3rd International Conference on

Information and Communication Technologies and Development

PROCEEDINGS
April 17-19, 2009
Carnegie Mellon University in Qatar
Education City, Doha, Qatar
Host Organization:

Organizing Partner:

Platinum Sponsors:

Gold Sponsors:

Bronze Sponsors:

Media Partner:

Technical Sponsors:
 Table of Contents

Introduction 1

A Global Empirical Evaluation of New Communication Technology 3

Use and Democratic Tendency
Victoria Stodden and Patrick Meier

A Review of the Research on Mobile Use by Micro and Small 17

Enterprises (MSEs)
Jonathan Donner and Marcela Escobari

An Evaluation of the Use of ICT within Primary Education in Malawi 27

David Hollow and Paola Masperi

Claim Mobile: Engaging Conflicting Stakeholder Requirements in 35

Healthcare in Uganda
Melissa R. Ho, Emmanuel K. Owusu and Paul Aoki

Computer Games in the Developing World: The Value of Non- 46

Instrumental Engagement with ICTs, or Taking Play Seriously
Beth E. Kolko and Cynthis Putnam

Content Creation and Dissemination by-and-for Users in Rural Areas 56

Sheetal K. Agarwal, Arun Kumar, Amit Anil Nanavati, Nitendra Rajput

E for Express: “Seeing” the Indian State through ICTD 66

Renee Kuriyan and Isha Ray

Evaluating the Accuracy of Data collection on Mobile Phones: A Study 74

of Forms, SMS, and Voice
Somani Patnaik, Emma Brunskill and William Thies

FOLKSOMAPS – Towards Community Driven Intelligent Maps for 85

Developing Regions
Arun Kumar, Dipanjan Chakraborty, Himanshu Chauhan

HIV Health Information Access using Spoken Dialogue Systems: 95

Touchtone vs. Speech
Aditi Sharma Grover, Madelaine Plauché, Etienne Barnard, Christiaan Kuun

ICT4What? – Using the Choice Framework to Operationalize the 108

Capability Approach to Development
Dorothea Kleine
ICTD for Healthcare in Ghana: Two Parallel Case Studies 118
Rowena Luk, Matei Zaharia, Melissa Ho, Brian Levine and Paul M. Aoki

Improving Child Literacy in Africa: Experiments with an Automated 129

Reading Tutor
G. Ayorkor Mills-Tettey, Jack Mostow, M. Bernadine Dias, Tracy Morrison
Sweet, Sarah M. Belousov, M. Frederick Dias, Haijun Gong

Improving Literacy in Rural India: Cellphone Games in an After-School 139

Program
Matthew Kim, Anuj Kumar, Shirley Jain, Akhil Mathur, and John Canny

Kelsa+: Digital Literacy for Low-Income Office Workers 150

Aishwarya Lakshmi Ratan, Sambit Satpathy, Lilian Zia, Kentaro Toyama,
Sean Blagsvedt, Udai Singh Pawar, Thanuja Subramaniam

Mapping the Dynamics of Social Enterprises and ICTD in Cambodia 163

Kelly Hutchinson and Alemayehu Molla

Political Incentives and Policy Outcomes: Who Benefits from Technology- 173
Enabled Service Centers?
Jennifer Bussell

Results from a Study of Impact of E-government Projects in India 183

Subhash C. Bhatnagar and Nupur Singh

The Contribution of User-Based Subsidies to the Impact and Sustainability 192

of Telecenters – the eCenter Project in Kyrgyzstan
Michael L. Best, Dhanaraj Thakur and Beth E. Kolko

Poster Papers 201

A Speech Enabled Indian Language Text-to-Braille Transliteration 201

System
Tirthankar Dasgupta and Anupam Basu

Analyzing Statistical Relationships between Global Indicators through 212

Visualization
Prabath Gunawardane, Erin Middleton, Suresh Lodha, Ben Crow and
James Davis

ATMosphere: A System for ATM Microdeposit Services in Rural Contexts 222

Michael Paik and Lakshminarayanan Subramanian
Building a Transportation Information System Using Only GPS and Basic 233
SMS Infrastructure
Ruth E. Anderson, Anthony Poon, Caitlin Lustig, Waylon Brunette, Gaetano
Borriello, Beth E. Kolko

