
Research attempting to understand mobile use behaviors at the BOP is relatively sparse.
Most studies tend to take ethnographic or qualitative approaches focusing on their unique
behaviors or project success stories (Horst & Miller, 2005, 2006; Sey, 2011; Burrell 2010;
Cartier, Castells, and Qui, 2005; Molony, 2006). Analyses of a handful of large-scale surveys on
the BOP mobile users have typically been limited to mere description of use patterns or
demographic comparisons. In fact, far too little attention has been paid to the motivational
drivers or needs behind their behaviors (Heeks, 2012; Banerjee & Duflo, 2011).
This dissertation investigates the factors affecting mobile2 use behaviors among owners
of mobile phones at the BOP in South Asia. It asks what motivational factors drive the BOP
owner’s decisions to use mobiles, with a particular interest in the utilization of mobile phones for
more than making or receiving voice calls. It also attempts to explore how such motivational
factors interact with the user’s contextual and socio-demographic factors. In doing so, it
conceptualizes the poor as rational decision-makers with heterogeneous preferences and
behavioral motivations, contrasting the conventional notion of the poor as a vulnerable mass only
classified by their income.

2

In this dissertation, the term ‘mobile’, ‘mobile phone’, ‘mobile telephony’ are used interchangeably referring not
only to hardware (i.e. handset) but also to related software and services offered on mobile phones such as voice call,
SMS, applications, information services, etc.

3

Extending on the well-known technology adoption theories such as the Theory of
Planned Behavior (TPB) and Technology Acceptance Model (TAM), the dissertation first
examines the fundamental drivers behind mobile use, including technological utilities (perceived
usefulness, PU), social influence (subjective norms, SN) and enabling contextual factors
(perceived behavioral control, PBC). Second, to expand the theories’ ability to draw more
practical implications relevant to the M4D practitioners (Bagozzi, 2007; Benbast & Barki, 2007),
it explores the antecedents to PU and PBC to address ‘what makes a mobile phone useful’ and
‘what contributes to the enabling contexts’ among the BOP mobile owners. Third, the
dissertation points out the limitations of such adoption theories in their tendency to overlook
socio-demographic effects and attempts to investigate whether and where such demographic
effects intervene in the motivational process.
Using a large-scale multi-country random sample (N=4,023), the dissertation provides
empirical evidence on the motivational drivers behind more-than-voice use. It first tests the
proposed model simultaneously with structural equation modeling (SEM) technique. It also takes
a novel approach to define a multi-dimensional effect of socio-demographic factors using a
cluster analysis method to identify four sub-demographic groups. The study analyzes the effects
of socio-demographics by comparing the model coefficients between different sub-groups.
The structure of this dissertation is as follows: chapter 1 reviews the research contexts in
relation to the structural changes fostering mobile adoption in developing countries. It reviews
increasing M4D initiatives and discusses the research gaps in the current mobile use studies as
well as the rationales for the current study. Chapter 2 introduces the theoretical framework and
hypothesis development based on the reviews on the technology adoption theories as well as
discussing the strength and limitations of the TPB. Chapter 3 explains the methodological

4

approach and the data, followed by empirical results in chapter 4. Chapter 5 discusses the
findings, research implications, limitations, and possible directions for future research.

5

CHAPTER 1: RESEARCH CONTEXTS

1.1. Mobile Explosion in Developing Countries
Mobile communication has experienced exponential growth in developing countries.
Between 2000 and 2010, mobile penetration in low- and middle-income countries surged from a
mere 4 per 100 inhabitants to 72 per 100, exhibiting an outstanding total increase of 1,500
percent (World Bank, 2012). The diffusion of mobile telephony in these nations is unprecedented
not only for its speed and volume but also for its breadth, reaching to diverse populations
(Chaudhuri, 2012; Kalba, 2008).
Such rapid uptake of mobile telephony in Africa, Asia, Middle East and Latin America
was not foreseen by many telecommunication scholars who observed complex barriers in fixed
telephony provision in developing countries (Sey, 2008). Guttman (1986) once predicted that, at
the growth pace of fixed telephony in 1985, Sub-Saharan Africa would have to wait until 2027 to
achieve a teledensity of 1 per 100 habitants. He pointed out there exist several fundamental
problems inherent in the African telecommunication sector such as over-regulation by central
government, low finance and under-investment, a lack of human capacity and large rural areas.
While fixed-telephony and Internet penetration still exhibit sluggish growth in these
regions, mobile telephony made a breakthrough in developing countries and became available,
affordable and accessible to the majority of the population within the last decade. In fact, how
this mobile revolution occurred cannot be explained by pointing to a single cause. Prior to
individuals’ demand for mobile, there existed several structural changes and innovations from
the supply-side involving multiple entities across technology, regulation and business (see Figure
1).
6

Figure 1: Structural Changes behind Mobile Adoption in Developing Countries

First, in the realm of technology, the commercialization of digitized cellular network
technologies in the 1990s provided a cost-efficient means to cover wide rural and sparsely
populated areas. Together with mobile phone’s unique characteristics of mobility, alwaysconnected and personal device, mobile telephony emerged as a viable new business opportunity
in developing countries.
Second, in the realm of regulation, a wave of telecom reforms since the late 1990s has
transformed the telecom sector across the world. These changes transformed the telecom sector
in many developing countries from state-owned monopolies to a privatized, liberalized and
independently-regulated marketplace (Melody, 1999). In case of the fixed-line market, however,
incumbent operators often continue to dominate the market even after privatization, by
maintaining close links with government or hampering competition with their dominant power.
In contrast, the timely introduction of mobile telephony in the midst of telecom reform allowed
the effective implementation of the improved regulation and an open and competitive market
environment in the mobile services. Important aspects include the lowering of entry barriers for
new operators, encouraging competition among multiple operators, and the creation of a dynamic
private sector with less state interference (Bauer, 2010). These reformed regulatory and market
7

environments became essential pre-conditions for declining prices, expanding network coverage
and increasing the user base (ITU, 2011; GSMA, 2006).
Third, the liberalized market also opened a door for foreign mobile operators who were
seeking opportunities outside the saturated western markets. Several European mobile operators
(e.g. Vodafone, Orange, Telefónica) and newly emerged multinational operators from the Global
South (e.g. MTN and Bharti Airtel) aggressively expanded their business across Africa and Asia.
At the same time, handset manufacturers also came up with low-cost handsets suitable for
emerging markets in response to the need of multinational mobile operators while feeling similar
pressure to move beyond the saturated Western handset market. Spurred by Nokia, which
pioneered a strategy for affordable low-cost phones since 2002 (Nokia Press Release, 2008), the
industry quickly expanded its battleground of competition to under USD 100 low-cost phones.
The fierce competition between major manufactures sparked various innovations relating to cost
reduction involved in product design, manufacturing and distribution, and eventually lowered the
cost substantially below USD 30 (Information Week, 2005).
Fourth, mobile operators in developing countries also came up with innovative business
strategies, such as the so-called ‘budget telecom network model’ (Samarajiva, 2010). These
budget telecom operators tried to create economies of scale by reducing the transactional cost in
handling a large number of customers who typically generate miniscule ARPU (Average
Revenue per User). They offered pre-paid plans and calling-party-pays systems which enabled
better control over expenditure for the low-income users, and widened the customer base via
simplified registration procedures and minimum top-up fees (Samarajiva, 2010; Barrantes &
Galperin, 2008; Sey, 2010; ITU, 2011). The budget operators also reduced the capital cost by
maximizing network capacity by upgrading the existing base stations with software solutions

8

instead of investing in building new network infrastructure (ITU, 2011). To reduce the
operational cost, they also substituted the distribution channels with pre-existing local kiosks or
informal dealers across the country (ITU, 2011).
As a consequence of the interrelated innovations from multiple stakeholders, mobile
communication became more available with the expanding network covering over 75 percent of
the population in developing countries (ITU, 2012). It also became more affordable with the
falling cost of handsets and voice calls. The total cost of ownership (TCO) of mobile service in
77 developing countries continued to decrease from an average annual cost of USD 13.16 in
2007 to USD 10.88 in 2009, with 12 countries offering the service for less than USD 5 (Nokia
Research, 2007, 2009; ITU, 2011). Therefore, as connectivity via mobile continues to increase,
the issues relating to mobile communication in developing countries are now gradually shifting
from ‘how to provide access’ to ‘what services can be delivered’ over the newly introduced
connectivity.

1.2. ICTs and Mobile for Development
The newly introduced connectivity at the BOP invites a new hope that a variety of
innovative services can be delivered directly to the hands of the poor. Mobile for Development
(M4D) refers to the growing research and practices to utilize mobile services in a way to bring
positive impacts on the lives of the poor (InfoDev 2012, GSMA 2011; Duncombe, 2012; Kumar
& Svensson, 2012; Donner, 2010).
Indeed, many development agencies and NGOs are exploring the potential of mobile
phones as a cost-effective platform to carry development services such as education, healthcare,

9

financial, and agricultural programs. Spurred by evidence of the mobile phone’s positive impact
on economic activities (Jensen, 2007; Abraham 2007; Aker, 2008), and successful cases like
mobile banking service in Kenya (Hughes & Lonie, 2007; Morawczynski, 2009), mobile phones
are increasingly perceived as a smart platform for existing intervention programs. Numerous
M4D projects (i.e. m-health, m-education, m-government, m-banking, m-agriculture, memployment etc.) are currently in trial across the world (UNDP, 2012; IFC 2011; GSMA
mWomen, 2012). As of May 2013, there exists over 1,320 live M4D projects and services are
offered across the world (GSMA Intelligence, 2013). For instance, mobile phones are used for
disseminating agricultural tips and market price information to rural farmers, connecting
healthcare workers and patients in remote areas, providing learning materials to children, linking
up buyers and sellers or employers and job seekers, and so forth.
In its origin, M4D can be seen as a branch of a broader movement called ‘Information
and Communication Technology for Development’ (ICTD) which explores how new media
technologies, including but not limited to computers, Internet and mobile phones, can facilitate
socioeconomic development of the disadvantaged communities across the world (Toyama &
Donner, 2008; Unwin, 2009; Heeks 2008, 2010). While its target technology has shifted from
computing systems to mobile phones in recent years, the M4D activities share much of the basic
rationales behind ICTD.
The proponents of ICTD typically address three rationales: first, information and
communications technologies possess intrinsic power that can foster innovation, efficiency and
empowerment in developing societies. Thus providing ICTs in those communities can spur the
process of development and poverty reduction. Second, digital ICT networks introduce a new
form of capital via boundless dissemination of information and knowledge as well as

10

opportunities for networking and communications (Castells, 1996). Consequently the gap
between those who are connected and those who are disconnected and neglected is increasingly
widening. Hence, it is critical to provide the poor with access to ICTs in order to embrace the
disconnected in the march towards the inclusive information society (WSIS, 2005; ITU 2011,
2012). Third, ICTs appear to be a new means to improve the designs and operations of existing
international development programs. It is increasingly acknowledged that mobile applications
and computer software can be used to enhance cost-effectiveness of development project
operation while creating participatory routes for the poor to engage with such programs (UNDP,
2012).
In fact, such rationales are founded on an implicit assumption that access to technologies
is bound to create positive impacts and that more access results in more economic and social
gains for the poor. Under this assumption on the direct impact of ICT on development, studies
found that the ICTs, particularly mobile phones, have positive impacts on national GDP growth
in developing countries (Waverman et al., 2005; Qiang et al., 2009; Sridhar & Sridhar, 2006). It
can also enhance market efficiency by resolving the problems of asymmetrical information and
price dispersion commonly found in developing economies (Jensen, 2007; Abraham, 2007; Aker,
2008). For small and micro- entrepreneurs (SMEs), ICTs are also found to contribute to
increased labor productivity (Esselaar et al., 2007) and business growth (Chew et al., 2011).
Nevertheless, research on the impacts of ICT on development is still in a nascent stage.
The conceptual and empirical linkage between ICT use and developmental impacts are currently
ambiguous as the findings to date are inconclusive and even contradictory. In fact, there are only
a handful of studies showing evidence that ICTD projects yield tangible and replicable success in
developmental outcomes (Heeks, 2002, 2010; Chaudhuri, 2012). Most impact studies are

11

typically anecdotal case studies that are too specific to be generalized (Sey & Fellow, 2011;
Gomez et al., 2012). On the other hand, several researchers noted that many ICTD projects have
been unsuccessful in bringing desired impacts (Heeks, 2002; Warschauer, 2004; Kuriyan et al.,
2006; Best & Kumar, 2008; Trucano, 2009; Jackson et al., 2011). In particular, some scholars
point out that these failures are often caused by the mismatch between the intended benefits and
local needs and the unexpected barriers that ICTs cannot overcome. Blattman, Jensen, and
Roman (2003) found that mere provision of market information over ICT had limited impact on
improving the livelihoods of rural farmers due to several structural and cultural hindrances. They
suggested that ‘information is a necessary but not sufficient condition for development’ and the
intended benefits can often obscured by local complexity. Thomas and Parayil (2008) also
suggested that social development can only be achieved via broader social policy intervention,
concluding that “providing access to ICTs through rural kiosks alone will not bring about
development and change”.
Furthermore, the unintended or even counteractive impacts of ICTs on development are
also discussed. Recent studies find that popular uses of Internet in developing countries tend to
be entertainment or recreational activities in preference to much hoped informational or business
uses (Sey & Fellows, 2009). Similarly, mobiles are mostly used for talking with friends and
family rather than business or economically productive uses (Rashid & Elder, 2009; Donner
2006). The adverse effects of ICT provision were also noted even long before the ICTD
movement started. Rogers (1995) has rightly conceptualized it as an ‘innovation paradox’ where
the ‘individuals who might need the benefits of innovation are generally the last to adopt an
innovation’. In other words, the benefits gained from the ICT access tend to reach first to those
who have resources to afford and utilize ICTs and the gap between the ICT-haves and have-nots

12

may be worsened. Recent studies indeed reveal that Internet cafes or telecenter users in
developing countries tend to be younger males with relatively higher education and income,
which indicates the emerging digital divide within developing countries (Sey & Fellows, 2009;
Rangaswamy, 2008; Furuholt et al., 2007).
To sum, the early approach to M4D was based on a two-stage model assuming mobile
access (input) will lead directly to socioeconomic development (output). The inconclusive
findings on the impact of ICTs on development suggest that the underlying assumption of the
two-stage linear model may have been too limited to explain the actual process between ICT
uses and development. The findings on ICTD failures and unintended consequences also suggest
that the relationship between ICTs and development are far more complex involving multiple
paths linking multiple factors.

1.3. Understanding Mobile Use at the BOP
The conceptual framework of M4D needs to be further specified in order to capture the
complex path from mobile access to developmental outcomes. In particular, the process needs to
be deconstructed into multiple stages including users’ m-service adoption decision, use behaviors,
behavioral changes and related developmental impacts (Figure 2). In this regard, understanding
people’s mobile use behavior can hold a missing link between access and development. The
current M4D research is in need of more studies examining how people actually adopt and use
mobile phones and available m-services in their everyday life and how such uses affect
behavioral changes.

13

Figure 2: Multi-Stage Process of Mobile for Development (M4D)

‘Mobile use’ is a dynamic and multi-faceted concept which changes through constant
interactions with user’s different needs, motivations and social and contextual influences. Over
the last decade, mobile use behaviors in advanced countries have been studied extensively in
relation to service adoption (Rice & Katz, 2003; Teo & Pok, 2003; Nysveen, 2005; Kim, 2008;
Cho, 2011), gratifications sought (Leung & Wei, 2000; Pedersen, 2005; Peters, 2009),
domestication process (Haddon, 2001; Ling, 2005; Ito, 2005; Selwyn, 2003) and appropriation
behaviors (Carroll et al., 2002; Bar et al., 2007; Wirth et al., 2008). Nevertheless, relatively little
attention has been paid to mobile use in developing countries, particularly among the
disadvantaged groups.
James & Versteeg (2007) point out that a considerable discrepancy exists between what
we can interpret from the global statistics on mobile penetration and what actually happens in
Africa due to the unique behaviors of the local mobile users. These behaviors emerge as part of
the survival strategies of the ‘information have-less’ to avoid the cost burden (Cartier, Castells,
14

and Qui, 2005; Zainudeen et al., 2006). For instance, studies found that sharing mobile phones is
a common practice in many African and Asian countries (Burrell, 2010; Sey, 2009; LIRNEasia,
2009). Some people carry only a SIM card without handset while others own multiple SIM cards
to swap networks for cheaper prices or better coverage (InfoDev, 2013; LIRNEasia, 2012). Users
in developing countries also employ inventive practices such as ‘beeping’ which users leave
intentional missed-call signs that can be decoded into pre-arranged messages such as ‘call me
back’ or ‘I’m doing fine’ (Donner, 2007).
Several studies also reveal ambiguous relationships between mobile use and economic
development. Molony (2007) finds that Tanzanian traders prefer face-to-face conversations over
mobile use when discussing business due to the lack of trust on mobile phones. It was also
suggested that mobile phones in these regions are mostly used for chatting with friends and
family rather than expanding business relationships (Donner, 2006; Sey, 2011; Chew et al., 2011;
Rashid & Elder, 2009; Souters et al., 2006). They caution that it is uncertain whether such
personal uses are relevant to economic benefits as the boundary between personal and business
or economic uses often blurs in the livelihoods of the poor (Sey, 2011; Donner, 2006). Horst &
Miller (2006) also observe how Jamaicans use mobile phones for ‘link-up’, a short exchange of
greetings to maintain social networks which can become economic support networks when
necessary. In addition, there exist several studies exploring how mobile phones are used among
social minority groups in developing countries, such as among migrants (Madianou & Miller,
2011; Wallis, 2011; Qui, 2008; Lin & Tong, 2008) or political groups (Rheingold, 2002).
Overall, the studies exploring mobile use behaviors in developing countries are still
limited not only in their quantity but also in their epistemological and methodological diversity.

15

Hence, the following section discusses four research gaps in the current mobile for development
studies and proposes rationales behind this dissertation research.

1.4. Research Gaps in the Current M4D Studies
The dissertation proposes the following four areas which call for further attention in the
current mobile for development research: (a) the need for more quantitative social science
approaches, (b) need for shifting the perspective on the poor as individual decision-makers, (c)
need for exploring user motivations in the context of socio-demographic constraints, and (d) the
need for studying more-than-voice use of mobile at the bottom of the pyramid.
1.4.1. Need for Quantitative Social Science Approaches: So far, most studies on the poor’s
mobile use tend to employ qualitative or ethnographic approaches (Gomez et al., 2012; Patra et
al., 2009). While these qualitative studies offer richer details of mobile use that often reveal
emergent behaviors specific to the poor’s use contexts, the collection of findings often tend to be
anecdotal and scattered across specific research sites. One risk of the limited diversity of
methodological approaches can be that, as the research domain is relatively nascent, the M4D
researchers in general are thirsty for empirical evidence. Accordingly, a limited quantity of
empirical studies in the field may lead to the undesirable situation where the findings conditioned
to a specific small sample are over-stretched to describe broader populations in different contexts.
In other words, studies based on a small sample of Tanzanian rural farmers may be
misappropriated to understand Tanzanian urban students or over-extended to explain Tanzanians,
Sub-Saharan Africans or even the African poor in general.
On the other hand, there exist only a few studies employing quantitative methodologies
but the analyses mostly remain at a descriptive level. Some of these studies are based on large16

scale industry surveys reporting the poor’s mobile access, ownership, usage patterns and related
demographic attributes (Souters et al., 2007; LIRNEasia, 2009, 2012; GSMA mWomen, 2011;
InfoDev, 2012, 2013). Although these studies provide useful information to understand the
current status of mobile use at the BOP, the results are often analyzed in mere counting numbers
of users and non-users while the associations beyond demographic attributes are rarely explored.
In this regard, quantitative social science approaches can provide a good guideline for
understanding mobile use behaviors in the Global South. Studies with a representative random
sampling and rigorous analysis can broaden the generalizablity of research findings to address
overall behavioral patterns across countries or regions. It can also theorize causal relationships or
valid associations among scattered motivational and behavioral tendencies to explain why and
how these behaviors occur and to predict future behaviors.
1.4.2. Conceptual Shift on the Poor: A conventional approach to define the poor is to
categorize them by their income level such as those below the poverty line of USD 2 a day or the
extreme poverty line of USD 1.25 a day (World Bank, 2011). Alternatively, the poor in
developing countries are often perceived as an amorphous mass labeled by geographic regions
(e.g. Sub-Saharan Africans) or location of residence (e.g. rural villagers or urban slum dwellers).
This simplistic labeling of five billion people mostly serves our convenience to classify the
unknown. Indeed, there exists a greater level of variations within the poor not only regarding
their demographic factors but also psychological factors such as preferences, motivations and life
aspirations. While the mobile users in advanced countries are extensively studied for their
demographics and individual preferences, little is known about the poor as individuals especially
in countries where national ID, address and accurate census data are unavailable.

17

In this regard, the dissertation proposes a conceptual shift, looking at the poor as active,
independent and rational individuals who possess heterogeneous behavioral motivations, in
contrast to the conventional notion viewing them as a mass of heterogeneous victims. It follows
the argument made by Prahalad (2004) who calls for a conceptual shift on the Bottom of the
Pyramid (BOP) as the untapped market currently ignored by the large private sector as well as on
the poor as rational decision-makers who are active in sense-making within their given
constraints. Similarly, in studying the poor’s mobile use, this dissertation argues for the need to
perceive the poor as heterogeneous individuals who make rational choices based on their
cognitive assessment on mobile use and their contextual conditions.
1.4.3. Exploring Motivations within Socio-demographic Constraints: In the current M4D
research, a common approach to survey data analysis is to compare the pattern of mobile service
users and non-users by demographic attributes and describe the difference in relation to age,
gender, income, education, region, and so forth. Others investigate the causal directions from
demographics to use behaviors (Wesolowski et al., 2012; Hilbert, 2011; Zainudeen et al., 2010;
Guierrez & Gamboa, 2010) and such approaches are also common in the practice of digital
divide research (Busy, 2000; Hoffman & Novak, 2000; Van Dijk, 2005; Hargittai & Hinnant,
2008; Hsiesh et al., 2008, 2011; Livingstone & Helsper, 2010; Deursen & Dijk, 2010; Wei &
Hindman, 2011).
Analyzing the mobile use patterns in relation to demographic factors certainly provides
useful description of what is happening at the BOP. Nonetheless, such an approach involves two
potential limitations: one is the effect of demographic attributes can be exaggerated in a way that
its interpretation may suggest demographic and structural conditions are major determinants of
the current use behaviors without considering individual differences. Another issue lies in its
18

efficacy of drawing practical implications because, although the research may address the current
‘gaps’ between gender, education and income, such structural barriers relating to demographic or
socioeconomic factors are practically difficult to overcome via short-term M4D intervention
programs.
Up until now, in development research, far too little attention has been paid to the poor’s
psychological reasoning and motivational drives behind their behaviors (Heeks, 2012; Banerjee
& Duflo, 2011). In this regard, the behavioral science approaches in communications,
psychology and information systems can provide a new direction to the mobile use studies
(Pedersen & Ring, 2003). If qualitative approaches explore the range of user motivations by
observing their behaviors and questioning the reasons, the behavioral studies delineate the
internal functions of multiple user motivations by modeling the process and examine their
relative importance within the cognitive process. Defining user motivation can both enrich our
understanding of BOP’s mobile adoption and use behaviors revealing specific user needs behind
M4D programs. For practitioners, such an approach can provide a useful framework for project
evaluation as well as suggest the points of intervention that they can influence through service
design, promotion and marketing programs.
Nevertheless, the problem lies in that there are largely two groups of researchers in media
and IS research focusing only on one side of the coin. Studies exploring user motivations have a
tendency to focus mostly on psychological factors while taking insufficient account of the effects
of socio-demographic conditions. Although recently scholars examine the moderating effects by
one or two demographic effects such as gender or age, it is a common practice to control
biological, socioeconomic, and cultural factors in order to maximize the assessments on the

19

motivational relationships. In contrary, as discussed earlier, there are researchers who see the
socio-demographic effects as the primary determinant of the digital social inequality.
Neither approach captures the accurate picture of the reality of mobile use if they focus
on only one aspect of the two related issues. Hence, the dissertation aims to look at the
interaction of both sides and proposes to examine the relationships between motivational factors
in the context of socio-demographic differences. The study posits that it is important to
investigate whether and where such socio-demographic factors intervene in the process of
individual’s cognitive process in mobile use, and how they interact with behavioral motivations.
1.4.4. More-than-voice Use of Mobile: Mobile phones are increasingly becoming multifunctional computing devices. Even the basic mobile phone owned by most of the poor comes
with several built-in applications (i.e. alarm clock, calendar, simple games, music players etc.)
and it is also open for potential services that can be introduced over the existing phone. Although
voice calling is the primary function of mobile phone, it is not the only service that a mobile
phone delivers.
Following conventional approaches in media studies, mobile use can be defined largely
in four ways, including ‘binary acceptance’ in adoption studies (Rogers, 1995; Davis et al.,
1989), ‘the quantity of exposure’ in uses and gratification studies (Rosengren, 1974; Papacharissi
& Rubin, 2000; Ruggiero, 2000), ‘taming’ in domestication studies (Silverston, 1994; Haddon,
2001) and ‘re-invention’ in appropriation studies (Orlikowski, 1993; Rogers, 2003; Carroll et al.,
2002). Overall, in the tradition of mass media research, media use is often defined as the length
of time spent on watching television, reading a newspaper or talking on the phone. However, as
Wirth et al. (2008) point out, this simplification to technology use fails to embrace the multifunctional and multi-faceted nature of new communications technologies which also have a
20

potential to evolve with changing needs of the users. For instance, ‘time’ measurement (i.e. hours
using internet) is not sufficient for studying Internet use as the medium offers a venue to a
variety of options for different activities that potentially have different effects on individuals.
Therefore, several researchers suggested new approaches to understand Internet use including a
typology of web activities, the purposes of use, the level of use intensity or the potential impacts
of use (Kraut et al., 2004; Howard et al., 2001; Papacharissi & Rubin, 2000; Hargittai & Hinnant,
2008). As the mobile phone is also a multifunctional and multipurpose technology as in the case
of Internet, the study proposes alternative ways to define mobile use.
Furthermore, most M4D programs presently employ services beyond voice calls. In fact,
Short Messaging Services (SMS) is currently the most popular method employed by such mservices (GSMA Intelligence, 2013). M-banking and m-health services are primarily based on
SMS in delivering its services and most m-agriculture services also adopt SMS as a means to
disseminate information. Various reasons support its popularity. SMS is available on all mobile
phones regardless of manufacturers or network systems. The technology is robust and reliable; it
works well even in the rural areas with patchy coverage as SMS messages can be stored in the
network’s server and can be forwarded when the phone appears within a signal range (Fitzgerald
et al., 2010). Moreover, SMS can be easier and more cost-effective to disseminate information to
a wide population when compared to a few-minute-long call, voice message or Interactive Voice
Response (IVR) system.
Despite the importance of more-than-voice use of mobile phones for M4D, there exist
only a limited number of studies exploring mobile use beyond voice calls in developing countries.
While the m-services based on SMS and more-than-voice features are proliferating, studies find
that user behavior of utilizing mobile services beyond voice calls are still scanty, in particular

21

among the BOP mobile users. A 2008 survey of South Asian countries reports that around 30
percent of the BOP mobile owners in India and Bangladesh had ever sent or received SMS
(LIRNEasia, 2008). Similarly, only 37 percent of the BOP women in four developing countries
had sent SMS regardless of their literacy levels (GSMA mWomen, 2012). The level of user
adoption was found to be low even in the case of SMS-based intervention services designed for
the poor. Zainudeen & Ratnadiwakara (2011) point out that the awareness of the BOP users of
such information and banking services was less than 20 percent while their actual use was much
lower. From an experiment of SMS-based healthcare service in rural Uganda, Chib et al. (2012)
found that the response rate to such intervention was again as low as 20 percent despite
participation incentives, and the effect on health knowledge was only limited.
Therefore, the dissertation calls for further attention on studying mobile use beyond voice
calls. It proposes to conceptualize mobile use in two ways: first, the diversification of mobile use
which means the utilization of the given functions available on a mobile phone and the second,
voice-only use versus SMS use that distinguish the difference between using mobile phones as a
mere telephone or using it a data communication device.
Based on the above rationales, the dissertation addresses the following research questions:
What are the factors affecting mobile use among the mobile owners at the BOP?
RQ1: What are the main motivational drivers of more-than-voice mobile use among the BOP
mobile owners?
RQ2: How do BOP mobile owners’ motivations behind more-than-voice mobile use interact with
their socio-demographic attributes?

22

CHAPTER 2: THEORETICAL FRAMEWORK

The dissertation’s theoretical framework is mainly drawn from the Theory of Planned Behavior
(TBP) (Azjen & Fishbein, 1980). This chapter first reviews several theories concerning people’s
motivational drives and cognitive process in technology adoption. It then discusses the
comparative advantages of the TPB to the current study context, followed by some of the
remaining limitations in explaining the BOP’s mobile use behaviors. It suggests how these
conceptual and practical limitations can be supplemented by merging it with other parallel
theories and proposes an extended model.
2.1. Review of Technology Adoption Theories
As discussed, behavioral science approaches within the field of communications, social
psychology and information systems provide good guidelines to understand the mobile user
behaviors. In particular, theories of technology adoption shed much light on deconstructing user
need and motivational drives behind user behaviors (Pedersen & Ring, 2003; Wirth et al., 2008).
These theories are not only applied to explain adoption behavior but also the continued use of
technology. As suggested in the domestication approach (Haddon, 2001), ‘technology adoption’
is not limited to the initial acquisition of hardware or services, but can encompass the ongoing
adaptation of new services offered via the already adopted technology. In its broad sense,
therefore, adoption is a part of an ongoing process of technology use. In the context of multifunctional technology such as mobile phones and the Internet, existing theories of technology
adoption can provide a useful framework to explain how people use their mobile phones by
adopting new services and expanding their use spectrum.

23

2.1.1. Brief Summary of Technology Adoption Theories: Understanding how and why people
accept or reject new technology has been addressed by many researchers in diverse disciplines.
Several theories have been developed to explain key drivers behind people’s technology
adoption. Diffusion of Innovation (DOI) is one strong pillar addressing how adoption is
determined by an individual’s perceptions on the attributes of innovation. Developed by Rogers
(1962), the theory proposes five attributes of innovation: relative advantage, compatibility,
trialability and observability. Later, DOI researchers from Information Systems research added
three more attributes: image, voluntariness, and result demonstrability (Moore & Benbasat,
1991). Compared to other adoption theories, DOI offers the most comprehensive framework
covering the multifaceted utilities of innovation (i.e. technology), including instrumental and
technological utility, social and symbolic utilities and accessibility. It also takes account of social
and cultural factors such as interpersonal and mass communication channels, social system and
socioeconomic factors (Rogers, 2003, p.170).
From social psychology, the Theory of Reasoned Action (TRA) and its subsequent
version of the Theory of Planned Behavior (TPB) constitute the second pillar of adoption
research. The primary goal of these theories is to explain and predict the process of behavioral
change, and it is also widely applied in technology adoption studies since ‘technology adoption’
can be understood as a type of change in one’s own way of acting. The TRA explains that
behavior is fundamentally determined by one’s intention to perform a target behavior, and this
intention is driven by two motivational factors: Attitude (A) and Subjective Norm (SN). These
motivational factors reflect the underlying salient beliefs one may have about the behavior and
one’s subjective assessment of its relative effects (Ajzen & Fishbein, 1975, 1980). The TRA
provides parsimonious and powerful causal explanations to human behavior in general (Conner

24

& Armitage, 1998). However, its application is limited to behaviors under volitional control. In
other words, TRA tends to lose its explanatory power if the behavior requires certain conditions
beyond one’s own control such as particular skills, resources, information or opportunities that
are not freely available all the time (Fishbein, 1993; Conner & Armitage, 1998; Ajzen, 1991).
Hence, Ajzen (1991) added the third variable called Perceived Behavioral Control (PBC)
to the TRA’s two motivational factors (A and SN) and introduced the Theory of Planned
Behavior (TPB). PBC includes one’s own evaluations on both internal control factors (e.g.
emotion, physical and mental deficiencies, skills, abilities, willpower etc.) and external control
factors (e.g. available resources, opportunities, barriers, supports or objections from others etc.).
The TPB offers a parsimonious framework with only three constructs that seek to explain all
kinds of human behaviors regardless of research context. Its simplicity and the efficacy to
explain human behaviors enabled the theory to be widely applied not only in social psychology
and communications but also applied research in health communications, technology adoption,
environmental research, and other fields. More recently, Fishbein (2000) reviewed several
theories predicting behaviors3 and proposed the Integrative Model (IM), which retains the main
constructs of the TPB but PBC is replaced with Self-efficacy. In addition, the path between the
behavioral intention and behavior is moderated by ‘skills & abilities’ and ‘environmental
factors’.
Evolved from the TRA, the Technology Acceptance Model (TAM) and its subsequent
versions (TAM2, the Unified Theory of Acceptance and Use of Technology) integrate a third
pillar developed from the Information Systems (IS) research that specifically addresses
technology adoption in the organizational contexts. The TAM is one of the most widely used
3

The reviewed theories include the TRA, the TPB, the Theory of Subjective Culture and Interpersonal Relations,
Transtheoretical Model of Behavioral Change, the Information/Motivation/Behavioral-skills model, the Health
Belief Model, Social Cognitive Theory.

25

theoretical frameworks in explaining user adoption of information technology. Grounded in the
tradition of TRA, it shares several key assumptions including that users make a rational
assessment, and that adoption behavior is under one’s volitional control. It also follows the TRA
in that behavior is explained directly by one’s behavioral intention and that this intention is, in
turn, determined by one’s attitude about the technology in question, formed by one’s salient
beliefs. Nevertheless, the TAM departs from the TRA in the specification of the attitudinal
factors in a way to be more relevant to the context of technology use. Replacing Attitude in the
TRA, TAM postulates two key fundamental concepts that determine users’ behavioral intention
behind technology adoption: Perceived Usefulness (PU) and Perceived Ease of Use (PEU). In
addition, the TAM assumes that the two fundamental determinants of PU and PEU are universal
regardless of types of technology, and their measures are offered a priori as an ‘off-the-shelf’
package (Davis et al., 1989).
However, the closed nature of the TAM with the pre-determined set of variables and
measurements made it inevitable for researchers to develop an upgraded version. Researchers
soon tested additional variables in different research contexts and began to dispute the TAM’s
validity (Agarwal & Prasad, 1999; Karahanna et al., 1999; Gefen & Straub, 1997; Karahanna &
Limayem, 2000; Hong & Tam, 2006; Yousafzai et al., 2010, Lee et al., 2003). Accordingly, an
extended version, named TAM2, was created with seven new variables as antecedents to PU and
two moderators borrowed from the TRA and DOI: Subjective Norms, Image, Job Relevance,
Output Quality, Result Demonstrability, Experience and Voluntariness.
In 2003, TAM researchers announced a new model called the Unified Theory of
Acceptance and Use of Technology (UTAUT) based on thorough analysis of the eight
comparable models, including the TRA, TPB, Decomposed TPB, DOI, SCT, Motivational

26

Model, the Model of PC Utilization (Venkatesh et al., 2003). The UTAUT is the outcome of the
TAM scholars’ effort to create a holistic and powerful model (Venkatesh et al., 2003). It consists
of four key perceptions on technology covering similar constructs in other models: performance
expectancy (embracing perceived usefulness, attitude, relative advantage, compatibility, outcome
expectancy), effort expectancy (perceived ease of use, complexity), social influence
(observability, subjective norms), and facilitating conditions (perceived behavioral control, selfefficacy). It also acknowledged the effects of user differences in relation to demographics and
specified age, gender, and experience as moderators.
Most recently, Venkatesh et al. (2012) released the UTAUT2 which targets general
consumers’ adoption and use of end-user technologies and services. The new model retains the
four main UTAUT predictors (performance expectancy, effort expectancy, social influence,
facilitating conditions) and three moderators (age, gender, experience). Then, it added three new
variables relevant to general consumers (hedonic motivation, price value, habit) as predictors to
BI. And the effects of the moderators have been adjusted from moderating all the BI predictors
in the previous model to affect only FC and three newly added variables. At the same time, the
moderator Voluntariness is dropped. Finally, the path from FC is previously directed only to B
but now it has both paths to BI and B. Similarly, Habit also exerts influence on both BI and B
while experience also moderates BI - B path.
Social Cognitive Theory (SCT) is another social psychology theory influential in
technology adoption studies. Founded by Bandura (1986), SCT offers a set of influential
conceptualization including triadic reciprocality, self-efficacy, outcome expectations, selfreflection, self-regulation, etc. Among others, self-efficacy is the most widely used construct in
technology adoption studies. Self-efficacy refers to ‘one’s belief in capability to organize and

27

execute a particular course of action’ (Bandura, 1986). It is a dynamic self-evaluation formed
through four agentic experiences: enactive learning (one’s own performance), vicarious
experience (observational learning), social persuasion, and physical and emotional status
(Bandura 1997; Pajares, 2002). Unlike self-esteem or self-confidence, therefore, self-efficacy is a
form of an optimal balance between one’s ability and desire, or a reflection of both personal and
sociostructural influences. In the domain of technology use, the concept was further elaborated
as ‘computer self-efficacy’ (Compeau & Higgins, 1995) or ‘Internet self-efficacy’ (Larose &
Eastin, 2004) to measure one’s assessment in skills and capability to perform a task on a
computer or on Internet.

2.1.2. Comparison of the Adoption Theories: Due to their conceptual similarities, researchers
became interested in comparing the theories’ explanatory power. Some suggest the TAM
explains more variance in intention than the TRA or the TPB (Davis et al., 1989; Mathieson,
1991) while others find that the TPB or DOI performs better than the TAM (Talyor & Todd,
1995; Plouffe et al., 2001). On the other hand, the founders of the UTAUT present that it
outperformed the eight competing models. The problem of these comparative analyses is,
however, their tendency to obsess with comparing ‘R-squared’ as if it is the only indicator to
theory’s power. The competence of theory should be considered from multiple angles while
direct comparison may not be possible if the theories are designed for different research domains,
behavior in question, research contexts and the population of study. Hence, this study shifts the
focus from the competitive power of the theories to the commonalities of the theories and
examines which theory can offer the most basic framework to study mobile use at the BOP.

28

As illustrated in Figure 3, the theories reviewed in this chapter have emerged from
different disciplines but became closely inter-related in the process of model validation,
extension and maturation. The TAM was derived from the TRA, and its sequential model TAM2
borrowed several constructs from DOI in the efforts to improve the model’s explanatory power.
Also, the UTAUT attempted to merge the TRA, TPB, TAM and SCT among others. For
instance, its facilitating condition (FC) is a mixture of three constructs: PC Utilization Model’s
FC, TPB’s PBC and DOI’s compatibility (Venkatesh et al., 2003, p.454). Similarly, Taylor and
Todd (1995) introduced the Decomposed TPB (D-TPB) which elaborates the TPB with
constructs from the TAM (PU, PEOU), DOI (compatibility) and SCT (self-efficacy). On the
Figure 3: The Evolution of Technology Adoption Theories

29

other hand, the TPB’s PBC is conceptually similar to SCT’s self-efficacy as Ajzen himself
admitted (Ajzen, 1991) and Fishbein (2000) actually substitutes PBC with self-efficacy in his
Integrated Model (IM).
Indeed, these theories share several assumptions. First, they all assume that technology
users are active and rational decision-makers. Second, they posit that human behavior is
primarily triggered by need, motivation and perception. Third, technology adoption is a
deliberative process where the decisions are made based on careful assessment of available
information. Finally, except DOI, they all tend to pursue a parsimonious framework explaining
the essence of the decision-making process only with a few key constructs. In addition, there
exists a degree of conceptual similarity in the key constructs included in these adoption theories.
Common factors across these studies can be extracted as four groups: Technological Utility,
Social Influence, Use context under control and Contexts beyond user controls. A more detailed
review on the adoption theories and their conceptual commonalities can be found in Appendix 1.

2.2. Theoretical Strengths of the TPB
This study chose the TPB as the baseline framework due to its theoretical advantages of
generality, openness and parsimony. First, the TPB framework provides a more general
framework to explore users’ diverse beliefs on mobile use in everyday settings. The TAM and
the UTAUT were originally designed to study technology adoption in organizations where the
users share similar purposes (e.g. productivity) and use contexts (e.g. working in offices) while a
series of trainings is typically offered before a technology is implemented. Coming from the
Information Systems tradition, it also pays more attention on the technology’s characteristics
(usefulness and ease of use) and relatively less on contextual controls. On the other hand, the

30

TPB provides a comprehensive set of motivations that can be applied to any kinds of human
behavior in diverse setting as long as the behavior in question is specifically defined. Such
advantages of generality can be more suitable for a multi-functional use of mobile phones in
everyday contexts.
Second, the TPB offers a more open framework that allows researchers to explore any
beliefs unique to the target behavior as well as to identify additional variables if they can
improve the variance explained in the target population (Ajzen, 1991; Fishbein & Cappella,
2006). This open nature of the TPB is distinct from the TAM-family theories, which aim to offer
a set of pre-determined constructs with the off-the-shelf measures. The current study is an initial
effort expanding the research to the under-studied population at the BOP in South Asia and their
relatively new behavior of mobile use. The open nature of the TPB encourages not only
replicating the model onto different samples but also exploring new attributes pertinent to the
studied behavior, thus making it suitable for the current study.
Figure 4: Theory of Planned Behavior

31

Third, the TPB offers the most comprehensive but parsimonious framework to
understand technology use. As Fishbein (2000) once put it, ‘there are only a limited number of
variables that must be considered in predicting and understanding any given behavior’ and the
TPB explains this general behavior with three comprehensive constructs: attitude, subjective
norms and behavioral control. Unlike the TAM-family of theories, it does not specify technical
usability but rather embraces it within PBC from a user’s perspective.
In fact, it can be suggested that the construct ‘ease of use’ overlaps both with technology
characteristics and contextual controls. Its meaning possesses a degree of duality as it is unclear
‘perceived easiness’ is intrinsic to a technology’s characteristics (as suggested by the TAM) or
belongs to user’s ability to control a technology (as suggested by the TPB). For instance, a
person may find a technology is easy to use if either the technology has a user-friendly
characteristics or s/he has sufficient skills and knowledge to handle the target technology. Hence,
the study argues that ‘ease of use’ exists as an underlying belief system rather than a separate
main factor, and in this regard, it concludes that the TPB offers the most parsimonious
framework.
Recently, scholars in media effect studies began to question whether the TPB is an
appropriate framework to study media consumption which is increasingly perceived as habitual,
automatic, and driven by spontaneous behaviors rather than rational decision-making processes
(Hartmann, 2009; LaRose, 2010). For instance, LaRose (2010) argues that media consumption is
likely to be habitual as it becomes repetitive through dual conscious and non-conscious
processes, contrary to the conventional assumption of active, goal-directed, and deliberate
process. Similarly, Hartmann (2009) specifies that TPB may fail to provide an accurate model
for automatic media behavior unless the decision involves a degree of risk and investment (e.g.

32

time, cost, resources) or if the user is highly unfamiliar with the choice or anticipated behavioral
outcomes.
In this regard, this study assumes that the BOP’s mobile use behavior is closer to
conscious and goal-directed than habitual and automatic behavior. The boundary may blur when
considering mobile use as answering phone calls which requires spontaneous reaction from
users. Nevertheless, most BOP mobile users are on pre-paid mobile service so that each call or
SMS involves a cost per minute or message, which activates ‘executive control’ or deliberate
consideration between cost and outcome. At the same time, unlike users in the developed world
who are surrounded by electronic gadgets, most BOP users are unfamiliar with economic
communication devices and therefore trying or learning to use new mobile functions or services
may require a degree of time investment and deliberation over its benefits.

2.3. Proposed Model Extension and Hypotheses
Despite the theoretical advantages, there still remain some conceptual and practical limitations of
the TPB to fully capture the chosen study contexts. Thus the study proposes the following four
areas for model modification by merging the TPB framework with elements from
complementary theories and extending it with additional variables. Figure 5 shows the proposed
model and the related hypotheses are justified below.