Challenges in Health Information Systems Integration: Zanzibar 243

Experience
Edwin Nyella

Cross Technology Comparison for Information Services in Rural 252

Bangladesh
Faheem Hussain and Rahul Tongia

Decentralization, Clientelism and Popular Participation - Is There a 267

Role for ICTs to Improve Local Governance?
Björn-Sören Gigler

Design and Deployment of a Blood Safety Monitoring Tool 280

S. Thomas, A. Asuntogun, J. Pitman, B. Mulenga, and S. Vempala

Dimensions of IT Literacy in an Arab Region Study in Barkha(Oman) 288

Sherif M. Aziz

Emergency Communication and System Design: The Case of Indian 300

Ocean Tsunami
R. Chen, J. Coles, J. Lee, and F. R. Rao

Empowering Muslim Youth through Computer Education, Access, Use: 310

A Gender Analysis
Farida Khan and Rehanan Ghadially

eServices Provisioning in a Community Development Context through a 320

JADE MAS Platform
Mamello Thinyane, Alfredo Terzoli, Peter Clayton

Extending the Technology-Community-Management Model to Disaster 328

Recovery in Asia
Arul Chib and A.L.E. Komathi

Featherweight Multimedia for Information Dissemination 337

Gerry Chu, Sambit Satpathy, Kentaro Toyama, Rikin Gandhi, Ravin
Balakrishnan, S. Raghu Menon

ICT Governance in Higher Education: Case Study of the Rise and Fall 348
of Open Source in a gulf University
Sofiane M. Sahyraoui
ICTD State of the Union: Where Have We Reached and Where Are We 357
Headed?
Rabin Patra, Joyojeet Pal, Sergiu Nedevschi

Information Communication Techology and Sustainable Communities in 367

Africa: The Case of the Niger Delta Region of Nigeria (Feb. 2009)
Uduak A. Okon

Integrating Health Information Systems in Sierra Leone 379

Johan Sæbø, Edem Kwame Kossi, Romain Tohouri Golly-Kobrissa, Ola
Titlestad, Jørn Braa

Mobile Telephony Access and Usage in Africa 392

A. Chabossou, C. Stork, M. stork, Z. Zahonogo

Numeric Paper Forms for NGOs 406

Gusharan Singh, Leah Findlater, Kentaro Toyama, Scott Helmer, Rikin Gandhi,
Ravin Balakrishnan

Rajnikant’s Laptop: Computers and Development in Popular Indian 417

Cinema
Joyojeet Pal

Regulatory Independence and Wireless Market Development: A 427

Comparative Analysis of Two African Nations
Annemijn F. van Gorp and Carleen F. Maitland

Social Enterprises: A Vocational Entrepreneurship Framework for 437

Street Youth
Paul Javid, Kentaro Toyama, Manna Biswas

Speech vs. Touch-Tone: Telephony Interfaces for Information Access 447

by Low Literate Users
Jahanzeb Sherwani, Sooraj Palijo, Sarwat Mirza, Tanveer Ahmed, Nosheen
Ali, Roni Rosenfeld

The Case for SmartTrack 458

Michael Paik, Ashlesh Sharma, Arthur Meacham, Giulio Quarta, Philip
Trahanas, Brian Levine, Mary Ann Hopkins, Barbara Rapchak,
Lakshminarayanan Subramanian

Uses of Mobile Phones in Post-Conflict Liberia 468

Michael L. Best, Edem Wornyo, Thomas N. Smyth and John Etherton
Demos 478

An Automated Braille Writing Tutor with Multilingual Exercises 478

and Educational Games
M. Bernadine Dias, M. Freddie Dias, Sarah Belousov, Mohammed Kaleemur
Rahman, Saurabh Sanghvi, Imran Fanaswala, Wael Ghazzawi, Ameer
Abdulsalam, Noura El-Moughny, and S. Raghu Menon

Boosting European Market Access to Malian Mango Growers 479

Saskia Harmen

Creating a Mobile-Phone Based Geographic Surveillance System 480

for Asian Flu
Yibo Lin and Claire Heffernan

Design of a Blood Flow System 481

A. Osuntogun, S. Thomas, J. Pitman, S. Basavaraju, B. Mulenga and
S. Vempala

DISHA: DISease and Health Awareness for Children on Multiple 482

Input Devices
Mohit Jain, Aakar Gupta, Navkar Samdaria, Praveen Shekhar and Joyojeet
Pal

Freedom Fone: Dial-up Information Service 483

Bev Clark and Brenda Burrell

FrontlineSMS and Ushahidi – A Demo 484

Ken Banks and Erik Hersman

Global Youth Connectivity (GYC) – Using ICT for Peaceful Recovery 485
and Long-term Change
Anne Bertrand

Implementing E-Government Accessible to Illiterate Citizens 486

D. Kettani and A. El Mahdi

Improving Data Quality with Dynamic Forms 487

Kuang Chen, Harr Chen, Neil Conway, Heather Dolan, Joseph M.
Hellerstein, and Tapan S. Parikh

IWB4D – Interactive Whiteboards for Development 488

John Traxler and Lee Griffiths
Livestock, Learning and Diagnostics: New Directions in Veterinary 489
Tele-medicine
Jun Yu and Claire Heffernan

Metamouse: Multiple Mice for Legacy Applications 490

Kurtis Heimerl, Divya Ramachandran, Joyojeet Pal, Eric Brewer,
and Tapan Parikh

Mobile Phone Job Services: Linking Developing-country Youth with 491

Employers, via SMS
Amber Houssain, Mohammad Kilany, and Jacob Korenblum

MultiMath: Numeric Keypads for Math Learning on Shared Personal 492

Computers
Sunil Garg, Charlotte Robinson, Clint Tseng, Heather Underwood, Richard
Anderson, Joyojeet Pal

A New Generation of Open Source Data Collection Tools 493

Yaw Anokwa, Carl Hartung, Adam Lerer, Brian DeRenzi, Gaetano Borriello

RuralScope: An Information System for Tracking Rural Disbursements 494

Sai Gopal Thota, Rabin Ratra, Murali Medisetty, Sivananda Reddy, Vivek
Mungala, Joyojeet Pal

T-Cube Web Interface in support of Real-Time Bio-surveillance Program 495

Artur Bubrawski, Maheshkumar Sabhnani, Michael Knight, Michael Baysek,
Daniel Neill, Saswati Ray, Anna Michalska and Nuwan Waidyanatha

Web Search over Low Bandwidth 496

Jay Chen, Lakshminarayanan Subramanian, and Jinyang Li

Author Index 497

 ICTD 2009 – Organizing Committee
Honorary Chairs

Dr. Hessa Sultan Al-Jaber Dr. Kentaro Toyama

Secretary General Assistant Managing Director
ictQATAR Microsoft Research India (MSRI)

Conference Chair

Dr. M. Bernardine Dias

Carnegie Mellon University

Program Committee Chairs

Dr. Richard Heeks Dr. Rahul Tongia

University of Manchester Carnegie Mellon University

Advisory Board

Dr. Vallampadugai S. Arunachalam Dr. Francois Bar

Center for Study of Science, Technology, University of Southern California
and Policy (CSTEP), Bangalore

Dr. Michael Best Dr. Kenneth Keniston

Georgia Institute of Technology Massachusetts Institute of Technology

Dr. Balaji Parthasarathy Dr. Krithi Ramamritham

International Institute of Information Indian Institute of Technology (IIT),
Technology (IIIT), Bangalore Bombay

Dr. Raj Reddy Dr. AnnaLee Saxenian

Carnegie Mellon University University of California, Berkeley

Dr. Kentaro Toyama Dr. Ernest Wilson

Microsoft Research India University of Southern California

Senior Program Committee Members

Dr. Francois Bar Dr. Michael Best

University of Southern California Georgia Institute of Technology

Dr. Eric Brewer Dr. Chris Coward

University of California, Berkeley University of Washington
 Dr. Robert Davidson Dr. Hernan Galperin
City University of Hong Kong Universidad de San Andrés