33

Figure 5: Proposed Research Model

2.3.1. TPB Framework with Perceived Usefulness
As specified in the TPB, the study assesses the three main motivational drives and their
relationship with behavioral intention and actual use as the basis of research. However, it
proposes that attitude in the TPB needs to be further specified in the context of technology use
and, in this regard, perceived usefulness in the TAM is a more specific and robust construct to
capture the attitudinal values related to mobile use.
Attitude refers to ‘the degree to which a person has a favorable or unfavorable
evaluation of the behavior in question’ (Ajzen, 1991). This conceptualization seeks to be broad
and general in order to be applied to all kinds of human behavior. In practice, attitude is usually
measured by several bi-polar Likert-scale questions on whether the target behavior is good or
bad, favorable or unfavorable, wise or foolish, or satisfying or unsatisfying, and so forth. This
form of attitudinal measure is appropriate for most behaviors, such as exercise, recycling,

34

smoking, drinking, getting a mammogram, etc., on which people may have different judgment
based on their personal beliefs or experiences. Hence, attitude is a general but useful construct to
reflect this varying degree of evaluations among individuals, and especially more so if the
behavior in question involves some level of controversy or disagreement.
However, the behavior of ‘technology use’ is grounded in an instrumental nature where a
technology functions as a tool to fulfill users’ utilitarian values and such an instrumental attitude
is widely shared by most of users. For that reason, unless using the target technology involves
considerable harm or cost, it is unlikely that the users form a severely negative or unfavorable
attitude. Especially in the current study context where the users have already decided to purchase
a mobile phone, it is doubtful whether people possess considerably varying attitude across bipolar measures (i.e. negative – positive) towards mobile use and whether attitude measures
appropriately capture the instrumental values inherent to mobile use behavior.
In this regard, Perceived Usefulness of the TAM refers to ‘the degree to which an
individual believes that using a technology will increase his or her performance’. Originated
from the same root of the TRA, PU is a customized version of ‘attitude’ in the context of
technology use. It specifies the instrumental utilities within technology use and aims to measure
the degree to which the target technology serves the instrumental needs of the users. Its
robustness in predicting technology use has also been proven in extensive empirical studies
(Davis et al., 1989; Venkatesh et al., 2003; Hong & Tam, 2006; Yousafzai et al., 2010).
Therefore, this study integrates perceived usefulness in place of attitude in order to define what
instrumental utilities of mobile use drive people’s behavioral intention.
Subjective Norms (SN) refers to ‘the perceived social pressure to perform or not to
perform the behavior’(Ajzen, 1991). Influence from social groups is often confounded with other

35

similar constructs such as subjective norms, conformity, compliance, perceived critical mass, and
perceived network externalities (Cho, 2012; Cialdini & Goldstein, 2003). In this study, we
follow the TPB’s narrow definition which specifies one’s perceived pressure to conform to the
norms perceived from the people who are important or with whom one interacts in proximity
(e.g. family, peer, community, opinion leaders, etc.). The efficacy of SN in predicting behavioral
intention has been questioned as it is known to be a weak predictor of behavioral intention in the
TPB framework (Armitage & Conner, 2001; Manstead, 2011). Although several studies also
found social influence as a direct predictor of use intention in mobile services (Nysveen, 2005;
Hong & Tam, 2006; De Silva et al., 2011; Kim et al., 2008; Lu et al., 2010), the current study
pays relatively less attention to the effect and antecedents to SN.
Perceived Behavioral Control (PBC) refers to ‘the perception that performance of a
specific behavior is within a person’s volitional control’ (Ajzen, 1991). PBC may differ from the
actual control which to some extent dictates the occurrences of one’s behavior. Nevertheless, as
Ajzen (1991) suggests, what matters in the formation of one’s intention at the psychological
level is his/her perception over the contextual control they believe to possess. In the context of
mobile use, people’s intention to use mobile services is influenced by their perceived control
over the skills, resources and opportunities that they believe to facilitate mobile use.
In addition, the TPB’s original model acknowledges the effects of behavioral control on
the path between the intention and the behavior. This indicates the actual availability or
constraints of resources and opportunities given in the structural and environmental contexts
(Ajzen, 1991). As the behavioral intention to behavior path is widely supported in empirical
findings and hence taken for granted, it is occasionally omitted from explicit examination. In the
meta-analysis of the TPB studies from the developed world, PBC is found to add, on average,

36

only 1 percent of variance explained in behavioral intention (Sutton, 2003). However,
considering the potential contextual constraints at the BOP, the study follows the original
framework of the TPB specifying the effect of PBC on actual use behavior.
Hence, based on the previous research literature on the TPB and TAM, the following
hypotheses are formed as the basis of the study.
H1a: Perceived usefulness is positively associated with behavioral intention to use a mobile
phone.
H1b: Subjective norms are positively associated with behavioral intention to use a mobile
phone.
H1c: Perceived behavioral control is positively associated with behavioral intention to use a
mobile phone.
H1d: Behavioral intention to use a mobile phone is positively associated with actual use of a
mobile phone.
H1e: Perceived behavioral control (PBC) is positively associated with actual use of a mobile
phone.

2.3.2. Exploring the Antecedents to PU and PBC
Building on the TPB framework, the study aims to explore the antecedents to PU and PBC in
order to identify what makes people find a mobile phone useful and what contributes to the
enabling contexts for mobile use. The primary concern of the TPB is to predict a behavior in the
most parsimonious manner by extracting three fundamental determinants (i.e. A, SN, PBC) of
any behaviors. It pays relatively little attention to exploring the factors associated with these
fundamental determinants. According to the TPB, such factors preceding A, SN and PBC exist
as underlying ‘belief systems’ and they are assumed to be fully reflected within the three main
determinants of behavioral intention. It also assumes that all belief systems are equally weighted
37

as their relative influences are undefined (Ajzen, 1991). As the TPB became popular in applied
research such as health communications, however, a degree of specificity was required to
suggest the points of intervention for practitioners. In this regard, the theory suggests for
researchers to be specific in defining the target behavior in terms of the action, target, context
and time or to identify specific population groups such as high/low intention groups (Ajzen,
1991; Fishbein, 2006).
Nonetheless, the dissertation posits that, in the current study, it is important to explore the
belief systems as the antecedents and to examine their relative influences. This is because the
population of the current study has been largely ignored and little is known on the behavioral
motivations relating to their mobile use. Thus, exploring the antecedents can add much
information to our knowledge, which may not be fully captured by the three general
determinants. Furthermore, exploring the antecedents to PU and PBC can draw useful
implications for M4D practitioners. As critics point out, the pursuit of theoretical simplicity may
lead to the loss of useful information, and in this regard, the theory does not pay much attention
to what makes the useful service or the enabling contexts (Bagozzi, 2007; Benbasat & Barki,
2007). Suggesting people will use the technology if they find it useful or if they have sufficient
controls does not provide much practical information for those who want to design a useful mservice. Therefore, based on the existing literature on the motivations behind mobile use, the
study includes the following motivational factors: perceived benefits, trust, perceived ease of use,
social support and price value.
The perceived benefits of mobile formed from past experience of using mobile phones
can influence a user’s evaluation of the instrumental values and specifically the degree of
usefulness of mobile phones. Based on the previous M4D literature describing general benefits
38

of mobile phones in developing countries, this study explores the instrumental benefits of mobile
phone in terms of economic and efficiency benefits.
Perceived Economic Benefits refers to the ability to expand one’s own economic
resources by accessing information, obtaining work/business opportunities, and securing
financial resources. It includes benefits such as enhanced communications with financial partners
(e.g. banks, money lenders, employers or buyers), which can facilitate the expansion of one’s
economic resources regardless of activity types (e.g. information, financial, or social uses).
Several studies in the ICTD literature suggest the impact of mobile phones on economic gains
via dissemination of up-to-date market price information (Jensen, 2007; Aker, 2008). Although
evidence on direct economic gains via mobile use is still limited and blurred, a few pilot projects
appear to offer success stories, including profit increases among the farmers in Senegal through
mobile-based price delivery (Rashid & Elder, 2009). Chew et al. (2010) also showed that using
mobile phones has a positive impact on business growth among female micro-entrepreneurs in
India.
Perceived Efficiency Benefits refers to one’s ability to manage existing – as opposed to
obtaining new – resources in a more efficient manner. This includes saving time and cost by
reducing the number of trips or their corresponding expenses. Regardless of activities, the
concept focuses on ultimate gains in one’s efficiency in everyday life, including more efficient
use of time, money, knowledge and social capitals as a result of travel reduction or better
organization of existing social contacts such as arranging childcare or daily business. Abraham
(2007) found that mobile phones increased the efficiency of Indian fishermen by allowing them
to coordinate their catch with demand by helping them in finding the underserved markets while
reducing their time idling at sea. Boateng (2011) also suggested that mobile phones enable

39

Ghanaian traders to monitor and schedule their sales activities more efficiently. Not limited to
economic activities, we assume that mobile phones introduce an enhanced capability to
coordinate one’s everyday activities in quicker, cheaper, and better-organized ways. From these
arguments, the following two hypotheses are derived:
H2a: Perceived economic benefits are positively associated with perceived usefulness.
H2b: Perceived efficiency benefits are positively associated with perceived usefulness.

Trust is an important factor to take into account when discussing the adoption of
technology-enabled communications beyond face-to-face interactions. In the adoption studies
from the developed world, trust was considered an important motivation behind service adoption
mostly in relation to banking or e-commerce (Pavlou & Fygenson, 2006; Luarn & Lin, 2005;
Zhang & Mao, 2008). However, for BOP users who are not familiar with electronic
communications over technologies, trust and service reliability in mobile phones can affect their
level of perceived instrumental values of mobile phones. Indeed, several qualitative studies of
mobile use among the poor found that trustworthiness is an important indicator when considering
technology adoption. Mittal et al. (2010) found that the quality of information, timeliness and
trustworthiness are the three important factors influencing Indian rural farmers’ mobile-foragriculture services. Crandall (2012) also found that trust is an important factor to Kenyan
farmers’ adoption of SMS. Khodamoradi and Abedi (2011) also discuss that trust in technology
influences the adoption of any new technologies among Iranian individuals. Hence, the study
forms the following hypothesis:
H2c: Trust in mobile services is positively associated with perceived usefulness.

40

Perceived ease of use (PEOU) refers to ‘the degree to which an individual believes that
using a particular system would be free of effort’ (Davis et al., 1989). The concept is from the
Technology Acceptance Model (TAM) and the theoretical and empirical relationship with PU
and PEOU have been extensively supported by TAM research (Davis, et al., 1989; Venkatesh et
al., 2003; Hong & Tam, 2006; Yousafzai et al., 2010, Wu & Wang, 2004). However, the TAM
conceptualizes PEOU as a part of the characteristics of technology, which has a distinct effect on
behavioral intention in addition to its relationship to PU. On the other hand, the TPB does not
specify the ease of conducting a behavior as a separate construct and rather embraces it as a
user’s behavioral control relating to the resources and knowledge to perform a behavior. In this
regard, the study posits that, from a user’s point of view, PEOU is conceptually linked both with
the technology’s ease of use (e.g. interface) and the user’s perceived level of control (e.g. skills,
resources). For instance, users may answer that given software is easy to use because it has a
user-friendly interface or because the user has sufficient skills to handle the software. Hence, the
study argues that PEOU is a compounded construct involving both instrumental evaluation of
technology (convenience) and user’s level of behavioral control (skills to use a technology). It
conceptualizes PEOU as an antecedent to both PU and PEOU and forms the following
hypotheses:
H2d: Perceived ease of use is positively associated with perceived usefulness.
H2e: Perceived ease of use is positively associated with perceived behavioral control.

Although calling or answering a mobile phone is fairly intuitive, other functions on
mobile phones may be challenging for users unfamiliar with technology. In this sense,
availability of technical assistance is one of the important conditions facilitating technology use
(Park et al., 2008; DiMaagio et al., 2001). In the context of mobile use in developing countries,
41

studies found that technical assistance often occurs in social settings (social support), typically
through help from family members or friends (LIRNEasia, 2009). Therefore, BOP users who
lack skills and literacy are likely to be more comfortable with using mobile services if they have
assistance from members of their family or from friends who can help with using various
functions on a mobile phone, or know how to resolve technical problems. In developing
countries, technical assistance from the service providers or experts also tends to take place
through social interactions in local shops or top-up kiosks rather than through help-lines or
websites. The presence of these social support systems is a distinctive social influence, which is
more relevant to facilitating conditions to use technology. Therefore, the study hypothesizes:
H2f: Social support is positively associated with perceived behavioral control over
mobile use.

For the poor, who are often defined as people living under USD 2 a day, the cost of
mobile calls and SMS service is likely to play a more important role than for affluent consumers.
Studies show that the BOP users employ various strategies to reduce their spending on mobile
use, including limiting their outgoing calls or using encrypted signs via ‘missed-calls’ (Donner,
2007; Sey, 1009; Horst & Miller, 2006). On the other hand, other findings also point out that the
BOP users tend to spend a considerable amount of their income on mobile use (Aileen et al. 2011;
Galperin & Mariscal, 2007). To estimate the effect of ‘cost’ scholars may utilize subjective
measures such as ‘perceived cost’, typically measured by the degree of fiscal expenditure the
user perceives with respect to his/her own available fiscal resources. In fact, perceived cost is a
multi-dimensional construct because it not only depends on one’s available budget, but also on
whether the value obtained is commensurate with the cost incurred (Dodds et al., 1991;

42

Venkatesh et al. 2012). In other words, people may perceive the price to be slightly higher than
their expectation, but may be willing to use the service if it offers value to them, or vice versa.
Therefore, we suggest ‘price-value (PV)’ as a more comprehensive measure than ‘perceived
cost’. In this study, PV is defined as ‘the degree to which individuals perceive the
appropriateness of the cost in relation to one’s perceived benefits and preference of the service’.
Previous studies also found perceived cost relating to service values to be a significant predictor
of technology use (Kim et al., 2008; Hong & Tam, 2006; Venkatesh et al. 2012). Hence, we
propose the following hypothesis:
H2g: Price-value is positively associated with perceived behavioral control over mobile use.

2.3.3. The Effect of Socio-demographic Factors
As the primary interest of the TPB lies in the workings of psychological determinants, the
theory does not pay much attention to the effects of structural and demographic factors. Such
demographic effects are considered as indirect background factors to the extent to which they are
reflected in the underlying belief systems. As Ajzen notes (2011), the TPB assumes that the
salient beliefs possibly originate from a mixture of background sources such as personality,
emotional status, demographic variables, media exposure, and other information sources. Thus
the effect of socio-demographic factors is not specified in the theory. Accordingly, it is a
common practice for TPB researchers to consider the socio-demographic differences as control
variables in their analysis.
Nevertheless, it is questionable whether these perceptions fully embrace the effects of
socio-demographic factors especially in the case of the poor who are social and economically
constrained. In fact, there are many researchers in the field of digital divide studies who see
43

socio-demographic factors as primary determinants of the unequal adoption and use of
technology. Several studies find that the digital divide is a direct product of the socioeconomic
status or demographic factors such as gender, age, race, location, income and education (Busy,
2000; Bimber, 2000; Hoffman & Novak, 1998, 2000; Bonfadelli, 2002). Some studies also find
secondary effects of socio-demographical characteristics on motivations to use technology
(Hsiesh et al., 2008, 2011) and digital skills and literacy (Deursen & Dijk, 2010; Hargittai, 2005,
2008, 2010; Livingstone & Helsper, 2010) as well as usage patterns (Van Dijk, 2005; Hargittai &
Hinnant, 2008; Jung et al., 2001; Wei & Hindman, 2011). Similar findings are also reported by
studies in developing countries (Furuholt et al., 2008; Fong, 2009; Sey & Fellows, 2009).
Therefore, to better explain the BOP’s mobile use behavior, this dissertation posits that it
is necessary to include effects of socio-demographic attributes within the TPB framework. In
doing so, it raises two subsequent questions. First, it is important to test whether sociodemographic factors moderate the overall workings of the TPB model. Second, it is equally
important to examine where exactly such effects exist in the given model as the theory does not
specify the effects of socio-demographics within the relationship among motivational constructs.
In other words, within the three layers of the TPB theory, it is unclear whether such demographic
effects influence the underlying layer of belief systems, the workings of the three main
determinants on behavioral intention, or the final path from the intention to actual behavior.
Another issue is related to how socio-demographic factors are being defined. Typically,
studies tend to consider socio-demographic effects as a moderator by applying one or two onedimensional demographic variables such as gender, age or education. Yet, in reality, an
individual has multiple biological and socioeconomic characteristics that cannot be defined by a
single variable. Using a set of multiple demographic variables may improve the accuracy but it

44

fails to explain the interaction effects or any correlations between the demographic variables. For
instance, a female in mid-40s with a high school degree working as secretary in an urban city
would be categorized by five separate variables (gender, age, education, location and occupation)
and it is practically difficult to consider their overall effects of these inter-related variables.
Therefore, the current study attempts to measure the multi-dimensional aspects of sociodemographic factors by grouping observations into sub-samples with a set of multiple
demographic variables. And it aims to examine the effects of socio-demographics on the overall
model by comparing the group differences. Hence, the following hypothesis is formed:
H3: The effects of the motivational factors are moderated by socio-demographic factors.

45

CHAPTER 3: RESEARCH METHOD

3.1. Data
This dissertation used the secondary data from the fourth round survey conducted in 2011 as a
part of the LINREasia4’s Teleuse at the Bottom of the Pyramid (Teleuse@BOP) project. The
survey comprises 9,066 respondents of the ‘BOP teleusers’ from the five Asian countries of
Bangladesh, Pakistan, India, Sri Lanka, and Thailand. The ‘BOP’ is defined as the two lowest
strata of socioeconomic classifications (SEC). ‘Teleusers’ is defined as those who are between
age 15 and 60, and have accessed, but do not necessarily own, telephony services in the last three
months including fixed-line, mobile, and public phone services.
Countries. While the studied countries are located in the same geo-economic region, they exhibit
rather different characteristics in terms of their population, economic and social development
status, and their ICT environments (see Table 1). For instance, Bangladesh, Pakistan and India
are comparatively poorer with a lower GDP per capita, a large number of people living under the
poverty line and a smaller number of ICT users than Sri Lanka and Thailand. As the focus of the
current study is the mobile users at the Bottom of the Pyramid (BOP) across the countries, these
differences mainly provide context. Although the study does not neglect these varying conditions
of mobile use in each country, the primary interest lies in the generalizable pattern of behaviors
and motivational drives related to the specific group at the same socioeconomic strata. Rather
than comparing country-specific factors between the five nations (which would be difficult to do
in a systematic way in a cross-national study of five countries), the research is based on the
simplifying assumption that people at a certain socioeconomic stratum within the same geo4

LIRNEasia is a regional think-tank specialized in researching telecommunication policy and regulation across the
Asia Pacific region. The first round of Teleuse@BOP was conducted in 2005 followed by 2007 and 2009 surveys
investigating ICT access, adoption and use among the low-income populations in multi-countries in Asia.

46

economic region share similar ways of living and behavioral perceptions5. Hence, the study pays
more attention on the generalizable behavioral motivations at the BOP mobile users in the SouthAsian countries. For analyzing the psychometrics on behavioral motivations, the country-level
differences are further controlled in the structural equation modeling.
Table 1: Country Profiles

Income Level*

Low

GDP per capita, PPP*
Human Development Index
(rank)**
Life Expectancy at birth (yrs)**
Literacy Rate (adult, %)***
Poverty Indicators****
GINI Index
Poverty headcount ratio at $2 a
day (PPP) (% of population)
Poverty headcount ratio at
national poverty line (% of
population)
Income share held by highest
20%
Income share held by lowest
20%
ICT Indicators*****
Mobile subscriptions
per 100 inhabitants
Fixed-telephone subscribers
per 100 inhabitants
% of Internet users

597.0

Pakistan
179.1
Lowermiddle
772.9

146

146

136

92

103

69.2
57.7

65.7
54.9

65.8
62.8

75.1
91.2

74.3
93.5

32.1

30.0

33.9

36.4

39.4

76.5

60.2

68.8

23.9

4.1

31.5

22.3

29.8

8.9

16.9

41.4

40.0

42.8

44.6

46.7

8.9

9.6

8.5

7.7

6.8

62.8

67.1

69.9

91.6

127.3

0.6

2.5

3.2

16.3

9.5

6.3

10.0

12.6

18.3

26.5

Population (million)*

Bangladesh
154.7

India
1,236.7
Lowermiddle
1106.8

Sri Lanka Thailand
20.3
66.8
LowerUppermiddle
middle
1884.2
3352.5

* World Development Indicators (2012), **Human Development Index Report (2012), ***CIA World Factbook
(2013), ****Poverty and Equality Database (2012, except Pakistan, 2008), *****ITU ICT Statistics (2012)

5

As shown in Table 1, although the absolute numbers of people under the poverty line may vary between the
countries, the income share held by the lowest 20 percent display relatively similar portions across all five
countries (ranged between 6.8 and 9.6 percent of the total income), suggesting the status of relative poverty in
each country does not vary to a large degree.

47

BOP teleusers. Prahalad’s notion of the BOP is a conceptual reference to billions of the poor
residing in developing countries without discussing a specific definition or measures to identify
them. In practice, the poverty threshold of $1 or, more recently, $1.25 a day as established by the
World Bank is a common method to define the poor (Ravallion et al., 2009). However, when
conducting an individual-level survey, it is practically difficult to identify the poor by selfreported questions on their income and consumptions. Respondents often reject to reveal their
actual income or fail to recall the accurate level of their earnings and expenses especially if their
income is irregular or miniscule (de Mel, et al., 2009; Chew et al., 2011).
Therefore, the study used the level of education and the type of occupation as a proxy for
low socioeconomic status. It employed the socioeconomic classification (SEC), originally
Table 2: Sample SEC Criteria for India
EDUCATION

OCCUPATION

Unskilled Worker
Skilled Worker
Petty Trader
Shop Owner
Businessmen /
Industrialist
Of Employees
Self Employed
Clerical / Salesman
Supervisory Level
Officers / Executives –
Junior
Officers / Executives –
Middle / Senior

Literate Schoo Schoo
Some
Secon
Post
Post
Illite- but no l up to l 5 to
College
-dary
Graduate Graduate rate
formal
4
9
but not
school
- General Professional
school years years
graduate

E2
E2
E2
D
D
C
B1
D
D
D

E2
E1
D
D
C
B2
B1
D
D
D

E2
E1
D
D
C
B2
B1
D
D
D

E1
D
D
C
B2
B2
A2
D
D
C

D
C
C
B2
B1
B1
A2
B2
C
C

D
C
C
B1
A2
A2
A1
B1
B2
B2

D
B2
B2
A2
A2
A1
A1
A2
B1
B1

D
B2
B2
A2
A1
A1
A1
A1
B1
A2

C

C

C

C

B2

B1

A2

A2

B1

B1

B1

B1

B1

A2

A1

A1

* The categories of education level and occupation types were customized for each country’s context.