Dr. Shirin Madon Dr. Alemayehu Molla

London School of Economics Royal Melbourne Institute of Technology

Dr. Balaji Parthasarathy Dr. Krithi Ramamritham

International Institute of Information Indian Institute of Technology (IIT)-Bombay
Technology (IIIT)-Bangalore

Dr. Kentaro Toyama Dr. Tim Unwin

Microsoft Research India Royal Holloway University of London

Panels and Workshops Chairs

Dr. Joseph Mertz Dr. Joyojeet Pal

Carnegie Mellon University University of Washington

Publications Chairs

Dr. Yonina Cooper Dr. Thrishantha Nanayakkara

Carnegie Mellon University in Qatar Harvard University and University of
Moratuwa, Sri Lanka

Poster Chairs

Dr. Faheem Hussain Dr. Tapan Parikh

Carnegie Mellon University in Qatar University of California, Berkeley

Demo Chairs

Dr. Khaled Harras Bill Thies

Carnegie Mellon University in Qatar Massachusetts Institute of Technology

Regional Chairs

Africa
Dr. Gary Marsden Dr. Tim Waema
University of Capetown, South Africa University of Nairobi, Kenya

Middle East
Dr. Adnan Abu Dayya Dr. Fouad Mrad
Qatar University American University of Beirut
 East Asia
Dr. Jack Linchuan Qiu Rinalia Abdul Rahim
The Chinese University of Hong Kong Global Knowledge Partnership, Malaysia

South Asia
Anita Gurumurthy Dr. Umar Saif
IT for Change, India Lahore University of Management
Sciences, Pakistan

South and Central America

Dr. Nicolau Reinhard Dr. Osvaldo Rodriguez
Universidade de São Paulo, Brazil La Planta University, Argentina

North America
Dr. John Bennett Claudia Morrell
University of Colorado at Boulder Multinational Development of Women
in Technology

Europe
John Traxler
University of Wolverhampton, UK

Oceania
Christina Higa Dr. Esther Batiri Williams
University of Hawaii The University of the South Pacific

Technical Program Committee Members

Dr. Jessica Aalami Dr. Salam Abdallah

University of California, Berkeley Abu Dhabi University

Dr. Reuben Abraham Dr. Erwin Alampay

Indian School of Business, Hyderabad University of Philippines

Dr. Richard Anderson Dr. Peng HwaAng

University of Washington Nanyang Technological University

Akhtar Badshah Dr. V. Balaji

Microsoft ICRISAT

Dr. Anupam Basu Dr. John Bennett

Indian Institute of Technology University of Colorado, Boulder
(IIT)-Kharagpur, India
Dr. Subhash Bhatnagar Robert Bichler
Indian Institute of Management, University of Salzburg
Ahmedabad

Dr. John Canny Dr. Royal Colle

University of California, Berkeley Cornell University

Dr. Rahul De Dr. Andy Dearden

Indian Institute of Management, Bangalore Sheffield Hallam University

Dr. Uday Desai Dr. Antonio Diaz

Indian Institute of Technology University of Auckland
(IIT)-Bombay

Dr. Jonathan Donner Dr. Ayman Elnaggar

Microsoft Research Sultan Qaboos University

Dr. Kevin Fall Dr. Ping Gao

Intel/Berkeley University of Manchester

Dr. Alison Gillwald Dr. Gillian Green

University of Witwatersrand University of Bolton

Pat Hall Dr. Saskia Harmsen

Kathmandu University International Institute for Communication
and Development (IICD)

Dr. Claire Heffernan Dr. Bill Hefley

University of Reading Carnegie Mellon University

Dr. Heather Hudson Dr. Faheem Hussain

University of San Francisco Carnegie Mellon University in Qatar

Mahad Ibrahim Dr. Ashok Jhunjhunwala

University of California, Berkeley Indian Institute of Technology
(IIT)-Madras

Dr. Muhammadou M. O. Kah Matt Kam

American University of Nigeria University of California, Berkeley

Dr. Sherif Kamel Dr. Atreyi Kankanhalli

American University of Cairo National University of Singapore

Dr. Srinivasan Keshav Dr. G. R. Kiran

University of Waterloo London School of Economics
Dr. Dorothea Kleine Dr. Jim Koch
University of London Santa Clara University

Dr. Beth Kolko Richa Kumar

University of Washington Massachusetts Institute of Technology

Dr. Gillian Marcelle Dr. Victor Mbarika

WITS University Southern University and A&M College

Dr. Wagner Meira Dr. Michel Menou

UFMG – Federal University City University of London
at Minas Gerais

Dr. Harekrishna Misra Dr. Amit Mitra

Institute of Rural Management, Anand Cranfield University

Dr. Beda Mutagahywa Dr. Shrikant Naidu

University of Dar es Salaam Motorola Labs, India

Dr. Amit Nanavati Dr. Solomon Negash

IBM Kennesaw State University

Dr. Joyojeet Pal Dr. Tapan Parikh

University of Washington University of California, Berkeley

Dr. Fay Payton Francisco Proenza

North Carolina State University Food and Agriculture Organization
of the United Nations (FAO)

Dr. Ranjini Raghavendra Dr. Nicolau Reinhard

Indian Institute of Management, Bangalore University of Sao Paolo

Marijn Rijken Dr. Osvaldo Rodriguez

TNO (Netherlands Organisation for La Planta University
Applied Scientific Research)

Dr. Roni Rosenfeld Dr. Sofiane Sahraoui

Carnegie Mellon University American University of Sharjah

Dr. Tony Salvador Dr. Maung Sein

Intel University of Agder

Dr. Afzal Sher Jahanzeb Sherwani

Swedish Program on ICT in Developing Carnegie Mellon University
Regions (SPIDER)
 Dr. Nirvikar Singh Dr. Hettie Soriyan
University of California, Santa Cruz Obafemi Awolowo University

Dr. Christoph Stork Dr. Eswaran Subrahmanian

Wits University CMU/NIST/CSTEP

Dr. Lakshmi(narayanan) Subramanian Bill Thies

Courant Institute of Mathematical Sciences Massachusetts Institute of Technology

Dr. Cathy Urquhart Dr. Carinade Villiers

University of Auckland University of Pretoria

Dr. Timothy Waema Dr. Ernest Wilson

University of Nairobi University of Southern California

Dr. Adel El Zaim

International Development Research
Centre (IDRC) Canada

Local Organizing Committee Chairs

Murry Evans Elaine Farah

Carnegie Mellon University in Qatar ictQatar
Erin Stewart
Carnegie Mellon University in Qatar

Publicity Coordinators

Noha Al Afifi Andrea Zrimsek

Carnegie Mellon University in Qatar Carnegie Mellon University in Qatar

Logistics Coordinators
Renee Barcelona Sarah Belousov
Carnegie Mellon University in Qatar Carnegie Mellon University
Kara Nesimiuk Ermine Teves
Carnegie Mellon University in Qatar Carnegie Mellon University

Website Coordinators

M. Freddie Dias Daniel Freeman

Carnegie Mellon University Carnegie Mellon University

Local Organizing Committee Members

Ray Corcoran, Bob Gaus, Jim Gartner, Shamila Khader, Aaron Lyvers
Carnegie Mellon University in Qatar
 1

Introduction to the Proceedings of ICTD2009

Welcome to the 3rd International Conference on Information and Communication

Technologies and Development (ICTD2009). It is with great pleasure that we present
the ICTD2009 proceedings, which include all of the full papers presented at the
conference in Doha, Qatar, held on 17-19 April 2009.

ICTD is the premier series of scholarly conferences on the use of ICTs for
development, spanning technical and social science domains. The call for papers
attracted a record 250 submissions. All papers were put through a double-blind peer-
review process. The Program Co-Chairs assigned papers to our Senior Program
Committee members who oversaw a review process involving three Program
Committee reviewers per paper: one with deep expertise about the subject matter;
another with broad background in the area; and one drawn from an altogether
different discipline. Our continuing hope is that this encourages a convergence of
vocabulary and ideas within the ICTD field, while maintaining the integrity of
different disciplines. The Senior Program Committee members then meta-reviewed
the papers. Authors were allowed a brief rebuttal to reviewer comments before final
acceptance decisions were made and revisions were finalized.