48

developed by Market Research Society of India (MRSI), which is a widely used marketing and
survey framework in developing countries in Asia. It comprises five groups (A to E) based on
the education and occupational status of the chief wage earner of the households (see Table 2)
and only groups D and E were selected for the survey. In case of rural households where a large
number of people are self-employed or unemployed, it used the type of housing – Pucca (stable
house made with cement, bricks, and concrete), Semi Pucca and Kaccha (temporary, unstable
housing made with straws, mud, bamboo, leaves available in villages) – as a proxy to indicate
SEC groups. As discussed in the description in the following section, the current sample
displayed a range of characteristics of low socioeconomic status including low education,
unstable occupation, irregular or no income. In addition, the average monthly household income
was reported as USD 122.02 (SD=53.5) which is below the USD 1 a day criteria when divided
by four household members per day.

Sample Design. The study used a multi-stage stratified cluster sampling by Probability
Proportionate to Size (PPS). Covering all regions (e.g. states, provinces, districts) of each
country, the PPS was used to select a target number of urban and rural centers in each province.
Accordingly, urban and rural areas were randomly selected using PPS on a constant population
interval on geographically ordered centers within each region. Within each area, a starting point
(e.g. a prominent landmark such as a main road, park, hospital, school or well-known building)
was randomly selected with a fixed number of interviews to be conducted around each starting
point. The number of starting points within each area was determined in proportion to the
population of the selected center. From the starting point, the interviewers visited every fourth
household following the right (or left)-hand-rule when facing a junction or end of the street. One
respondent was chosen from each household for the survey. In households with more than one
49

valid respondent, a Kish-grid (random number chart) was used to randomly select the next
member of the household to be interviewed. For more details on the sampling methods, see De
Silva et al. (2008, 2011).
Data Collection. The structured survey questionnaire was first created in English and then
translated into local languages. Back-translation and pre-tests were conducted to modify any
obscure questions and words. The data collection was carried out from May to June 2011 by a
leading market research firm which won the bidding process based on their industry reputation
and survey experience. A face-to-face survey was conducted by intensely trained local
administrators who read out each question and marked the answers on behalf of the respondents.
A set of pictorial and text cards was used for Likert-scale or complex questions.
Sample Characteristics. The dissertation focuses on mobile phone owners only. Of the total
9,047 respondents6, 54.5 percent owned a personal mobile phone (N=4,924). As shown in Table
3, females accounted for 42.7 percent of the total BOP mobile owners and the mean age of the
mobile owners was 32.5 years (SD=11.3). 70.3 percent were married as people in the studied
countries tend to marry in their 20s. The sampled BOP owners mostly resided in rural areas
(68.7%) as the number of urban areas is growing but still limited in these countries. Another
measure of location also indicates that about 40 percent needed to walk more than 30 minutes to
the nearest town.
The majority of the BOP mobile owners were educated up to primary school (55.4%)
while 19.6 percent received no formal education. 18.3 percent graduated secondary school and
3.6 percent proceeded to higher education. About 10 percent of the BOP mobile owners reported

6

For descriptive analysis, the data has been weighted by gender, province group and urban-rural populations to
correct over- or under-sampling for particular socioeconomic groups.

50

Table 3: Sample Characteristics
All
Demographics
Gender

Bangladesh

Pakistan

India

(% of the total 4,924)
40.7
44.7
36.2
59.3
55.3
63.8

Sri
Lanka

Thailand

55.0
45.0

49.6
50.4

Female
Male

42.7
57.3

Mean Age

years
(SD)

32.5
(11.3)

30.9
(10.6)

30.8
(10.2)

32.5
(10.4)

34.4
(11.3)

35.5
(12.5)

Location

Urban
Rural

31.3
68.7

35.0
65.0

57.0
43.0

35.3
64.7

34.4
65.6

48.8
51.2

Education

No formal
Education
Primary
School
Secondary or
more

19.6

25.3

19.7

25.6

8.7

3.5

55.4

48.8

62.7

45.4

71.4

65.9

21.9

25.9

17.6

29.3

20.0

30.5

36.1

47.7

36.5

34.2

37.7

29.8

23.3

16.5

19.7

27.3

19.0

37.0

15.3
15.3

18.5
16.9

35.5
8.3

15.5
23.0

17.3
26.0

12.5
20.7

131.8
(60.2)

103.6
(32.8)

127.5
(50.1)

103.0
(54.5)

186.8
(57.8)

188.3
(48.6)

53.6
92.0
79.8
38.2
6.3

39.9
80.4
56.6
12.4
1.8

14.6
99.3
90.2
8.0
4.4

55.0
85.4
65.7
27.2
2.0

91.6
97.8
95.2
93.0
8.2

90.5
100.0
99.7
77.2
20.5

98.4

100.0

99.9

99.6

94.6

95.8

3.7
(2.5)

3.4
(2.3)

3.1
(2.2)

2.8
(1.9)

3.9
(2.2)

5.9
(3.1)

4.32
(2.7)

3.67
(2.6)

4.27
(2.6)

3.76
(2.5)

4.58
(2.3)

6.80
(2.9)

98.0

100.0

99.0

98.3

92.0

100.0

67.1

86.2

71.1

77.4

60.4

24.4

32.7

19.0

37.8

22.7

51.7

36.7

Occupation

Household
Income
Access to
Facilities

Unemployed
Unskilled
laborer
Skilled laborer
Self-employed
Monthly,
USD (SD)

bank account
Electricity
Television
Radio
Computers
Mobile Use Behaviors
Pre-paid Mobile
Years of Mobile Ownership
Monthly Mobile
Expenditure, USD (SD)
making
calls
missed
Type of Use
calls
SMS

51

to be illiterate in their own language. The majority of the BOP owners (64%) reported to have
some kind of occupation such as unskilled worker (23.3%), skilled worker (15.3%) or selfemployed in informal or agricultural business (15.3%). The remaining 36 percent included
housewives, students, job seekers and the retired. Only one in three people (33.7%) reported to
have regular income while the rest had irregular (23.4%) or no income (42.9%). Hence, the
aforementioned demographic characteristics indicate the sample consists of the lower
socioeconomic category of the population.
As for the mobile use behavior, almost all mobile owners use it for making or receiving
phone calls (99.5%). There was a high use of missed-calls for signaling a caller’s intention
(68.1%). Nevertheless, the use of non-voice functions was considerably lower than voice-related
uses. In relation to the non-voice features on mobile phones, the BOP mobile owners’ devices
were equipped with an average of eight features on their phone out of the fifteen suggested
functions on mobile phones7 (mean = 8.0, S.D.=3.49) and ever used around three to four features
on their mobile phone (mean =3.5, S.D.=2.87).
Overall, the sample characteristics were generally consistent across the five nations
except a few notable differences. India had slightly more males than females whereas the
Pakistanis sample had more urban residents working as skilled laborers. The household monthly
income and general access to facilities were higher in Sri Lanka and Thailand than other three
countries. The respondents with no formal education were lower in these two nations but the
percentage of people proceeding to secondary school or higher education was similar in all five
nations. Regarding mobile use patterns, almost all the respondents used a pre-paid mobile service.
The use of missed-call was notably low in Thailand but their use of SMS was close to the

7

This may include the respondents who may have such features but are unaware their presence on their phone.

52

average. On the other hand, Bangladesh users tend to use a mobile for mostly voice calls and
missed-call signs and their use of SMS was lower than other countries.
As shown in Table 4, the rankings of the most used mobile features other than voice call
show a notably different use pattern among the BOP mobile owners to typical Western users. For
instance, alarm clock was ranked as the most widely used feature and flashlight as the second
most used due to the limited availability of electricity or facilities connected with electricity (e.g.
toilet, street lights). On the other hand, multimedia features (e.g. camera, music player, video
player) and data-service related features (e.g. downloading, browser, MMS) were either not
present or not used by the majority.

Table 4: Ranking of the Most Used More-than-Voice Mobile Features
Feature Ever Used

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15

Alarm Clock
Flashlight
SMS
Games
Calendar
Calculator
FM radio
Music player
Camera
Video player
Connectivity
MMS
Download
WEB/WAP
browser
Converter
Total

Feature Present

Counted
N
2143
1931
1878
1819
1703
1490
1459
1308
1166
719
495
255
183

% Total
Respondents
49.4%
44.5%
43.3%
41.9%
39.2%
34.3%
33.6%
30.1%
26.8%
16.6%
11.4%
5.9%
4.2%

Counted
N
4446
2310
4177
4159
4344
4015
2497
1856
1684
1327
1075
1059
728

% Total
Respondents
92.1%
47.8%
86.5%
86.1%
90.0%
83.1%
51.7%
38.4%
34.9%
27.5%
22.3%
21.9%
15.1%

172

4.0%

685

14.2%

120
16841

2.8%

767
35127

15.9%

* Multiple response questions

53

3.2. Measures
All measures, except Perceived Benefits (PB), were adapted from prior studies with
minor changes in wording (Table 5). Rather than contextualizing the measures for the BOP, the
study chose to use widely used or sufficiently validated measures from the existing TPB and
information systems literature. By doing so, it intended to test the applicability of the Westernorigin standard measures as well as to facilitate the comparison of the findings between the overstudied developed world and the under-studied BOP population. Since most BOP respondents
are unfamiliar with complex psychometric questions, 5-point Likert scales were used instead of
the conventional 7-point ones (1 = strongly disagree, 3 = neither agree nor disagree, 5 =
strongly agree).
The measures for Perceived Benefits (PB) of mobile phones were constructed based on
the previous qualitative studies on mobile use at the BOP by LINREasia (LIRNEasia, 2009). The
researchers initially listed 12 exploratory items with 5-point Likert-scales asking about any
perceived improvement on various aspects of everyday life including economic, efficiency,
information, social benefits (1=No change; 5=Greatly improved). Exploratory factor analysis
(EFA) with varimax rotation was performed to identify the emergent factors with statistical
significance. Both scree plot and eigenvalue criteria clearly indicated two factors: economic and
efficiency benefits made up 60.6 percent of the total variance that was explained. Four items with
cross-loadings and weak scores were removed by the recommended rule of thumb (Costello and
Osborne, 2005). The EFA results and factor loadings are stated in Table 6.

54

Table 5: Measurement Items
Construct

Perceived
Usefulness

Name
PU1
PU2
PU3
PU4
SN1

Subjective
Norm

SN2
SN3
PBC1

Perceived
Behavioral
Control
Perceived
Ease of
Use

PBC2
PBC3
PEOU1
PEOU2

PV1
PV2

I think mobile services offer values for money

PEOU3

Trust
TRU2

Social
Support

SUPT1
SUPT2
INT1

Intention

Using a mobile phone is entirely within my control
I find mobile phone easy to use
I think learning how to use mobile phone is easy to me
My interaction with mobile phone is clear and
understandable
Based on my experience with mobile phone, I know it
provides good service
Based on my experience with mobile phone, I know it
is trustworthy
I think mobile services are reasonably priced

TRU1

Price
Value

Item
I find mobile phone to be useful in my life
Using mobile phone increases my chances of achieving
things that are important to me
Using a mobile phone helps me accomplish things
more quickly
I find a mobile phone gives me useful information
People who influence my behaviors think I should use
a mobile phone
I use a mobile phone because I want to use the same
service people around me use
I use a mobile phone because it is common to use it in
my community
I have the resources necessary to use a mobile phone
I have the knowledge and ability necessary to use a
mobile phone

INT2

When I have problems in using a mobile phone I can
get help from my friends/family members
When I have problems in using a mobile phone I can
get help from the service providers or experts
I intend to continue using mobile services in the future
I expect that I would use mobile phone frequently in
near future

55

Sources

Davis et al.
(1989)

Ajzen &
Fishbein
(1980)
Kim et al.
(2008)

Ajzen &
Fishbein
(1980)

Davis et al.
(1989)

Kim et al.
(2008)
Kim et al.
(2008);
Venkatesh et
al. (2012)
Teo&Pok,
(2003
Ajzen &
Fishbein(1980)
Davis et al.
(1989)

Table 6: Exploratory Factor Analysis on Perceived Benefits of Mobile
Factor
Economic
Benefits

Efficiency
Benefits

Name
ECB1
ECB2
ECB3
ECB4
EFB1
EFB2
EFB3
EFB4

Items
Ability to find out about employment/work opportunities
Ability to save money
Ability to access finance
Ability to plan and make decisions relating to your livelihood
Ability to reduce travel
Ability to act or contact others in emergency
Efficiency in your day to day work
Relationships with family and friends

Factor loadings
.814
.792
.781
.768
.820
.789
.747
.638

Dependent Variables: The dependent variable measures the binary division between
voice-only use of the mobile device versus SMS use as a proxy for more-than-voice mobile use.
SMS is currently the most popular means of delivering mobile for development (M4D) services
due to its wide availability, technical robustness and cost-efficiency of the service. However, the
low adoption and use of SMS is the biggest barrier to such M4D projects (Kang et al., 2013) and
investigating SMS use can draw important practical implications to the M4D practitioners.
Second, SMS is a text-based data service which makes a conceptual distinction between multifunctional mobile devices to a portable telephone. Before entering the mass adoption of
smartphone, SMS use indicates a stepping stone to more advanced data communication via
mobile and it is worth paying attention to its use behavior. Hence, a multiple-response question
asked ‘which of the following services have you used on your mobile phone?’ including voice
calls (99%), SMS to communicate with people (32.1%), SMS to access information and banking
services (0.8%), none of above (1%). A binary variable (Mean=0.34, SD=0.47) is created from
those who answered voice only and those who used SMS services (voice only=0, SMS use=1).

56

3.3. Data Preparation
Missing values. Unlike the typical problems in data from developing countries, the level of
missing values was relatively low as the data was collected via face-to-face interviews by trained
enumerators. The initial missing value analysis (i.e. missing value patterns) revealed 198 cases
with no response to all the Likert-scale independent variables. After removing the blank cases,
the percentages of missing values were at a negligible level for all variables (average 1.5 percent)
except some of the Perceived Benefits (PB). Little’s MCAR test was conducted to check if the
missing values were completely random and, as the test found the missing pattern to be not
random (chi-square = 4702.1, df=3739, sig=.000), mean replacement was preferred to listwise
deletion. On the other hand, 6 out of 10 Perceived Benefit (PB) items had more than 10 percent
missing values ranging from 14 to 21 percent of the total sample. It was identified that all the
questions with high missing values were related to employment and the patterns of missing
values were structured because housewives or the unemployed may have felt ineligible to answer
such questions. Hence, multiple imputation was conducted using demographic variables (gender,
age, education, employment, have income) and the rest of the relatively complete PB items.
Outliers. Tukey’s outlier labeling rules (Tukey, 1977) were applied to eliminate outliers for two
continuous variables (monthly mobile expenditure and household income). The inter-quartile
range (Q3-Q1) was multiplied by 1.5 to determine the upper and lower boundary (Tukey, 1977).
Additionally, the box-plot method was used to screen out the remaining outliers manually. The
procedure yielded in the total sample size of 4,062 comprising five countries: Bangladesh
(25.3%, N=1,027), Pakistan (24.2%, N=983), India (23.9%, N=969), Sri Lanka (16.5%, N=670)
and Thailand (10.2%, N=413).

57

Country Effects. As for the psychometric measures, Intra-class correlation (ICC) coefficient of
all the psychometric scores between the five studied countries was tested. The test shows that the
correlation coefficient is in an acceptable range (β=.755, F-test=5.456, p<.000) on the studied
measures of motivations among the five countries. At the same time, country effect was further
controlled in the analysis using structural equation modeling.

3.4. Reliability and Validity
Confirmatory factor analysis (CFA) using SPSS-AMOS (version 21.0) was employed to
assess the dimensionality and validity of all measures and latent factors. Item reliability was
assessed by reviewing standardized factor loadings and Cronbach’s alpha values for their internal
consistency. All items were loaded significantly (p <.000) with the loadings over .60 in general
(Table 7). Construct reliability, convergent and discriminant validity are shown in Table 8 and 9.
Cronbach’s alpha values and composite reliability were all satisfactory, exceeding the rule of
thumb of .70 (Churchill, 1979), except social support which is a formative construct measuring
informal support from friends and family and formal support from service providers.
Convergent validity was assessed by Average Variance Explained (AVE) scores. Most
constructs exceed the recommended rule of thumb of .50 while a few are close to the .50
benchmark8. Discriminant validity was evaluated by following the rules proposed by Fornell and
Larcker (1981) who suggested that the correlation between any two constructs should be lower
than the square root of AVE of the individual construct. Table 7 shows the construct correlation
matrix with the diagonal of the square root AVE. All factor correlations are lower than the
diagonal values, thus the constructs are distinct and safe for further analyses.
8

It has been suggested that AVE is related to the sample size as it can be improved by dropping cases (Ping, 2009;
Efron, 1981). Considering the large sample size, the marginally low AVE scores are accepted in this study.

58

To assess the overall fit of the measurement model, a set of goodness-of-fit indices are
reviewed: RMSEA (.040), CFI (.943), GFI (.954), TLI (.931) and a Chi-square value of 3076.6
(df = 409, p<.000). It should be noted that the Chi-square value is easily distorted by the large
sample size and thus an inadequate indicator for a sample size over 400 (Bentler & Bonett, 1980;
Kaplan, 2009; Arbuckle, 2011). Following the criteria suggested by Bentler & Hu (1999) and
Hooper et al. (2008)9, a combination of the above fit indices suggests the measurement model
has a good fit.
Table 7: Standardized Factor Loadings
Item
No.
1
2
3
4

PU

SN

PBC ECB

EFB

TRST

PEOU

PV

SUPT

INT

.700
.721
.690
.700

.627
.840
.771

.595
.708
.702

.649
.663
.540
.634

.733
.750

.771
.788
.720

.796
.821

.615
.730

.766
.774

.670
.617
.768
.731

* PU: Perceived Usefulness; SN: Subjective Norms; PBC: Perceived Behavioral Control; ECB: Economic Benefits;
EFB: Efficiency Benefits; TRST: Trust; PEOU: Perceived Ease of Use; SUPT: Social Support; INT: Intention

Table 8: Construct Statistics
Mean
PU
SN
PBC
ECB
EFB
PEOU
TRST
PV
SUPPORT*
INTENT

4.25
3.81
3.99
2.39
3.41
4.17
4.16
3.89
3.79
4.22

Cronbach’s
alpha
.796
.780
.706
.791
.711
.802
.709
.791
.620
.744

Composite
Reliability
.796
.793
.709
.791
.716
.804
.710
.791
.624
.744

AVE

√AVE

.494
.564
.481
.488
.389
.578
.550
.654
.624
.593

.703
.751
.694
.699
.623
.760
.742
.809
.675
.770

* Formative construct

9

Among the various fit indices available on structural equation modeling technique, it is generally recommended
that Root Mean Square Error of Approximation (RMSEA) value below .06, Comparative Fit Index (CFI) greater
than .90, Goodness-of-fit (GFI) above .090, Tucker-Lewis Index (TLI/NNFI) close to .095 are to be used as cut-off
criteria for determining how the model fits the sample data (Bentler & Hu, 1999; Hooper et al., (2008); Hair et al.
(2010))

59

Table 9: Factor Correlation Matrix and Discriminant Validity Assessment

PU
SN
PBC
ECB
EFB
PEOU
TRST

PV
SUPT
INT

PU
SN
PBC ECB EFB
0.703
0.487
0.751
0.610
0.610 0.694
0.312
0.356 0.301 0.699
0.375
0.145 0.281 0.378 0.623
0.701
0.489 0.689 0.234 0.294
0.642
0.510 0.619 0.291 0.349
0.475
0.548 0.577 0.279 0.217
0.470
0.378 0.488 0.260 0.148
0.653
0.653 0.584 0.274 0.324

60

PEO
U

TRU
S

0.760
0.618
0.466
0.356
0.543

0.742
0.548
0.452
0.645

PV

SUP
T

INT

0.809
0.558
0.483

0.675
0.390

0.770

CHAPTER 4: ANALYSIS

The empirical analysis consists of three sets of statistical assessments. First, the study
examined how the TPB’s main behavioral motivations and their related antecedents influence the
use intention by testing the proposed model with the full dataset via Structural Equation
Modeling (SEM). In fact, the path between use intention and actual use behavior is dropped in
the first analysis due to the limitations of cross-sectional data which does not allow timestructure involved in the process10. Second, the study evaluated whether there exist variability
between different socio-demographic groups. TwoStep Cluster analysis was performed to
identify different demographic groups within the BOP, and the differences in path coefficients
between multiple sub-groups were analyzed. Third, a supplementary analysis using a logistic
regression was conducted to explore the relationship between behavioral intention and actual
behavior in the population of the current study.

4.1. Path Analysis
The SEM technique was chosen for this analysis as it allows a simultaneous testing of multiple
direct and indirect relationships in a complex model. SEM enables us to examine latent variables
such as psychological perceptions and motivations that are difficult to observe with a single
measure but common in social sciences. The technique also improves the accuracy of model
testing by separating the errors within measurement model and structural path analysis so that the
10

The intention is measured by user’s willingness to continue using a technology and therefore the use behavior
should be measured with a time-interval through a follow-up survey. Although it is a widespread practice to
examine the intention-behavior path with a cross-sectional data in the field of TAM- or TPB-based studies, the
dissertation acknowledges the given limitations and therefore H1d and H1e are tested in a supplementary analysis
which explores the intention-behavior relationship using three different variables of more-than-voice use
behaviors.

61

measurement errors are screened out via confirmatory factor analysis before model testing. In
addition, when extending the existing theoretical model, SEM helps researchers identify the best
fitting model to the given dataset by offering multiple fit indices and tests to compare different
model modifications.
The results of the structural model testing are shown in Figure 6. The multiple fit indices
suggest a good model fit: CFI=.937, RMSEA=.049, GFI=.966, IFI=.937 and a Chi-square value
of 1617.2 (df=80, p-value <.001). The model achieved an R2 value of .404 for perceived
usefulness, .346 for perceived behavioral control, .297 for intention to use mobile phone.
Figure 6: Structural Model Testing

Overall, the model explained successfully the multiple direct and indirect factors
influencing the formation of behavioral intention. There was strong statistical support for all
three TPB constructs, suggesting that the TPB would be a suitable theoretical explanation of the
cognitive process behind user’s behavioral intention (H1 supported). Perceived usefulness (PU)
has the strongest effect on the behavioral intention (H1a: β=.359, p-value <.001) while subjective
62

norms (SN) also exhibit direct positive effects (H1b: β=.146, p-value <.001) and perceived
behavioral control (PBC) (H1c: β=.203, p-value <.001). The results suggest that the BOP mobile
owner’s intention to use mobile phones is primarily driven by the instrumental values of mobile
phones rather than social and contextual factors.
All the proposed antecedents exhibited statistically significant effects on PU and PBC.
Perceived ease of use (PEOU) exhibited the strongest association with both PU (H2d: β=.411, pvalue <.001) and PBC (H2e: β=.384, p-value <.001), indicating ‘ease-to-use’ feature is the most
important factor relating both to the instrumental values and enabling contexts behind mobile
use. It also implies that PEOU is empirically associated with both technological utilities and
contextual factors, conflicting to the conventional findings of the role of PEOU in the TAMbased studies. PU is also positively associated with the level of trust the users have over mobile
services (H2c: β=.241, p-value <.001). Relatively weaker but statistically significant
relationships were also found between PU and economic benefits (H2a: β=.077, p-value <.001)
and efficiency benefits (H2b: β=.119, p-value <.001) and. PBC was also found to be positively
associated with both social support (H2f: β=.136, p-value <.001) and price value (H2g: β=.241,
p-value <.001).

4.2. Multi-Group Comparison
Next, the study examined whether and where the effect of socio-demographic factors
exerts its influence on the given model (H3a and H3b) by applying the model up to behavioral
intention into different sub-groups. In identifying the sub-demographic groups, it employs a
novel approach to defining a multi-dimensional aspect of demographic characteristics. As
discussed earlier, people possess multiple demographic and socioeconomic identities that are
inter-related with each other. Defining the sub-group by a single measure, such as gender or age,
63

would miss out too much information on the user’s multi-dimensional demographic
characteristics.
In this regard, Two-Step cluster analysis provides an appropriate tool to explore multidimensional demographic clusters in the dataset (Nourusis, 2003; Bacher et al, 2004). Cluster
analysis is a widely-used exploratory technique to classify the sample groups by similarities in
attributes and distance between clusters based on a log-likelihood function. Among the various
clustering methods, Two-Step clustering can handle both continuous and categorical attributes
and reduces arbitrariness by adopting a clustering criterion such as Schwarz’s Bayesian inference
criterion (BIC) or Akaike’s information criterion (AIC) (Okazaki, 2006; Bacher et al, 2004).
Thus, a Two-Step cluster analysis on SPSS (version 21.0) is performed using the
following categorical variables: gender, marital status (married, single), age (teen, young adults,
older adults, elders), education (no education, primary school, secondary or more), occupation
(unemployed, unskilled, skilled laborers, self-employed), literacy, walking time to the nearest
town (less than 15 minutes, 15 to 30 minutes, 30 to 60 minutes, more than an hour) and monthly
household income.
As shown in Figure 7, four distinct groups emerged from the cluster analysis. Overall, the
clustering model reached the fair fit with a Silhouette measure of cluster cohesion and separation
of 0.5 (Kaufman & Rousseeuw, 1990). First, the Working Husband group (N=1,521) was
defined with the sharing characteristics of married (82%) males (100%) with an occupation as
skilled laborers (35%). Second, the Housewife group (N=1,046) consists of married (97%)
females (100%), unemployed (70%) with relatively higher household income (USD 140.8).
Third, the Student group contains unmarried (93%) people or teenagers (50%) who are
unemployed (66%) with a higher education of secondary school or more (52%) and also has a

64

higher household income (USD 140.7). Finally, the Poorest group contains illiterate (82%) older
adults (53%) with no formal education (100%) with an occupation as unskilled laborers (38%),
living in rural (36%) with a considerably lower household income (USD 94.4).
Using the four sub-groups, a multi-group comparison was conducted on AMOS version
21.0 and the statistical differences between the path coefficients were compared by z-test (Table
10). First, the study tests whether socio-demographic effects moderate the overall model by
comparing the coefficient differences between the full-sample model (default model) with each
socio-demographic group. Table 8 shows the z-test results on the co-efficient differences
between the default model and the four sub-groups. Overall, the statistically significant
differences are found in some of the paths (H1a, H1b, H3d, and H3g) and, therefore, the model is
partially moderated by the socio-demographic effects (H3a supported).
Figure 7: Four Sub-groups from the Cluster Analysis

65

Supporting the H3b, the moderation effects are partially present in all three parts
(antecedents, main constructs and BI-B path). As for the main TPB constructs, the poorest group
is less driven by PU (z-score = -2.474, p-value <.05) while the student group is more strongly
influenced by SN (z-score = 2.331, p-value <.05). As for the antecedents, only the effect of
PEOU on PU was found to be different among all the sub-groups. Housewives were more
strongly influenced by PEOU in their perception on usefulness (z-score=3.904, p-value <.001)
while the poorest group was least affected by PEOU (z-score= -3.011, p-value <.001). On the
other hand, the poorest group’s perception of their behavioral control was more strongly affected
by social support (z-score= 2.404, p-value <.05) than other groups. In addition, the model’s
explanatory power differs between the sub-demographic groups. The R2 value of BI was the
Table 10: The Path Coefficients and R2 Difference among Sub-groups11

H1a
H1b
H1c
H2a
H2b
H3c
H3d
H3e
H3f
H3g

PU – BI
SN-BI
PBC-BI
EBC-PU
EFB-PU
TRUST-PU
PEOU-PU
PEOU-PBC
PV-PBC
SUPPORT-PBC
R2
PU
PBC
BI

ALL
Estimates
0.409***
0.116***
0.199***
0.011***
0.028***
0.214***
0.364***
0.397***
0.184***
0.103***

Husband
z-score
0.666
-0.651
-0.966
-0.136
-0.377
-0.292
1.86*
-1.644
0.04
-0.256

Housewife
z-score
1.514
-1.215
1.278
-0.651
-1.075
-0.78
3.904***
1.632
0.216
-1.605

Student
z-score
-1.095
2.331**
-0.574
0.012
1.032
0.447
-2.743***
0.316
-0.505
0.225

Poorest
z-score
-2.474**
-0.231
1.002
0.131
0.465
0.362
-3.011***
0.291
0.226
2.404**

.404
.346
.287

.450
.318
.257

.425
.358
.350

.360
.335
.301

.361
.413
.256

* p<.10, ** p<.05, *** p<.001

11

z-score values were calculated by the comparison of the path coefficients between the default model and each subsample group.

66

highest among the housewife group (R2=.350) and the lowest among the poorest group
(R2=.256). The proposed antecedents to PU were most applicable in explaining the variance
among the husband group (R2=.450) while those of the PBC were best suitable to explain the
poorest group’s motivations (R2=.413). Hence, the results indicate that the different sociodemographic groups have different workings of their motivations behind PU, PBC and Intention.
Next, the detailed comparison between the selected sub-groups was analyzed. First, in
order to examine the gender-related difference, the husbands and housewife groups are compared
in details (Figure 8). Based on the z-test scores, four paths were found to be significantly
different, particularly regarding the effects of PEOU and PBC. The effects of PEOU on both PU
and PBC were stronger among the housewife group than the husbands with z-score value of
2.247 (p <.05) and 2.596 (p<.001), respectively. The effect of PBC was also marginally stronger
among the housewives in relation to intention (z-score=1.806, p<.10).
Figure 8: Path Comparison 1 - Husband vs. Housewife Group

67

Second, the working husband group and the poorest groups were compared in detail
(Figure 9). Overall, three paths were found to be different in a statistically significant way. The
husband group was more driven by PU than the poorest group (z-score=-2.642, p<.001) and their
perception on usefulness was also more strongly affected by PEOU than the poorest (z-score= 3.915, p<.001). On the other hand, the poorest group’s perception on behavioral control (PBC)
was more strongly affected by the degree of social support than the husband group (zscore=2.346, p<.05).
Figure 9: Path comparison 2 - Husband vs. the Poorest Group

4.3. Supplementary Analysis: Effects of Behavioral Intention on Behavior
In the above analyses, the dissertation examined the factors influencing the formation of
the intention to use mobile phones and how socio-demographic effects intervene in the model. In
this supplementary analysis, it examines how the intention is translated into actual use of morethan-voice mobile use. The path between the intention and use was dropped in the main analysis
68

due to the limitations of the current secondary dataset. The cross-sectional data does not allow
the time-interval involved in the actualization of the intention. In other words, the intention,
which is measured by user’s willingness to continue using a mobile phone, and the actual
behavior of mobile use cannot be measured concurrently as it requires a certain time lap for the
intention to be realized in the actual use.
When testing the full model with the use, there were only miniscule differences in the fit
indices: CFI=.924, RMSEA=.043, GFI=.961, IFI=.924 and a Chi-square value of 2020.4
(df=125, p-value <.001) and the model achieved an R2 value of .404 for perceived usefulness,
.346 for perceived behavioral control, .297 for intention to use mobile phone. Nevertheless, the
path between the intention and the use was insignificant while there was only a weak but
statistically significant effects from PBC (H1e: β=.054, p<.05) on use. The R2 value for mobile
use was limited to .03, suggesting that a potential limitation that the proposed model explains
adequately the processes involved in ‘the formation of behavioral intention’ to use SMS but fail
to explain ‘the actualization of the intention’.
One possible explanation for this insignificant relationship can be the neglected timeinterval involved in the process between the behavioral intentions to be actualized in the use
behavior. Another can be that there exist additional factors moderating or hindering the
actualization of the behavioral intention. Thirdly there can be a measurement issue of mobile use
to the extent to which the measure of actual mobile use (that is, voice only vs. SMS use) may be
incongruent to the measure of use intention12. Hence, additional analyses were conducted to

12

In fact, due to the limitation of the secondary data, the measure of the use intention refers to the intention to
continue using a mobile phone in general instead of a specific function of mobile phone (i.e. SMS use). However,
mobile use includes not only using a voice call but also using different functions available on mobile phones
including SMS use. Hence, the current measure of the intention does not conceptually incongruent to the measure
of the dependent variable (voice vs. SMS).

69

investigate these possibilities before accepting the non-significant relationship between BI and B
among the BOP mobile owners.
First, in order to examine the measurement issues relating to the use behavior, two
additional dependent variables were created to measure the different aspects of mobile use
(Table 11). The first new dependent variable operationalizes the quantity of mobile use in terms
of the amount of mobile expenditure per month. The second new variable measures the breadth
of use in relation to the features ever used on their personal mobile phones. These additional
variables aim to provide a more accurate measure of mobile use that may be referred in the
measure of the intention to use a mobile phone.
Table 11: Summary of Dependent Variables
Variable

Description
Voice only = 0, SMS use = 1

N

Mean

S.D

Median

4924

0.34

0.47

0

DV1

SMS Use

DV2

Monthly
mobile
expenditure

Value of monthly top-up amount
4226
(see below formula)

4.32

2.74

3.60

DV3

Breadth of
use

Features ever used on mobile
phones when answered
functions are present on their
own phone

3.48

2.87

3.0

4924

The quantity of mobile use (DV2) was created from three questions relating to the top-up
behaviors. Since most of the BOP mobile owners are on pre-paid service plans, there exist no
accurate data for monthly mobile expenditure at the level of BOP. The questions include ‘what
was the amount of your last top up/reload?’, ‘considering your normal usage, can you please tell
me from the day of your last top up how long do you expect your top-up will last?, and ‘how long
do you usually keep your phone with a zero balance before topping up?’. The local currency was

70

converted to USD Purchasing Power Parity (PPP) conversion rate for private consumption13 as
of 2011 (World Bank, 2011). Then, the monthly mobile expenditure was calculated by the
following formula:

/ 12

The breadth of mobile use (DV3) was created by counting the number of functions they
ever used on their mobile phones. As most mobile users in developing countries still own a basic
feature phone14, the researcher created a selected list of 15 most widely available applications on
general feature phones, including SMS, alarm clock, calendar, simple games, calculator,
converter, camera, video, music player, connectivity (USB slot, Bluetooth, Infra-red), MMS,
Web/WAP browser and downloading feature15. Since not all of such features are available on the
respondents’ phone and their usage depends on the availability, two multiple-response questions
contained in the survey were used ‘which of these features does your mobile handset have?’ and
‘which of these features have you ever used?’ Then the number of ever used features was
counted contingent upon such a feature being marked as present on their device.
Table 12 shows the correlation table between the three dependent variables. DV1
correlates low with DV2 as most features do not necessarily require additional fees. DV1 also
correlates low with DV3 which assumes that SMS use is still low among this population and

13

PPP conversion factor indicates the number of units of a country’s currency required to buy the same amounts of
goods and services in the domestic market as US dollar would buy in the US, which is a more accurate approach
to compare the cost of the same services and goods. The 2011 PPP rate for the studied countries can be found at
http://data.worldbank.org/indicator/PA.NUS.PRVT.PP
14
The current data shows only 4.5 percent of the total mobile owners possess a phone with touch screen (a proxy for
smartphone).
15
Voice call is excluded as all mobile owners are assumed to receive or make a call.

71

SMS cost can be cheaper than a few minute long voice calls. The correlation of DV2 and DV3 is
.37 which indicates that the two variables are marginally correlated but can still be considered as
distinct measures of mobile use.
Table 12: Pearson’s Product Moment Correlation among Dependent Variables
DV1
Voice vs. SMS

DV2
Mobile
expenditure

DV1

1

DV2

.151***

1

DV3

.367***

.177***

DV3
Used
features

1

*** p-value <.001, ** p-value <.05

A structural model was tested using the three dependent variables using the full sample.
Table 13 shows only the path coefficients relating to intention to use and the R2 values of mobile
use. The results show that the BOP user’s intention is irrelevant to their mobile monthly
expenditure (DV2: not supported) while the level of intention has a significant but negative
relationship with the diversity of use (DV3: supported but negative). Hence, all three measures of
mobile use, including voice/SMS use, the quantity of use, the breadth of use, have failed to
explain the positive relationship between the behavioral motivation to use mobile and the actual
use of mobile. Hence, the results reduce a potential risk involved in inadequate or inconsistent

Table 13: Model Testing with Multiple Mobile Use Measures
R2

Hypotheses

DV1 (voice vs. SMS)

Estimates
(p-value)
.036 (.215)

.030

Not supported

DV2 (expenditure)

.009 (.943)

.024

Not supported

DV3 (breadth of use)

-.289 (.000)

.010

Supported but negative path

INT – USE

72

measure of mobile use and suggest that the non-significant path of BI-B may relate to further
factors excluded in the proposed model. In other words, at the BOP, even though they have a
high intention to use mobile phone, their actual use may be hindered or facilitated by further
factors.
Next, the study investigated possible moderation effects between intention and behavior.
According to Fishbein (2000) who applied the TPB in the context of HIV/AIDS intervention
programs, behavioral intention can be hindered by two factors: skills & ability and
environmental constraints. He suggested that people have formed the desired intention but may
not be acting on it due to a lack of skills or the presence of surrounding constraints (Fishbein,
2000; Fishbein & Yzer, 2003; Fishbein & Cappella, 2006). Among the various constraints,
hence, the study focuses on the lack of skills as a moderating factor in the actualization of
behavioral intention to use a mobile phone.
While mobile use appears to be seemingly intuitive and easy for most people, using
more-than-voice functions actually requires sequential skills. For instance, to send or receive a
SMS message, users need to know how to browse the menu, open the application, type or
comprehend the message and manage received messages. However, the level of knowledge and
skills to use mobile phones has not yet been fully explored, particularly in the context of the BOP
mobile users. In the absence of a comprehensive set of objective measures of the user’s technical
skills to use a mobile phone, the degree of self-efficacy, which is an indirect measure from one’s
assessment on their own ability, can be the best available measure of one’s own capabilities
relating to mobile use. Self-efficacy is originally defined as ‘one’s own belief in his or her
capability to organize and execute a particular course of action’ (Bandura, 1997) and. When the
concept was transferred in the field of information systems (IS), the notion became adjusted to

73

more specific measure to address the users’ own technical ability to use a certain technology
(Compeau & Higgins, 1995). Previous studies found that self-efficacy is a significant factor
associated with user adoption of mobile banking (Luarn & Lin, 2005) and SMS advertising
(Zhang & Mao, 2008).
Based on this rationale, the study adopts mobile efficacy as an additional factor
moderating the relationships between behavioral intention to use SMS and actual use behavior
since the construct is a good proxy to indicate the level of user’s perceived skills to use mobile
phones in relations to their surrounding resources. The measures of mobile efficacy are shown in
the Table 14. An interaction term with intention and mobile efficacy was created where meancentering was used to prevent multicollinearity.
Table 14: Mobile Efficacy Measures and Item Statistics
Name

EFF1
Mobile
Efficacy

EFF2

EFF3

Measures
I am confident of using a mobile
phone if someone showed me
how to do it first
I am confident of using a mobile
phone if I could ask someone
for help when I got stuck
I am confident of using a mobile
phone if I had a lot of time to
try and use it

Loading

Mean

Alpha

CR

AVE

3.92

.740

.738

.485

.641

.706

.739

Logistic regression was performed with behavioral intention (BI), mobile efficacy (EFF),
interaction term (BI x EFF) and control variables (a selection of demographic variables) as well
as the dependent variable (SMS use). As shown in Table 15, the intention has a significant but
negative relationship with SMS use (β=-.538, p<.05) when all other things are considered equal.
Also, mobile efficacy has no significant relationship with SMS use on its own. However, the
74

interaction term of behavioral intention and mobile efficacy (BI x EFF) has a significant effect
on predicting SMS use (β=.129, p<.05). The results indicate that, when a high level of intention
is combined with high level of mobile efficacy, people are more likely to use mobile phones for
more than just voice (odds ratio =1.138).
Nevertheless, the relatively low pseudo R-square values (Nagelkerke R-square = .175,
Cox & Snell R-square=.126) and the large odds ratio of the constant term (Exp(b)=161.6)
indicate that there can be more factors relating to environmental constraints affecting the BOP
mobile owners’ more-than-voice use behavior. Further attention should be paid on exploring
potential barriers existing between the BOP user’s intention and actual use behavior.

Table 15: Effects of Intention, Mobile Efficacy and Interaction on Mobile Use
Variables

Estimates

S.E

Wald

Odds ratio

-0.538**

.222

5.859

.584

-0.389 (n/s)

.253

2.357

.678

BI x EFF

0.129**

.058

4.893

1.138

Education

0.876***

.082

113.475

2.402

Gender

-0.133 (n/s)

.088

2.287

.875

Age

-0.055***

.004

198.335

.947

Rural

-0.610***

.073

70.021

.543

Employment

0.007 (n/s)

.093

.005

1.007

5.085**

1.905

7.124

161.646

BI
EFF

Constant

Chi-square=16.7 (df=8, p<.05), Pseudo R square = .175
*** p-value <.001, ** p-value <.05

75

CHAPTER 5: DISCUSSION

5.1. Key Findings
This dissertation examined the motivational factors affecting more-than-voice mobile use
among the mobile owners at the Bottom of the Pyramid in South Asia. While mobile ownership
is rapidly increasing among the poor in developing countries, there is still a paucity of research
exploring their mobile use behaviors. In particular, mobile use of this newly joined user group
has been mostly understood at a descriptive level often in relation to their socioeconomic
constraints, and their psychological motivations and their internal workings in the cognitive
process have been largely unexplored. Furthermore, despite the mounting interests in providing
mobile for development (M4D) services using non-voice features of mobile technology, little
attention is paid on their more-than-voice use of mobile phones such as SMS or application uses.
The dissertation aimed to fill the current research gaps both in communication researchers and
M4D practitioners. Guided by the well-known behavioral theory of the TPB and TAM, the
dissertation explored the motivational drivers and their relatively degree of influences behind
mobile use behavior at the BOP. On top of the three fundamental behavioral motivations
(technological utilities, social influence and contextual factors) proposed by the adoption theories,
the study also examined the antecedents explaining ‘what makes a mobile phone useful’ and
‘what contributes to the enabling contexts for mobile phone use’ among the BOP mobile owners.
In addition, it investigated whether and where socio-demographic factors intervene in the
workings of the motivational drivers by comparing the results between different sociodemographic groups.

76

Table 16: Summary of Hypotheses Testing
H
From
Main TPB Model
H1a
Perceived Usefulness (PU)
H1b
Subjective Norms (SN)
H1c Perceived Behavioral Control (PBC)
H1d

Behavioral Intention (BI)

To
Behavioral
Intention (BI)
Use Behavior (B)

H1e Perceived Behavioral Control (PBC)
Antecedents
H2a
Economic Benefits (ECB)
H2b
Efficiency Benefits (EEB)
Perceived
H2c
Trust (TRUS) Usefulness (PU)
H2d
Perceived Ease of Use (PEOU)
H2e
Perceived Ease of Use (PEOU)
Perceived
H2f
Social Support (SS) Behavioral Control
H2g
Price Value (PV)
(PBC)
Moderation Effects
H3

Moderation effects of socio-demographic factors

Coefficient

Status

.359***
.146***
.203***
n/s

Supported
Supported
Supported
Not Supported

.054**

Supported

.077***
.119***
.241***
.411***
.384***
.136***
.241***

Supported
Supported
Supported
Supported
Supported
Supported
Supported

See Table 8

Supported

* significant at 0.10 level, ** significant at 0.05 level, *** significant at 0.01 level

As shown in the summary of the hypothesis testing in Table 16, the dissertation provides
empirical evidence for the following. First, the findings suggest that the overall theoretical power
of the TPB extends to the population at the BOP and it successfully explains the factors
influencing the formation of behavioral intention. The analysis found that their intention to use a
mobile phone (BI) is more strongly driven by the instrumental utility (PU) than social or
contextual factors. This contrasts to previous adoption studies in developing countries, which
emphasized the importance of subjective norms or skills and knowledge amongst non-western
users (Mao et al., 2005; Straub et al., 1997). While the poor’s behaviors are commonly assumed
to be determined by their structural constraints by the researchers focusing on the demographic
conditions, this result suggests that they also make a rational evaluation on the instrumental

77

utilities of mobile phones and this utilitarian evaluation exerts stronger effects on behavioral
intention than the availability of resources or knowledge.
Second, the dissertation extends the TPB framework with several antecedents to PU and
PBC, which are specifically related to the BOP mobile use contexts. The analysis of the
antecedents found that all hypothesized factors are significantly associated with PU and PBC. In
particular, it found ease of use (PEOU) exerts the strongest effects among the hypothesized
factors. It also offers a rather unconventional finding on the effect of PEOU in that PEOU is
conceptually and empirically related to both perceived instrumental utilities (PU) and contextual
control (PBC). In contrast to the conventional TAM studies, the dissertation hypothesized that
PEOU is a compounded construct associated with both instrumental evaluation of technology
(user friendliness) and users’ contextual evaluation of their skills and the resources required to
use a technology. By specifying it as the antecedents to both PU and PBC, the analysis proves
that PEOU has dual effects on both the level of instrumental evaluation and enabling conditions
among the BOP users.
Along with PEOU, the study also identified several other antecedents and their relative
importance relating to PU and PBC. Trust in mobile services was also positively related to the
level of instrumental utilities. To a lesser degree, the experienced benefits of mobile phones have
significant associations with PU where benefits of improved efficiency have a stronger
relationship than economic benefits. The findings suggest that the BOP mobile owners in general
evaluate mobile phones’ usefulness in relation to the easy-to-use features, trustworthiness and
efficiency gains in managing everyday events. On the other hand, both the level of perceived
cost (price value) and the availability of technical support in social contexts (social support) were
found to be statistically significantly associated with the enabling contexts for mobile use.

78

Third, the study points out the theoretical weakness of the TPB in consideration to the
effects of socio-demographic factors. While the TPB assumes that demographic effects exist as a
background factor underlying the belief systems, the study found that there exist the moderation
effects of socio-demographic factors not only in the belief systems (i.e. antecedents) but also in
the paths from the three main determinants (PU, SN, PBC) to BI. The dissertation took a novel
approach to identifying the socio-demographic effects by grouping the sample with a multidimensional demographic measures using TwoStep cluster analysis. It identified the significant
differences in the path coefficients and the model’s explanatory powers between the four
demographic groups. The findings indicate that the TPB should pay further attention to
specifying the characteristics of the studied population and the moderating effects of background
factors in order to improve its explanatory accuracy as well as to draw practical implications for
the intervention.
In addition, the study examined the degree of different effects of the motivational factors
on BI between sub-demographic groups. In comparing the husbands and housewives, it found
that the housewives’ intention to use mobile is more influenced by the contextual factors (PBC)
than the husbands group. And their perceptions on both PU and PBC are more strongly driven by
PEOU than their counterpart. When comparing the husbands to the poorest group, the study
identified that husbands’ intention is more influenced by the level of PU and this PU is in turn
also more strongly driven by PEOU than the poorest group. On the other hand, the poorest group
found the social support as the stronger indicator to PBC than the husband group. In other words,
it suggests that the instrumental utility in relation to the level of easy-to-use convenience are
important factor behind the husband group’s motivation to use mobile phone while the poorest

79

group is affected by the contextual factor in relation to the availability of technical support in
social settings.
Finally, the study identified that the TPB-based framework is appropriate for explaining
the formation of behavioral intention, but is critically limited in explaining the actual use
behavior among the population at the BOP. As the relationship from behavioral intention (BI) to
behavior (B) has been widely proved in social psychology, researchers in the field of technology
adoption studies occasionally omit the BI-B path in their analysis as if it is a given fact (Lee et al.,
2003). However, in the population at the BOP, the study found that the user’s behavioral
intention to use mobile does not lead to the actual behavior of more-than-voice use and the BI-B
path might be hindered by external factors. The supplementary analysis reduced the possibility of
measurement errors in mobile use behaviors, and explored the potential moderation effects with
mobile efficacy. The results found, although tentatively, that only the interaction term with
intention and mobile efficacy was significant when all other factors are considered equal. This
means that the BOP mobile users tend to use more for more than voice if their high intention to
use mobile is combined with high level of efficacy in using mobile phones. Nonetheless, the
dissertation concludes that, in the population at the BOP, there may exist further barriers
hindering the actualization of the behavioral intention and further studies should pay more
attention on exploring external factors moderating the BI-B path.

5.2. Research Implications
The dissertation contributes to the existing literature in communications, information
systems and ICT for development research in the following ways. Focusing on the previously
untapped populations at the BOP, the dissertation fills the critical research gaps in our

80

knowledge on the newly joined mobile users from developing countries. The poor’s mobile use
behaviors have been mostly approached from ethnographic or qualitative methods while a few
quantitative studies conducted in developing countries often use the samples from university
students or company employees who in general belong to the social elites in low-income
countries. By using the large random sample specifically targeting the BOP mobile owners and
applying the rigorous statistical analyses, the dissertation provides empirical evidence on mobile
use behaviors that are generalizable to the BOP population in South Asia.
In particular, the dissertation explored the understudied topic of psychological
motivations and cognitive process behind mobile use at the BOP. By applying the westernoriented technology adoption theories in the context of the BOP, the study identified that the
TPB-based framework can be successfully extended to explaining and predicting the mobile
users at the BOP but its theoretical power is limited in explaining their actualization of the
intention. In particular, it found that, as with the users in the developed world, the poor’s
intention to use mobile phone is primarily formed by the instrumental utilities than social or
contextual factors although this intention is, unlike the Western users, supposed to be hindered
by the external barriers. Hence, the dissertation concludes that three constructs specified in the
TPB and TAM, in particular the effects of instrumental utilities, can be universal in explaining
technology use. However, it draws a caution for researchers who attempt a mere replication of
the Western-oriented theories in the context of the poor and calls for further attention on
exploring the barriers in the actualization of the behavioral intention.
The study also contributes to the theoretical advancement of the TPB by identifying its
theoretical limits in under-estimating the socio-demographic effects as a background factor. In
this regard, the study proves that there exist the socio-demographic factors intervene in the TPB

81

model not only in relation to the belief systems but also to the effects of the main determinants.
For the theory’s explanatory accuracy and its potential practical implications, the dissertation
recommends the TPB researchers to be specific in defining the population characteristics as well
as the intervening effects of the socio-demographic factors in their analysis. It also points out a
problem in focusing only on one side of coin, in that, the researchers interested in motivational
factors often stifle out demographic factors while those interested in policy implications focus
only on demographic factors as the key determinants of social inequality. The dissertation calls
for researchers to see the both sides of the coin to be closer to the reality.
For the TAM researchers, the dissertation suggests that PEOU can be a compounded
construct which embraces both technology’s characteristics and user’s skills to use a technology
and it provides empirical evidence that PEOU is strongly associated with both PU and PBC in
the population of the BOP. Further attention is needed to evaluate the conceptual, empirical or
measurement issues relating to the PEOU when it is applied to the general users outside
organizational contexts. In addition, the dissertation enriched the possible practical implications
that can be drawn from the adoption theories by exploring the antecedents to what makes a
mobile useful and what contributes to the enabling factors. Accepting the critics (Bagozzi, 2007;
Benbast & Barki, 2007), the dissertation suggests that, if the theory exists not for the sake of the
theory, it should pay attention less on the theory’s parsimoniousness and more on further
indicators explaining the main constructs.
The dissertation also draws several implications for the M4D practitioners. First, the
dissertation calls for a conceptual shift in perceiving the poor as active, independent and rational
individuals. It assumes that the poor also makes decisions based on their own preferences and
evaluations of their resources. By investigating the internal functions of their motivations, the

82

study found that the poor’s intention to use mobile phones is mostly driven by the instrumental
utilities and less by the contextual constraints. Hence, the practitioners should pay primary
attention to providing the useful services from the user’s point of view before considering the
issues in cost and skills. It also found that easy-to-use features are the fundamental drivers
behind their perception on usefulness and contextual control, followed by service reliability and
perceived price value.
Second, by identifying a multiple demographic groups and their different functions of the
motivational factors, the study found that there exist much socio-demographic variations within
the poor and such different affects the functions of the studied motivations within each
demographic group. For instance, it found that the husband group is more driven by the
instrumental utilities while the housewives are by conceptual control. It also found that the
poorest consider the availability of technical support at social settings as an important factor to
enabling their mobile use. This draws an important implication for the development community
that ‘under-2-dollar-a-day’ is a simplistic criterion to categorize the poor and further attention is
needed to understand the diverse socio-demographic groups comprising the poor. It also suggests
that the M4D practitioners should be clear in understanding their target population’s
characteristics and needs. For instance, if they design a job-search service for working males, it
should focus more on delivering outcome utilities of the services. On the other hand, if they
target female housewives, the service should consider supporting their technological skills and
limited resources. For the services targeting the rural poor, the program should be equipped with
technical support in informal and social manner such as hiring a local assistant or campaigner.

83

5.3. Limitations and Further Research
The dissertation has several limitations. First, although the current study successfully defined the
process behind the formation of the behavioral intention, the model fails to explain the full set of
factors contributing to more-than-voice mobile uses. Further studies should explore the external
and internal barriers hindering the behavioral intention at the level of BOP. Second, the
dissertation examined the motivations behind general mobile phone use, which limits the
specificity of the practical implications it can draw. Further study may use the proposed
framework to assess more specific use behaviors particularly within the practice of targeted M4D
intervention programs (mobile banking, mobile for health programs, etc.). Such an approach will
enable the practitioners to monitor and evaluate how motivational factors influence the adoption
of a specific service. Third, although the dissertation makes an initial attempt to examine the
utilities involved in mobile phone use, it limits the scope of the benefits as direct instrumental
gains due to the limitations of using a secondary dataset. Further study should include a broader
set of utilities perceived from mobile use including entertainment and recreation, emotional
support, security, symbolic status and so forth. Finally, the role of social influence needs more
attention. The study uses a narrow definition of social influence in relation to everyday
interaction in close social relationships. Further research is encouraged to explore the multiple
layers of social effects from vicarious learning, mass media and public campaign, community
norms, social structure and cultural factors.

84

5.4. Concluding Remarks
The contour of mobile communications is rapidly expanding across the globe spreading
into the lives of the previously unconnected populations at the Bottom of the Pyramid. We are in
urgent need of understanding how they adopt and use mobile phones and what it means to have a
mobile phone in their everyday lives. Such stories at the BOP can neither be captured from the
incomplete statistics we get from the ITU or national census in developing countries, nor be fully
understood from scattered case studies. Certainly, there should be more research focusing on this
severely under-represented group of users.
This dissertation made a small but new attempt to understand their mobile use behaviors
by tackling the untapped topic of the motivational drives behind mobile use and extending a
western-oriented technology adoption theory in the context of the BOP. Although the current
study is focused on general use of mobile phone (i.e. SMS) in everyday contexts, the research
model used in this study can be extended to studying different types of mobile use behaviors.
Also it can provide a useful framework for project evaluation if the practitioners wish to examine
the factors behind their success or failure of a specific mobile-based intervention programs.
In this regard, the dissertation raises a critical concern towards mere replication of the
western-oriented TPB- or TAM-based models in the different socio-cultural contexts of BOP
users. As the dissertation reveals, such adoption theories may be adequate to explain the
psychological process within the formation of the user intention, but they fail to provide the full
picture of how such intention is actualized or hindered among the BOP users. It also shows that
the BOP is not a homogenous group and there exist considerable socio-demographic differences,
which results in different functions of motivational drives. Hence, further research is encouraged
to embrace mixed-methods to discover hidden factors unique to the users at the BOP that can

85

complement the conventional adoption models. Understanding people precedes establishing a
theory and we are still at the very beginning of developing an understanding of these new users
at the BOP. Many missing pieces of the jigsaw puzzle remain, leaving it far from completion.
The field is in need of more diverse, creative, pro-active studies that can collectively build up
evidence helping us understand what is really happening at the Bottom of the Pyramid.

86

APPENDIX

87

APPENDIX: Detailed Review of Technology Adoption Theories
Diffusion of Innovation (DOI)
Diffusion of Innovation (DOI) is one of the most extensively researched and interdisciplinary
theories in studying people’s adoption behaviors. Diffusion theory originated in anthropology
and rural sociology in the 1950s and became refined as a comprehensive theory by Everett
Rogers in 1962. Since then, DOI became a dominant paradigm in the adoption research,
particularly between the 1970s and 1990s, across several disciplines such as sociology,
communications, public health, education, marketing and management, and so forth.
According to Rogers, (1995), innovation is referred to as ‘an idea, practice or object that
is perceived as new by an individual or other unit of adoption’. Thus, DOI’s research topics are
not limited to new technology, but cover to a wide range of new ideas, goods and service such as
fertilizers, health information, pedagogical approaches, media campaigns as well as ICTs. On the
other hand, diffusion is ‘a process in which an innovation is communicated through certain
channel over time among members of a social system’. Diffusion is also perceived as a special
type of communication process in which participants exchange messages about perceived new
ideas in order to reach a mutual understanding. In this sense, DOI is primarily interested in the
‘process’ of how innovation is accepted, communicated and disseminated in personal,
interpersonal and social domains. Therefore, DOI has a broader research scope surrounding
adoption behavior, including not only the individual’s adoption decision makings (i.e. the
attributes of innovation and the innovation-decision process), but also the S-curve rate of
diffusion in a given society, the adopter characteristics of adopters (e.g. innovators, earlyadopters, early-majority, later-majority, laggards), diffusion networks, the role of change agents,
and the consequences of diffusion. In sum, DOI theory aims to understand the universal
mechanisms behind human behaviors and social changes surrounding innovation.
88

Regarding the adoption process at the individual level, DoI analyzes user’s perceptions
on new technology/innovation as fundamental drivers behind adoption decision, which is similar
to other theories discussed in this study. However, while others offer a three-stage framework
(perception-intention-decision), DOI defines the decision process in five stages over time:
knowledge-persuasion-decision-implementation-confirmation. This process is largely influenced
by five factors, including perceived attributes of innovations, type of innovation decision,
communications channel, the nature of the social system, and change agents (Rogers, 2003,
p.222).
Figure 10: The Attributes of Innovation

Relative
Advantage
Complexity
Compatibility
Trialability
Observability
Voluntariness

The degree to which an innovation is perceived as better than the idea it
supersedes
The degree to which an innovation is perceived as difficult to understand
and use
The degree to which an innovation is perceived as being consistent with the
existing values, past experiences, and needs of potential adopters
The degree to which an innovation may be experimented with on a limited
basis
The degree to which the results of an innovation are visible to others
The degree to which adoption is perceived as being an optional innovation
decision

89

Here, what attracted most attentions among the technology adoption researchers is the
‘attributes of innovation’ as it explains what characteristics of technology lead to adoption or
rejection and how to improve technological or service features. Originally, Rogers proposes five
attributes of innovation: relative advantage, compatibility, trialability and observability. Later,
DOI researchers from IS research defined three more variables: image, voluntariness, result
demonstrability (Moore & Benbasat, 1991).
Overall, DOI offers a comprehensive framework to study what perceived characteristics
of technology triggers adoption and their relative effects on the decision-making. In contrast to
other theories seeking model parsimony, DOI proposes the eight specific factors covering the
multifaceted utilities of new technology, including instrumental and technical utility (relative
advantages, compatibility, complexity), social and symbolic utility (observability, result
demonstrability, image) and accessibility (trialability, voluntariness). It also takes account of
social and cultural factors such as interpersonal and mass communication channels, social system
and socioeconomic variables (Rogers, 2003, p.170). While other theories are micro-level theories
focusing purely on individuals, DOI can be categorized as a ‘middle range theory’ which
organizes a body of findings from replicated studies into a structured system of principle
(Merton, 1968).
However, while Roman (2003) suggests there is room for further diffusion study in the
context of ICTD, the theory holds certain limitations to be applicable to the M4D. The DOI
theory assumes that adoption is only a matter of time, and the late majority and laggards will
eventually follow the adoption process. This is a rather one-dimensional view on technology
adoption. It may explain the almost universal adoption of mobile phones or the Internet but fails
to explain the varying degree of ICT utilization to the extent of the digital divide of skills and use

90

(Van Dijk, 2003). As Rogers summarizes the common criticism to DOI, past diffusion research
tended to perceive innovation as a positive and beneficial force (Pro-Innovation bias, p. 106).
The rejection or reinvention are briefly mentioned but not integrated into the main theory (Al
Qeisi, 2009) and, in some cases, the late adopters or laggards are individually blamed as
‘irrational’ or ‘traditional’. Rather than analyzing the social and systemic faults, the research
tradition of DOI tend to view an individuals responsible for their own problems (Individualblame bias, p. 118). Second, the DOI frameworks have been extensively replicated both in
developed and developing contexts. The research has entered a static period both theoretically
and methodologically to the extent that Rogers himself calls for no “more-of-the-same research”
(Wirth et al, 2008; Meyer, 2004; Rogers, 2003). The established findings on the similarity of the
adoption process between the developed and developing countries adds no value to the M4D
research to discover different mobile use behaviors among the poor. Finally, while the theory
involves social and cultural factors within its conceptualization, it is unclear how these
surrounding factors are related in the adoption process, which makes simultaneous empirical
assessment too challenging.

Theory of Reasoned/Planned Action (TRA/TPB)
Theory of Reasoned action (TRA) is one of the most influential theories explaining primary
causes of human behavior. As opposing to the dominant notion in social psychology of the 1970s
that ‘attitude causes behavior’ (Peak, 1955), Ajzen and Fishbein (1975, 1980) explains that
behavior is determined by one’s behavioral intention to perform a target behavior, and this
intention is affected by two motivational factors specific to the behavior in question: Attitude (A)
and Subjective Norm (SN). These motivational factors are estimated by the salient beliefs one
may have about the behavior and one’s subjective assessment on their relative effects. For
91

instance, Attitude is a function of the sum of salient behavioral beliefs weighted by the
evaluation of the outcome. SN is also the sum of salient normative beliefs about other people’s
opinions, multiplied by one’s motivation to comply with it.
The TRA provides parsimonious and powerful causal explanations to human behavior in
general (Conner & Armitage, 1998). However, its application is limited to behaviors under
volitional control. In other words, TRA tends to lose its explanatory power if the behavior
requires particular skills, resources, information or opportunities that are not freely available all
the time (Fishbein, 1993; Conner & Armitage, 1998; Ajzen, 1991).

Figure 11: The TRA and the TPB

Attitude (A)
Subjective Norm (SN)
Perceived Behavioral
Control (PBC)
Actual Control

The degree to which a person has a favorable or unfavorable
evaluation or appraisal of the behavior in question
The perceived social pressure to perform or not to perform the
behavior
The perception that performance of a specific behavior is within a
person’s volitional control
The extent to which a person has the skills, resources, and
other prerequisites to perform a given behavior
92

Hence, Ajzen (1991) adds the third variable called Perceived Behavioral Control (PBC)
on top of the TRA’s two motivational factors (A and SN) and introduces Theory of Planned
Behavior (TPB). PBC is the individual’s perception of the degree that behavior in question is
easy or difficult to perform (Ajzen, 1991). These perceptions include both internal control factors
(e.g. emotion, physical and mental deficiencies, skills, abilities, willpower etc.) and external
control factors (e.g. available resources, opportunities, barriers, supports or objections from
others etc.). And some of the salient control factors are weighted by the perceived power of each
factor to facilitate or inhibit the behavior in question (Ajzen, 1991; Conner & Armitage, 1998).
The TPB has several theoretical advantages. First, the TPB offers a parsimonious
framework with only three constructs that can explain all kinds of human behaviors regardless
research contexts. Thus, the TPB has been widely applied to study a variety of topics across
disciplines such as information systems, social psychology, marketing, health communications,
media effects, marketing, environmental studies and so forth. Second, the TRA and the TPB
offer an open framework which a set of salient beliefs are unique to each behavior in question
(Ajzen 1991; Fishbein & Cappella, 2006). Ajzen (1991) also emphasizes that the TPB is in
principle open to additional predictors if it can capture a significant proportion of the variance.
Therefore, researchers are encouraged to explore different behavioral, normative, control belief
systems surrounding a particular behavior and customize the model specific to their research
topics.
Nevertheless, the TRA and the TPB are also without its criticism. First, as they assume
that individuals are rational decision-makers, they fail to take account of irrational, automatic or
habitual behaviors (LaRose, 2011; Hartmann, 2009; Ajzen, 2011). Second, they also assume that

93

everyone is motivated to perform a certain behavior and thus fails to explain non-adoption or
reject of behavior (Talyor & Todd, 1995; Al-Qeisi, 2009). Third, it presents insufficient
conceptual understanding and measurement to SN and PBC as well as affective motivations
(Manstead, 2011; Ajzen, 2011, Armitage & Corner, 2001).

Technology Acceptance Model (TAM/TAM2/UTAUT)
The Technology Acceptance Model (TAM) is one of the most widely used theoretical
frameworks in explaining user adoption of information technology. It is also one of the most
critically debated frameworks within the IS research (Lee et al, 2003; Bagozzi, 2007). Since its
inception by Fred Davis in 1986, the TAM has undergone a series of major theoretical
advancement throughout model introduction, validation, extension and elaboration (Lee et al.
2003). Due to relentless efforts to expand its theoretical power, the model has continued to
evolve into succeeding models such as the TAM2 (2000) and the UTAUT (2003).
Its research domains were also expanded from a word processor system at work to
various information end-user technologies and mass market services such as personal computers,
mobile phones, emails, e-commerce and banking, e-government, 3G mobile services, etc. (Lu et
al, 2008; Yousafzai et al, 2010, Park et al, 2011). Its target population has also expanded from
employees at a firm level to households and individual consumers (Yousafzai et al, 2010; LopezNicolas et al, 2008) while the geographic area of interest has broadened from the US to other
regions, including developing countries (Brown, 2002; Park et al, 2009).
The TAM aims to understand “what causes people to accept or reject information
technology” (Davis, 1989), and explains the cognitive process of decision-making in user
adoption behavior. Among the various theoretical foundations (Davis, 1989), the TAM is most
firmly grounded in the tradition of TRA and shares several key assumptions (Ajzen & Fishbien,
94

1980; Davis et al, 1989). First, the TAM accepts the TRA’s underlying assumption that users
make a rational assessment, and adoption behavior is under one’s volitional control (Ajzen,
1991). Second, it accepts the fundamental assumption of the TRA that behavior (B) is explained
directly by one’s behavioral intention (BI) to use the given technology (Assumption 1: B → BI).
Third, it agrees that this intention is, in turn, determined by one’s attitude about the technology in
question, formed by one’s salient beliefs and motivational factors.
Figure 12: The Original TAM

Perceived Usefulness
(PU)
Perceived Ease of Use
(PEOU)

the user’s subjective probability that using a specific application
system will increase his/her job performance
the degree to which the user expects the target system to be free of
efforts

Nevertheless, originally developed in an organizational setting with a newly introduced
information technology, the TAM departs from the TRA in the specification of the attitudinal
factors. Replacing Attitude in the TRA, TAM postulates two key fundamental concepts that
determine user’s BI: Perceived Usefulness (PU) and Perceived Ease of Use (PEU). The TAM
posits that BI is a joint function of PU and PEU (Assumption 2: BI = PU + PEU), while ease of
use can enhance perceived job performance or save time, thus affects the degree of PU
(Assumption 3: PEU→PU). In addition, the TAM assumes that the two fundamental
determinants of PU and PEU are universal, and are offered a priori (Davis et al, 1989).

95

However, the closed nature of the TAM with the pre-determined set of variables and
measurements made it inevitable for researchers to release an upgraded version. Researchers
soon tested additional variables in different research contexts and began to dispute the TAM’s
validity (Agarwal & Prasad, 1999; Karahanna et al, 1999; Gefen & Straub, 1997; Karahanna &
Limayem, 2000; Hong & Tam, 2006; Yousafzai et al, 2010, Lee et al, 2003). For instance,
Venkatesh & Morris (2000) found significant effects of gender and experience. They also
restored Subjective Norm back in the model, which was intentionally omitted in the original
model due to the weak theoretical foundations and empirical evidence (Davis et al, 1989;
Venkatesh & Davis, 2000). Accordingly, the extended version named as the TAM2 was created
with seven new variables as antecedents to PU and two moderators borrowed from the TRA and
DOI: Subjective Norms, Image, Job Relevance, Output Quality, Result Demonstrability,
Experience and Voluntariness.
Figure 13: TAM2 Extension

96

Similarly in 2003, TAM researchers announced a new succeeding model called the
Unified Theory of Acceptance and Use of Technology (UTAUT) based on thorough analysis of
the eight comparable models, including the TRA, TPB, Decomposed TPB, DOI, SCT,
Motivational Model, the Model of PC Utilization (Venkatesh et al. 2003). The UTAUT is the
outcome of the TAM scholars’ effort to create a holistic and powerful model (Venkatesh et al,
2003). It consists of four key perceptions on technology covering similar constructs in other
models: performance expectancy (embracing perceived usefulness, attitude, relative advantage,
compatibility, outcome expectancy), effort expectancy (perceived ease of use, complexity),
social influence (observability, subjective norms), and facilitating conditions (perceived
behavioral control, self-efficacy). It also acknowledged the effects of user differences in relation
to demographics and specified age, gender, and experience as moderators. Based on longitudinal
Figure 14: The Unified Theory of Acceptance and Use of Technology (UTAUT)

Performance
Expectancy
Effort Expectancy
Social Influence
Facilitating Conditions

the degree to which an individual believes that using a specific
system will increase his or her performance
the degree to which an individual believes that using a particular
system would be free of efforts
degree to which an individual perceives that important others believe
he/she should use the new system
the degree to which an individual believes that an organizational and
technical infrastructure exists to support use of the system

97

Figure 15: The UTAUT2

fieldworks in multiple settings, the UTAUT was proven to outperform other models by
explaining 69 percent of the variance in behavioral intention and about 50 percent for the
behavior (Venkatesh et al, 2003).
Most recently, Venkatesh et al. (2012) released the UTAUT2 which targets general
consumers’ adoption and use of end-user technologies and services. The new model retains the
four main UTAUT predictors (Performance expectancy, Effort expectancy, Social Influence,
Facilitating Conditions) and three moderators (age, gender, experience). Then it re-specified the
model with the following three points: first, it added three new variables relevant to general
consumers (Hedonic Motivation, Price Value, and Habit) as predictors to BI. Second, the effects
of the moderators have been adjusted from moderating all the BI predictors in the previous
model to affect only FC and three newly added variables while PE, EE, SI are intact from the
moderators. At the same time, the moderator Voluntariness is dropped. Third, the path from FC

98

is previously directed only to B but now it has both paths to BI and B. Similarly, Habit also
exerts influence on both BI and B while experience also moderates BI - B path.
The 20 year long survival of the TAM in the fast-moving IS research demonstrates the
theoretical strength of the model. First, unlike other comparable theories discussed in this study,
the TAM is specifically tailored for the context of information technology. Second, the original
TAM offers an intuitive and parsimonious framework to understand a complex nature of
technology adoption. Third, it is also a powerful and robust theory which consistently
outperformed other comparable theories in terms of the explanatory power (Talyor & Todd,
1995; Bagozzi, 2009; Yousafzai et al, 2010). Fourth, the TAM also provides a set of sound
measurement scales that are well-proven in terms of item reliability as well as convergent,
discriminant, factorial, and nomological validity. All these advantages allow researchers to
replicate the study easily to other domains, and thus enable the accumulation of empirical
findings and the generalization of the theory. Overall, the TAM and its successors brought
significant contribution to the adoption research by uniting several comparable theories and
enhancing research rigors with consistency and standardization (Lee et al, 2003).
Nevertheless, as the TAM created a dominant research paradigm in the 1990s and 2000s,
the theory raised several criticisms. These criticisms are noteworthy if the TAM/UTAUT to be
an appropriate vehicle to explain mobile adoption among the poor in developing countries. First,
except the recent UTAUT2, critics caution on replicating the model for different contexts other
than organizational setting (Bouwman & Wijngaert, 2009; Karahanna & Limayem, 2000; Moon
& Kim, 2001). Originally designed for employees’ adoption of company IT systems, these
theories assume training and open access that are usually offered free of charge while the task is
primarily related to one’s profession. On the other hand, a mobile phone is a multi-purpose

99

consumer technology used in everyday context. Unlike company employees or university
students, BOP users may lack material resources, skills, and prior experience with relevant
technology (Musa, 2006; Park et al, 2009). Second, the TAM-based models are limited in their
relevance for practitioners since they do not explain ‘what actually makes a system useful’
(Benbasat & Barki, 2007; Lee et al, 2003), nor do they explain what contributes to the
facilitating conditions of a technology’s adoption. Suggesting that people will adopt a technology
if they find it useful and easy or if enabling conditions are present does not provide actionable
implications to practitioners who want to understand how to build a system with optimal results
in terms of both adoption and continued optimal use. Third, most TAM studies use convenient
samples of university students whose lifestyles tend to be congruent whereas their perceptions on
technology may differ from the general public (Legris et al, 2003; Lee et al, 2003). Fourth, the
extensions of the TAM have been also criticized for being over-crowded with concepts borrowed
from other theories in a way of mere ‘patchwork’ which fails to provide sufficient theoretical
grounds for such addition (Bagozzi, 2007; Benbasat & Barki, 2007). At the same time, the
simplistic view on technology use as a mere quantity of time or frequency does not help us
understand the consequence of such use (Benbasat & Barki, 2007).

Social Cognitive Theory
Social Cognitive Theory (SCT) is another social psychology theory to understand the overall
functioning of human motivation, belief systems and behavior. Founded by Bandura (1986),
SCT offers a set of influential conceptualization including triadic reciprocality, self-efficacy,
outcome expectations, self-reflection, self-regulation, etc. These constructs are now widely
adopted in media effect studies, marketing and technology adoption. According to Bandura
(2001), SCT assumes human is a proactive actor who constantly tries to make sense of his
100

thoughts, actions and life circumstances by organizing, regulating and evaluation their own
action. It also takes a holistic approach (i.e. reciprocal triadic determinism) that any behavioral
or social phenomenon is a form of continuous bidirectional interactions between personal
factors, behaviors and environmental influences (Pajares, 2002). In other words, rather than
ruling out socio-structural influences (e.g. class, gender, race, value, culture etc), SCT posits that
such structural and contextual factors are naturally internalized by each individuals over time and
reflected in one’s own perceptions (i.e. self-efficacy). In addition, SCT explains human
incentives consist of five factors: sensory/novel, social, monetary, activity and status incentives,
which creates expected outcomes of future behaviors.
Among others, self-efficacy is the most widely used construct in technology adoption
studies. Self-efficacy refers to ‘one’s belief in capability to organize and execute a particular
course of action’ (Bandura, 1986). It is a dynamic self-evaluation formed through four agentic
experiences: enactive learning (one’s own performance), vicarious experience (observational
learning), social persuasion, and physical and emotional status (Bandura 1997; Pajares, 2002).
Self-efficacy then influences the expected outcomes of one’s future behavior. According to
Bandura, experiencing these four sources occurs within a tight network of sociostructural system
and, throughout this iterative process, people internalized the sociostructural influences. Unlike
self-esteem or self-confidence, therefore, self-efficacy is a form of an optimal balance between
one’s ability and desire, or a reflection of both personal and sociostructural influences. In the
domain of technology use, the concept was later further elaborated as ‘computer self-efficacy’
(Compeau & Higgins, 1995) or ‘Internet self-efficacy’ (Larose & Eastin, 2004) to measure the
one’s assessment in skills and capability to perform a task on computer or the Internet.

101

SCT is a comprehensive framework that captures not only human motivations but also
environmental factors influencing behaviors. In particular, by measuring self-efficacy as a proxy,
it enables us to estimate the immeasurable operation of biological, social, political and cultural
factors which affected an individual throughout his or her lifetime. In the absence of accurate
measures of the objective skills and ability, self-efficacy also provides the most accurate
estimation of one’s own capability. In this sense, the self-reported measures have more validity
than other approaches.
Nevertheless, the main constructs and assumptions of SCT are not delineated in one
comprehensive model while most paths are bi-directional or multi-directional. Also, the theory
does not offer sufficient guidelines on measurements and apart from self-efficacy, much of
theoretical conceptualizations are still in the process of development particularly in the context
of technology use. Thus, it is still theoretically challenging to apply the full scope of SCT in
explaining a particular behavior.

102

BIBLIOGRAPHY

103

BIBLIOGRAPHY

Abraham, R. (2007). Mobile Phones and Economic Development: Evidence from the Fishing
Industry in India. Information Technologies and International Development, 4(1), 5-17.
Agarwal, R. & Prasad, J. (1999). Are Individual Differences Germane to the Acceptance of New
Information Technologies? Decision Sciences, 30(2), 361–391.
Agüero, A., Silva, H. De, & Kang, J. (2011). Bottom of the Pyramid Expenditure Patterns on
Mobile Phone Services in Selected Emerging Asian Countries. Information Technologies
& International Development, 7(3), 19–32.
Ajzen, I. & Fishbein, M. (1980) Understanding Attitudes and Predicting Social Behavior,
Prentice-Hall, Englewood Cliffs, NJ.
Ajzen, I. (1991). The Theory of Planned Behavior. Organizational Behavior and Human
Decision Processes, 50(2), 179–211.
Ajzen, I. (2011). The Theory of Planned Behavior: Reactions and Reflections. Psychology &
Health, 26(9), 37–41.
Aker, J. (2008). Does Digital Divide or Provide? The Impact of Cell Phones on Grain Markets in
Niger, Center for Global Development Working Paper No. 177
Aker, J., & Mbiti, I. (2010). Mobile Phones and Economic Development in Africa. Center for
Global Development Working Paper, 24(3), 207–232.
Al-Qeisi, K. I. (2009). Analyzing the Use of UTAUT in Explaining an Online Behavior: Internet
Banking Adoption. Unpublished Doctoral Dissertation. Brunel University, UK.
Arbuckle, J. L. (2011). IBM SPSS Amos 20 user’s guide. Armonk, NY: IBM.
Armitage, C. J., & Conner, M. (2001). Efficacy of the Theory of Planned Behaviour: a metaanalytic review. The British journal of social psychology, 40(4), 471–99.
Bacher, J., Wenzig, K., & Vogler, M. (2004). SPSS TwoStep Cluster: A First Evaluation.
Lehrstuhl für Soziologie.
Bandura, A. (1986). Social Foundations of Thought and Action: A Social Cognitive Theory,
Prentice-Hall, NJ.
Bandura, A. (2001). Social Cognitive Theory of Mass Communications. In Bryant, J. & Zillman,
D. (eds.) Media Effects: Advances in Theory and Research (2nd ed., 121-153) Hillsdale,
NJ: Lawrence Erlbaum.

104

Banerjee, A, & Duflo, E. (2011). Poor Economics: A Radical Rethinking of the Way to Fight
Global Poverty. Public Affairs: New York.
Bar, F., Pisani, F., & Weber, M. (2007). Mobile Technology Appropriation in a Distant Mirror:
Baroque Infiltration, Creolization and Cannibalism. May, 15(2007), 20-21.
Barrantes, R., & Galperin, H. (2008). Can the Poor Afford Mobile Telephony? Evidence from
Latin America. Telecommunications Policy, 32(8), 521–530.
Bauer, J. M. (2010). Current Status of ICT and Telecommunications in LDCs. In ITU (2011) The
role of ICT in Advancing Growth in Least Developed Countries: Trends, Challenges and
Opportunities 2011. International Telecommunications Union: Geneva.
Benbasat, I., & Barki, H. (2007). Quo Vadis, TAM? Journal of the Association for Information
Systems, 8(4), 211-218.
Bentler, P.M. & Bonnet, D.C. (1980). Significance Tests and Goodness of Fit in the Analysis of
Covariance Structures. Psychological Bulletin, 88 (3), 588-606.
Bentler, P.M & Hu, L.T. (1999). Cutoff Criteria for Fit Indexes in Covariance Structure
Analysis: Conventional Criteria versus New Alternatives. Structural Equation Modeling,
6 (1), 1-55.
Best, M. L., & Kumar, R. (2008). Sustainability Failures of Rural Telecenters: Challenges from
the Sustainable Access in Rural India (SARI) Project. Information Technologies and
International Development, 4(4), 31–45.
Bimber, B. (2000). Measuring the Gender Gap on the Internet, Social Science Quarterly, 81 (3),
September 2000.
Blattman, C., Jensen, R., & Roman, R. (2003). Assessing the Need and Potential of Community
Networking for Development in Rural India. The Information Society, Special Issue :
ICTs and Community Networking, 19(5), 349-364.
Boateng, R. (2011). Mobile phones and Micro-Trading Activities – conceptualizing the link.
Info, 13(5), 48-62.
Bonfadelli, H. (2002). The Internet and Knowledge Gaps: A Theoretical and Empirical
Investigation. European Journal of Communication, 17 (1), 65–84.
Brown, I. T. J. (2002). Individual and Technological Factors Affecting Perceived Ease of Use of
Web-based Learning Technologies in a Developing Country. The Electronic Journal of
Information Systems in Developing Countries, 9, 1-15.
Burrell, J. (2010). Evaluating Shared Access: social equality and the circulation of mobile
phones in rural Uganda. Journal of Computer-Mediated Communication, 15(2), 230–250.
105

Busy, E. P. (2000). Social Access to the Internet, the Harvard International Journal of
Press/Politics, 5(1), 50-61.
Carroll, J., Howard, S., Peck, J., & Murphy, J. (2007). From Adoption to Use: the Process of
Appropriating a Mobile Phone. Australasian Journal of Information Systems, 10(2), 38–
48.
Chaudhuri, A. (2012). ICT for Development: Solutions Seeking Problems? Journal of
Information Technology, 27(4), 326–338.
Cartier, C., Castells, M., & Qiu, J. L. (2005). The Information Have-Less: Inequality, Mobility,
and Translocal Networks in Chinese Cities. Studies in Comparative International
Development, 40(2), 9–34.
Castells, M. (2006) The Rise of the Network Society, Oxford University Press: Oxford, UK.
Chew, H. E., Levy, M. R., & Ilavarasan, P. V. (2011). The Limited Impact of ICTs on
Microenterprise Growth: A Study of Businesses Owned by Women in Urban India.
Information Technologies and International Development, 7(4), 1–16.
Chib, A., Wilkin, H., Ling, L., Hoefman, B., & Hajo, B. Van. (2012). You Have an Important
Message! Evaluating the Effectiveness of a Text Message HIV/AIDS Campaign in
Northwest Uganda. Journal of Health Communications, 17(June), 146–157.
Cho, H. (2011). Theoretical Intersections Among Social Influences, Beliefs, and Intentions in the
Context of 3G Mobile Services in Singapore: Decomposing Perceived Critical Mass and
Subjective Norms. Journal of Communication, 61(2), 283-306.
Churchill Jr, G. A. (1979). A Paradigm for Developing Better Measures of Marketing
Constructs. Journal of marketing research, 64-73.
Cialdini, R. B., & Goldstein, N. J. (2004). Social Influence: Compliance and Conformity. Annual
review of psychology, 55(1974), 591–621.
Compeau, D., & Higgins, C. (1995). Application of Social Cognitive Theory to Training for
Computer Skills. Information Systems Research, 6(2), 118–143.
Conner, M., & Armitage, C. J. (1998). Extending the Theory of Planned Behavior: A Review
and Avenues for Further Research. Journal of Applied Social Psychology, 28(15), 1429–
1464.
Costello, A.B. & Osborne, J.W. (2005). Best Practices in Exploratory Factor Analysis: Four
Recommendations for getting the most from your analysis. Practical Assessment,
Research & Evaluation. 10 (7).

106

Crandall, A. (2012). Kenyan Farmer’s Use of Cell Phones: Calling Preferred over SMS. In
Proceedings of Mobile for Development 2012. M4D 2012 28-29 February 2012, New
Delhi, India, Karlstad University, Sweden.
Davis, F. D. (1986). A Technology Acceptance Model for Empirically Testing New End-User
Information Systems: Theory and Tesults. Unpublished doctoral dissertation,
Massachusetts Institute of Technology, Cambridge.
Davis, F. D. (1989). Perceived Usefulness, Perceived Ease of Use, and User Acceptance of
Information Technology. MIS Quarterly, 13(3), 319.
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.
Deursen, a J. a M. van, & Dijk, J. a G. M. van. (2010). Internet Skills and the Digital Divide.
New Media & Society, (December 2010).
DeMaagd, K. & Kang, J. (2010). A Synthesized Model of ICT IN and FOR Development.
Conference paper presented at International Telecommunications Society Asia-Pacific
Regional Conference, Tokyo, June 28-30
DiMaggio, P., Hargittai, E., Neuman, R. and Robinson, J. (2001) Social Implications of the
Internet, Annual Review of Sociology. Vol. 27: 307-336
Dodds, W. B., Monroe, K. B., & Grewal, D. (1991) Effects of Price, Brand, and Store
Information on Buyers’ Product Evaluations. Journal of Marketing Research, 28(3), 307.
Donner, J. (2006). The Use of Mobile Phones by Microentrepreneurs in Kigali, Rwanda:
Changes to Social and Business Networks. Information Technologies & International
Development, 3(2), 3.
Donner, J. (2007). The Rules of Beeping: Exchanging Messages via Intentional "Missed Calls"
on Mobile Phones. Journal of Computer-Mediated Communication, 13(1), article 1.
Donner, J. (2008). Research Approaches to Mobile Use in the Developing World: A Review of
the Literature. The Information Society, 24(3), 140–159.
Donner, J., & Tellez, C. A. (2008). Mobile Banking and Economic Development: Linking
Adoption, Impact, and Use. Asian Journal of Communication, 18(4), 318–332.
Donner, J., & Toyama, K. (2009). Persistent Themes in ICT4D Research: Priorities for Intermethodological Exchange. Proceedings of the 57th session of the International Statistics
Institute, (June), 1–10.

107

Donner, J. (2010). Framing M4D: The Utility of Continuity and the Dual Heritage of “Mobiles
and Development”. The Electronic Journal of Information Systems in Developing
Countries, 44, 1–16.
Donner, J., & Escobari, M. (2010). A Review of Evidence on Mobile Use by Micro and Small
Enterprises in Developing Countries. Journal of International Development, 22, 641–
658.
Duncombe, R. (2012). Understanding Mobile Phone Impact on Livelihoods in Developing
Countries: A New Research Framework. Center for Development Informatics, University
of Manchester, UK
Esselaar, S., Stork, C., Ndiwalana, A. & Deen-Swarray, M. (2007). ICT Usage and Its Impact on
Profitability of SMEs in 13 African Countries, Information Technologies and
International Development, 4(1), Fall 2007, p.87-100.
Fishbein, M. (2000). The Role of Theory in HIV Prevention. AIDS care, 12(3), 273–8.
Fishbein, M., & Cappella, J. N. (2006). The Role of Theory in Developing Effective Health
Communications. Journal of Communication, 56(s1), S1–S17.
Fitzgerald, R., Spriggs, J., & Keosothea, N. (2010). Enhancing Communications in Developing
Countries Using SMS Technology: The Case of Agricultural Value Chains in Cambodia.
International Journal of Continuing Engineering Education and Life-Long Learning,
20(1), 72–83.
Fong, M. W. L. (2009). Digital Divide between Urban and Rural Regions in China. The
Electronic Journal of Information Systems in Developing Countries, 36(6), 1-12.
Forestier, E., Grace, J., & Kenny, C. (2002). Can Information and Communication Technologies
Be Pro-poor? Telecommunications Policy, 26, 623–646.
Fornell, C., & Larcker, D. F. (1981). Structural Equation Models with Unobservable Variables
and Measurement Error: Algebra and Statistics. Journal of marketing research, 382-388.
Furuholt, B., Wahid, F., & Kristiansen, S (2006). Gaming or Gaining? Comparing the Use of
Internet cafés in Indonesia and Tanzania, International Information and Library Review,
40, 129-139.
Furuholt, B., & Kristiansen, S. (2007). A Rural-Urban Digital Divide? Regional Aspects of
Internet Use in Tanzania. Electronic Journal of Information Systems in Developing
Countries, 31(6), 1-15.
Gefen, D., & Straub, D. W. (1997). Gender Differences in the Perception and Use of E-mail: An
Extension to the Technology Acceptance Model. MIS quarterly, 389-400.

108

Gomez, R., Baron, L., & Fiore-Silfvast, B. (2012). The Changing Field of ICTD: Content
Analysis of Research Published in Selected Journals and Conferences, 2000--2010. ICTD
2012 Conference Proceedings, March 12-15, 2012, Atlanta, GA, USA, 65–74.
Gutierrez, L., & Gamboa, L. (2010). Determinants of ICT Usage among Low-Income Groups in
Colombia, Mexico, and Peru. The Information Society, 26(5), 346–363.
Guttman, W. (1986). Telecommunications and Sub-Saharan Africa: The Continuing Crisis.
Telecommunications Policy, 325–340.
GSMA (2006). Regulation and the Digital Divide, Price Waterhouse Coopers/GSM
Associations: London, UK.
GSMA mWomen (2012). Striving and Surviving: Exploring the Lives of Women at the Base of
the Pyramid, GSMA mWomen Program, London, UK
Haddon, L. (2001). Domestication and Mobile Telephony. Paper presented at the ‘Machines that
Become Us’ conference, Rutgers University, New Brunswick, NJ, USA, 18–19 April.
Hair, J. F., Black, W. C., Babin, B. J., & Anderson, R. E. (2010). Multivariate data analysis.
Upper Saddle River, NJ: Prentice Hall.
Hargittai, E., & Hinnant, A. (2008). Digital inequality Differences in Young Adults' Use of the
Internet. Communication Research, 35(5), 602-621.
Hartmann. T. (2009). Action Theory, Theory of Planned Behavior and Media Choice. In
Hartman, T. (ed.) Media Choice: A Theoretical and Empirical Overview, Routledge:
New York.
Heeks, R. (2002). Information Systems and Developing Countries: Failure, Success, and Local
Improvisations. The Information Society, 18(2), 101–112.
Heeks, R. (2007). Theorizing ICT4D Research. Information Technologies and International
Development, 3(3), 1–4.
Heeks, R. (2008). ICT4D 2.0: The Next Phase of Applying ICT for International Development.
Computer, 41(6), 26–33.
Heeks, R. (2010). Do Information and Communication Technologies (ICTs) Contribute to
Development? Journal of International Development, 22(5), 625–640.
Heeks, R. (2012). Deriving an ICT4D research agenda: a commentary on “Information and
communication technologies for development (ICT4D): solutions seeking problems?”
Journal of Information Technology, 27(4), 339–341.

109

Hilbert, M. (2011). Digital Gender Divide or Technologically Empowered Women in
Developing Countries? A typical case of lies, damned lies, and statistics. Women’s
Studies International Forum, 34(6), 479–489.
Hoffman, D., Novak, T & Schlosser, A. (2000). The Evolution of the Digital Divide: How Gaps
in Internet Access May Impact Electronic Commerce, Journal of Computer-Medicated
Communication, Vol.5, No.3.
Hong, S., & Tam, K. (2006). Understanding the Adoption of Multipurpose Information
Appliances : The Case of Mobile Data Services. Information Systems Research, 17(2),
162-179.
Hooper, D., Coughlan, J. & Mullen, M. R. (2008). Structural Equation Modeling: Guidelines for
Determining Model Fit. The Electronic Journal of Business Research Methods. 6(1). 5360.
Horst, H., & Miller, D. (2005). From Kinship to Link-up: Cell Phones and Social Networking in
Jamaica. Current Anthropology, 46(5), 755–778.
Horst, H., & Miller, D. (2006). The Cell Phone: An Anthropology of Communication. Berg
Publishers.
Hsieh, J. J., Rai, A. & Keil, M. (2008). Understanding Digital Inequality: Comparing Continued
Use Behavior Models of the Socio-Economically Advantaged and Disadvantaged, MIS
Quarterly, Vol. 32, No.1, p97-126.
Hsieh, J. J. P.-a, Rai, A., & Keil, M. (2011). Addressing Digital Inequality for the
Socioeconomically Disadvantaged Through Government Initiatives: Forms of Capital
That Affect ICT Utilization. Information Systems Research, 22(2), 233-253.
Howard, P. E., Rainie, L., & Jones, S. (2001). Days and Nights on the Internet: The Impact of a
Diffusing Technology. American Behavioral Scientist, 45(3), 383-404.
Hughes, N., & Lonie, S. (2007). M-PESA: Mobile Money for the “Unbanked” Turning
Cellphones into 24-Hour Tellers in Kenya. Innovations: Technology, Governance,
Globalization, 2(1-2), 63–81. doi:10.1162/itgg.2007.2.1-2.63.
IFC (2011). IFC Mobile Money Study 2011: Summary Report. International Finance
Corporation. Washington DC, USA.
Information Week (2005). Motorola Introduces Dirt Cheap Cell Phone. News article September
21, 2005. www.informationweek.com
InfoDev (2013). ICT for Development: Contributing to the Millennium Development Goals. The
World Bank Group. November 2003. Washington DC, USA

110

InfoDev (2012). Mobile Usage at the Base of the Pyramid in South Africa. InfoDev publication
prepared by Research ICT Africa and Intelcon. December 2012, Washington DC, USA.
InfoDev (2013). Mobile Usage at the Base of the Pyramid in Kenya. InfoDev publication
prepared by iHub Research and Research Solution Africa. January 2013. Washington
DC, USA.
Ito, M. (2005). Mobile Phones, Japanese Youth, and the Re-placement of Social Contact. In
Mobile Communications (pp. 131-148). Springer London.
ITU (2011). The role of ICT in Advancing Growth in Least Developed Countries: Trends,
Challenges and Opportunities 2011. International Telecommunications Union: Geneva.
ITU (2013). Measuring the Information Society 2013. International Telecommunications Union:
Geneva.
Jackson, S. J., Pompe, A., & Krieshok, G. (2011). Things Fall Apart: Maintenance, Repair, and
Technology for Education Initiatives in Rural Namibia. Proceedings of the 2011
iConference. 83–90.
James, J., & Versteeg, M. (2007). Mobile Phones in Africa: How Much Do We Really Know?
Social indicators research, 84(1), 117–126.
Jensen, R. (2007). The Digital Provide: Information (Technology), Market Performance, and
Welfare in the South Indian Fisheries Sector. Quarterly Journal of Economics 122(3):
879-924.
Jung, J.-Y., Qiu, J. L., & Kim, Y.-C. (2001). Internet Connectedness and Inequality: Beyond the
“Divide.” Communication Research, 28(4), 507–535.
Kalba, K. (2008). The Global Adoption and Diffusion of Mobile Phones. Center for Information
Policy Research, Harvard University: New Haven, USA
Kang, J., Cousaris, C. & Maity, M. (2013). Texting among the Bottom of the Pyramid: Enabling
Factors on SMS Use among Low-Income Mobile Users in Asia, a paper presented at the
63rd Annual Conference of the International Communications Association, London, UK,
June 17-21 2013.
Kaplan, D. (2009). Structural Equation Modeling: Foundations and Extensions (Vol. 10). SAGE
Publications, Incorporated.
Kim, S. (2008). Moderating Effects of Job Relevance and Experience on Mobile Wireless
Technology Acceptance: Adoption of a Smartphone by Individuals. Information &
Management, 45(6).

111

Kim, G., Park, S. & Oh, J. (2008). An Examination of Factors Influencing Consumer Adoption
of Short Message Service (SMS).Psychology and Marketing, 25(8), 769–786. John Wiley
& Sons.
Kleine, D. (2010). ICT4WHAT?-Using the Choice Framework to Operationalise the Capability
Approach to Development. Journal of International Development, 22(5), 674–692.
Karahanna, E., & Straub, D. W. (1999). The Psychological Origins of Perceived Usefulness and
Ease-of-Use. Information & Management, 35(4), 237–250.
Karahanna, E., & Limayem, M. (2000). E-mail and V-mail Usage: Generalizing Across
Technologies. Journal of organizational computing and electronic commerce, 10(1), 4966.
Kaufman, L. R., & Rousseeuw, P. PJ (1990) Finding Groups in Data: An Introduction to Cluster
Analysis. Wiley Series in Probability and Statistics. New York: Wiley.
Khodamoradi, S. & Abedi, M. (2011). The Role of Information and Communication
Technologies (ICT) in Agricultural Development. Journal of American Science. 7 (3).
Kraut, R., Olson, J., Banaji, M., Bruckman, A., Cohen, J., & Couper, M. (2004). Psychological
research online: report of Board of Scientific Affairs' Advisory Group on the Conduct of
Research on the Internet. American Psychologist, 59(2), 105.
Kumar, V., & Svensson, J. (2012). Proceedings of Mobile for Development 2012. M4D 2012 2829 February 2012, New Delhi, India, Karlstad University, Sweden
Kuriyan, R., Toyama, K., & Ray, I. (2006). Integrating Social Development and Financial
Sustainability: The Challenges of Rural Computer Kiosks in Kerala. 2006 International
Conference on Information and Communication Technologies and Development, 121130.
LaRose, R. (2010). The Problem of Media Habits. Communication Theory, 20(2), 194–222.
LaRose, R., & Eastin, M. (2004). A Social Cognitive Theory of Internet Uses and Gratifications:
Toward a New Model of Media Attendance. Journal of Broadcasting & Electronic
Media, Online Publication 01 October 2004.
Lee, Y., Kozar, K., & Larsen, K. (2003). The Technology Acceptance Model: Past, Present, and
Future. Communications of the ACM, 12, 752–780.
Leung, L., & Wei, R. (1999). Who Are the Mobile Phone Have-Nots? Influences and
Consequences. New Media & Society, 1(2), 209–226.

112

Lin, A., & Tong, A. (2008). Mobile Cultures of Migrant Workers in Southern China: Informal
Literacies in the Negotiation of (New) Social Relations of the New Working Women.
Knowledge, Technology & Policy, 21(2), 73–81.
Ling, R. (2005). The Socio-linguistics of SMS: An Analysis of SMS Use by a Random Sample
of Norwegians. In R. Ling & P. Pedersen (Eds.), Mobile communications: Renegotiation
of the social sphere (pp. 335 – 349). London: Springer.
LIRNEasia. (2008). Teleuse at the Bottom of the Pyramid Data Booklet: Findings from a SixCountry Study in Emerging Asia. LIRNEasia available at http://lirneasia.net/wpcontent/uploads/2008/04/Teleuse-at-BOP-Data-Booklet-LIRNEasia.pdf
LIRNEasia. (2009). Teleuse@BOP3: A Qualitative Study Center for Knowledge Societies.
LIRNEasia publication prepared by Center for Knowledge Societies: Colombo, Sri
Lanka.
LIRNEasia. (2012). Teleuse@BOP4: A Qualitative Study Center for Knowledge Societies.
LIRNEasia: publication prepared by Center for Knowledge Societies: Colombo, Sri
Lanka.
Livingstone, S., & Helsper, E. (2009). Balancing Opportunities and Risks in Teenagers’ Use of
the Internet: the Role of Online Skills and Internet Self-Efficacy. New Media & Society,
12(2), 309-329.
Lopez-Nicolas, C., Molinacastillo, F., & Bouwman, H. (2008). An Assessment of Advanced
Mobile Services Acceptance: Contributions from TAM and Diffusion Theory Models.
Information &Management, 45(6), 359-364.
Lu, Y., Deng, Z., & Wang, B. (2010). Exploring Factors Affecting Chinese Consumers’ Usage
of Short Message Service for Personal Communication. Information Systems Journal,
20(2), 183-208.
Luarn, P., & Lin, H. (2005) Toward an Understanding of the Behavioral Intention to Use Mobile
Banking. Computers in Human Behavior, 21(6), 873-891.
Madianou, M., & Miller, D. (2011). Mobile Phone Parenting: Reconfiguring Relationships
Between Filipina Migrant Mothers and Their Left-behind Children. New Media &
Society, 13(3), 457–470.
Manstead, A. S. R. (2011). The Benefits of a Critical Stance: A Reflection on Past Papers on The
Theories of Reasoned Action and Planned Behaviour. The British journal of social
psychology, 50(3), 366–73.
Mathieson, K. (1991). Predicting User Intentions: Comparing the Technology Acceptance
Model with the Theory of Planned Behavior. Information Systems Research. 2. 173-191.
113

Melody, W. (1999). Telecom Reform: Progress and Prospects. Telecommunications Policy,
23(1), 7–34.
Merton, R. K. (1968). Social Theory and Social Structure. New York: Free Press.
Mittal, S., Gandhi, S., & Tripathi, G. (2010). Socio-economic Impact of Mobile Phones on Indian
Agriculture. New Delhi: Indian Council for Research on International Economic
Relations.
Molony, T. (2007). I Don’t Trust the Phone; It Always Lies: Trust and Information and
Communication Technologies in Tanzanian Micro- and Small Enterprises. Information
Technologies and International Development, 3(4), 67–83.
Moore, G. C., & Benbasat, I. (1991). Development of an Instrument to Measure the Perceptions
of Adopting an Information Technology Innovation. Information systems research, 2(3),
192–222.
Morawczynski, O. (2009). Exploring the Usage and Impact of ‘Transformational’ Mobile
Financial Services: the Case of M-PESA in Kenya, Journal of Eastern African Studies,
(3): 509-525.
Nokia (2008). Nokia Introduces Affordable Mobile Devices and Services that Make the Internet
Available for Emerging Market. Nokia Press Release. November 04, 2008.
Nokia (2009). Affordability Key in Bringing Digital Inclusion. Expanding Horizons, 1st Quarter
2009, 9-12.
Norusis, M. J. (2003). SPSS 12.0 Statistical Procedures Companion. Upper Saddle River, NJ:
Prentice-Hall.
Nysveen, H. (2005). Intentions to Use Mobile Services: Antecedents and Cross-Service
Comparisons. Journal of the Academy of Marketing Science, 33(3), 330-346.
Okazaki, S. (2006). What Do We Know About Mobile Internet Adopters? A Cluster Analysis.
Information & Management, 43(2), 127-141.
Orilkowski, W. J. (1992). The Duality of Technology: Rethinking the Concept of Technology in
Organizations. Organization Science. 3(3). P.398-427.
Pajares, F. (2002) Overview of Social Cognitive Theory and of Self-efficacy, paper available at
http://www.emony.edu/education/mfp/eff.hml
Papacharissi, Z. & Rubin, A. (2000). Predictors of Internet Use. Journal of Broadcasting &
Electronic Media, 44(2), p.175-196.

114

Park, N., Roman, R., Lee, S., & Chung, J. E. (2009) User Acceptance of a Digital Library
System in Developing Countries: An Application of the Technology Acceptance Model.
International Journal of Information Management, 29(3), 196-209.
Pavlou, P. A., & Fygenson, M. (2006). Understanding and Predicting Electronic Commerce
Adoption: An Extension of the Theory of Planned Behavior. MIS Quarterly, 30(1), 115–
143.
Pedersen, P. E. (2005). Adoption of Mobile Internet Services: An Exploratory Study of Mobile
Commerce Early Adopters. Journal of organizational computing and electronic
commerce, 15(3), 203-222.
Pedersen, P. E. P., and Ling, R. (2003). Modifying Adoption Research for Mobile Internet
Service Adoption: Cross-disciplinary Interactions. Proceedings of the 36th Annual
Hawaii International Conference on System Sciences, 2003. 1–10.
Peters, O. (2009). A Social Cognitive Perspective on Mobile Communication Technology Use
and Adoption. Social Science Computer Review, 27(1), 76-95.
Plouffe, C. Hulland, J and Vandenbosch, M. (2001). Research Report: Richness Versus
Parsimony in Modeling Technology Adoption Decisions: Understanding Merchant
Adoption of a Smart Card-based Payment System. Information Systems Research. 12:
208-222.
Prahalad, C. K. (2004). The Fortune at the Bottom of the Pyramid: Eradicating Poverty through
Profits. Upper Saddle River, NJ: Wharton School Publishing.
Qiang, C. Z. W., & Rossotto, C. M. (2009). Economic Impacts of Broadband. 2009 Information
and communications for development: extending reach and increasing impact, 35. World
Bank Publications.
Qiu, J. L. (2008). Working-class ICTs, Migrants, and Empowerment in South China. Asian
Journal of Communication, 18(4), 333–347.
Rangaswamy, N. (2008). Telecenters and Internet Cafes : the Case of ICTs in Small Businesses.
Asian Journal of Communication, 18(4), 365-378.
Rashid, A. T., & Elder, L. (2009). Mobile Phones and Development: An Analysis of IDRCSupported Projects. Electronic Journal of Information Systems in Developing Countries,
36(2), 1-16
Ravallion, M., Chen, S., & Sangraula, P. (2009). Dollar a Day Revisited, World Bank Economic
Review, 23(2), 163-184

115

Rheingold, H. (2002). Smart Mobs: The Next Social Revolution, Perseus Books, Cambridge,
MA.
Rice, R., & Katz, J. (2003). Comparing Internet and Mobile Phone Usage: Digital Divides of
Usage, Adoption, and Dropouts. Telecommunications Policy, 27, 597–623.
Rogers, E. M. (1995). Diffusion of Innovations, New York: Free Press, 4th ed.
Ruggiero, T. (2000) Uses and Gratification Theory in the 21st Century, Mass Communication &
Society, 3(1), p.3-37.
Samarajiva, R. (2010). Leveraging the Budget Telecom Network Business Model to Bring
Broadband to the People 1. Information Technologies and International Development, 6
(Special Edition), 93–97.
Selwyn, N. (2003). Apart from Technology: Understanding People’s Non-use of Information and
Communication Technologies in Everyday Life. Technology in society, 25(1), 99-116.
Sey, A. (2008). Mobile Communication and Development: A Study of Mobile Phone
Appropriation in Ghana. Unpublished Doctoral Dissertation. University of Southern
California, USA.
Sey, A. (2009). Exploring Mobile Phone-Sharing Practices in Ghana. Info, 11(2), 66–78.
Sey, A., & Fellows, M. (2011). Loose Strands: Searching for Evidence of Public Access ICT
Impact on Development. Proceedings of iConference 2011, February 8-11, 2011, Seattle,
WA, USA
Sey, a. (2011). “We Use it Different, Different”: Making Sense of Trends in Mobile Phone Use
in Ghana. New Media & Society, 13(3), 375–390.
Silverstone, R. (1994). Television and Everyday Life. London: Rutledge.
Souter, D, Scott, N., Garforth, C., Jain, R., Mascararenhas, O. and McKemey, K. (2007) The
Economic Impact of Telecommunications on Rural Livelihoods and Poverty Reduction: A
Study or Rural Communities in India (Gujarat), Mozambique, and Tanzania. Report by
Commonwealth Telecommunications Organization for UK Department for International
Development, KaR Project 8347
http://www.telafrica.org/R8347/files/pdfs/FinalReport.pdf
Sridhar, K. and Sridhar, V. (2004). Telecommunications Infrastructure and Economic Growth:
Evidence from Developing Countries, National Institute of Public Finance and Policy.
Working Paper No. 14. New Delhi, India.
Straub, D., Keil, M., & Brenner, W. (1997). Testing the Technology Acceptance Model across
Cultures: A Three Country Study, Information & Management 33 (1), pp. 1–11.
116

Taylor, S., & Todd, P. A. (1995). Understanding Information Technology Usage: A Test of
Competing Models. Information systems research, 6(2), 144–176.
Teo, T. S., & Pok, S. H. (2003). Adoption of WAP-enabled Mobile Phones among Internet
Users. The International Journal of Management Science, 31(6), 483–498.
Thomas, J. J., & Parayil, G. (2008). Bridging the Social and Digital Divides in Andhra Pradesh
and Kerala: A Capabilities Approach. Development and Change, 39(3), 409-435.
Toyama, K. (2011). Technology as Amplifier in International Development. Proceedings of the
iConference 2011, February 8-11, 2011, Seattle, WA, US (pp. 75–82). ACM.
Trucano, M. (2009). What Do We Know about the Effective Uses of Information and
Communication Technologies in Education in Developing Countries? In Assessing the
effects of ICT in education: indicators, criteria and benchmarks for international
comparisons (p. 61). France: OECD Publishing.
Tukey, J. W. (1977). Exploratory Data Analysis. Reading: MA, 231.
UNDP (2012). Mobile Technologies and Empowerment: Enhancing Human Development.
United Nations Development Programme, New York, NY.
Unwin, T. (2009). ICT4D: Information and Communication Technology for Development.
Cambridge University Press, UK.
Van Dijk, J. (2005) The Deepening Divide: Inequality in the Information Society, Sage
Publications: USA
Venkatesh, V., & Davis, F. D. (2000). A Theoretical Extension of the Technology Acceptance
Model: Four Longitudinal Field Studies. Management science, 46(2), 186-204.
Venkatesh, V., & Morris, M. G. (2000). Why Don’t Men Ever Stop to Ask for Directions?
Gender, Social Influence, and Their Role in Technology Acceptance and Usage
Behavior. MIS Quarterley, 24(1), 115-139.
Venkatesh, V., Morris, M., & Davis, G. (2003). User Acceptance of Information Technology:
Toward a Unified View. MIS quarterly, 27(3), 425-478.
Venkatesh, V., Thong, J., & Xu, X. (2012). Consumer Acceptance and Use of Information
Technology: Extending the Unified Theory of Acceptance and Use of Technology. MIS
Quarterly, 36(1), 157–178.
Wallis, C. (2011). Mobile Phones without Guarantees: The Promises of Technology and the
Contingencies of Culture. New Media & Society, 13(3), 471–485.
Warschauer, M. (2003). Technology and Social Inclusion: Rethinking the Digital Divide. MIT
Press, USA.
117

Warschauer, M. (2004). The Rhetoric and Reality of Aid: Promoting Educational Technology in
Egypt. Globalisation, Societies and Education, 2(3), 377-390.
Waverman, L., M. Meschi and M. Fuss (2005). The Impact of Telecoms on Economic Growth in
Developing Countries, Vodafone Policy Paper Series, 2, March, London, UK.
Wei, L., & Hindman, D. B. (2011). Does the Digital Divide Matter More? Comparing the Effects
of New Media and Old Media Use on the Education-Based Knowledge Gap. Mass
Communication and Society, 14(2), 216-235.
Wesolowski, A., Eagle, N., Noor, A. M., Snow, R. W., & Buckee, C. O. (2012). Heterogeneous
Mobile Phone Ownership and Usage Patterns in Kenya. PloS one, 7(4), e35319.
Wirth, W., Von Pape, T., & Karnowski, V. (2008). An Integrative Model of Mobile Phone
Appropriation. Journal of Computer-Mediated Communication, 13(3), 593–617.
World Bank (2012). Information and Communications for Development 2012: Maximizing
Mobile. InfoDEV and World Bank publication: Washington D.C., USA
Wu, J., & Wang, S. (2005). What Drives Mobile Commerce? An Empirical Evaluation of the
Revised Technology Acceptance Model. Information & Management, 42(5), 719–729.
Yousafzai, S. Y., Foxall, G. R., & Pallister, J. G. (2010). Explaining Internet Banking Behavior:
Theory of Reasoned Action, Theory of Planned Behavior, or Technology Acceptance
Model? Journal of Applied Social Psychology, 40(5), 1172-1202.
Zainudeen, A., Samarajiva, R., & Abeysuriya, A. (2006). Telecom Use on a Shoestring: Strategic
Use of Telecom Services by the Financially Constrained in South Asia. World Dialogue
on Regulation for Network Economies (WDR) Dialogue Theme 3rd Cycle Discussion
Paper, WDR0604, Version 2. February 2006.
Zainudeen, A., Iqbal, T., & Samarajiva, R. (2010). Who’s Got the Phone? Gender and the Use of
the Telephone at the Bottom of the Pyramid. New Media & Society, 12(4), 549–566.
Zainudeen, A., & Ratnadiwakara, D. (2011). Are the Poor Stuck in Voice? Conditions for
Adoption of More-Than-Voice Mobile Services. Information Technologies International
Development, 7(3), 45–59.
Zhang, J., & Mao, E. (2008). Understanding the Acceptance of Mobile SMS Advertising among
Young Chinese Consumers. Psychology & Marketing, 25(August), 787-805.

118