Ultimately, 19 papers were selected for oral presentation, and another 27 papers were
chosen as full papers for poster presentation; an acceptance rate of just over 18%.
These papers represent some of the best work being done in ICTD today. They focus
on a wide variety of development goals, and involve a broad and innovative range of
digital technologies. They draw from all continents of the global South, and focus on
all stages of the ICTD lifecycle: from readiness through design and adoption to use
and impact. They also tell us about all levels, from the individual through
communities and projects to ICTD programmes and policies. We hope that you will
find them an insightful, provocative, and informative contribution to our fast-growing
field of research and practice.

We wish to thank our keynote speakers William H. Gates, Chairman of Microsoft

Corporation and Co-Chairman of the Bill and Melinda Gates Foundation, and Dr.
Carlos A. Primo Braga, Director of Economic Policy and Debt in the Poverty
Reduction and Economic Management Network (PREM) at The World Bank for their
insightful presentations. All the other conference presenters also deserve our gratitude
for the variety of content and insights they added to the conference program.

We also need to thank a number of people without whom the program could not have
been put together. First, we wish to thank our Honorary Chairs, Dr. Hessa Sultan Al-
Jaber, Secretary General of The Supreme Council of Information and Communication
Technology (ictQatar), and Dr. Kentaro Toyama, Assistant Managing Director of
Microsoft Research India (MSRI) for their tremendous support in making this
conference a success. We are of course deeply indebted to our Program Committee:
those who did the hard work of reviewing and our senior PC members who managed
the review process so effectively; all together it is they who ultimately steer the course
of ICTD. We thank the Regional Chairs for their assistance in promoting the
conference, and our Advisory Board which provided guidance and moral support. We
also thank the Publication Chairs, Yonina Cooper and Thrishantha Nanayakkara, who
 2

made this proceedings possible; and Faheem Hussain and Tapan Parikh, who carried
the load in organisation of the poster presentations.

The conference program for ICTD2009 went well beyond papers, and was
significantly enhanced by demonstrations, organised by Bill Thies and Khaled Harras,
and by a series of panels and workshops, organised by Joe Mertz and Joyojeet Pal.
We are very grateful for their input and hard work.

This conference would not have been possible without the tireless efforts of a number
of organizers and volunteers, notably our local organizing committee chairs Elaine
Farah, Murry Evans, and Erin Stewart, and the many others on the local organizing
committee, especially Dean Charles Thorpe, Sarah Belousov, Ermine Teves, Renee
Barcelona, Kara Nesimiuk, Andy Zrimsek, Noha Al Afifi, Shams Hassan, Aaron
Lyvers, Shamila Khader, Ray Corcoran, Bob Gaus, Freddie Dias, Daniel Freeman,
and Jim Gartner. If there are others we did not name explicitly, it is our lapse.

The success of the conference is in part due to our many sponsors and partners. We
are extremely grateful to our organizing partner, The Supreme Council of Information
and Communication Technology (ictQatar), to our media partner Al Jazeera
Children’s Channel (JCC), to our technical sponsors IEEE and ACM, and to our
financial sponsors, the Qatar National Research Fund (Platinum Sponsor); Canada’s
International Development Research Centre (Platinum Sponsor); Qatar Telecom
(Platinum Sponsor); ExxonMobil (Gold Sponsor); Microsoft Corporation (Gold
Sponsor); IBM (Bronze Sponsor); the Computer Science program at Carnegie Mellon
University in Qatar (Bronze Sponsor); and other sponsors who asked not to be
publicly acknowledged.

Finally, we are indebted to Carnegie Mellon University in Qatar for hosting the
conference at their campus in Education City, and to the numerous employees of
Carnegie Mellon University (in both the Pittsburgh and Doha campuses), especially
the TechBridgeWorld team, who truly made this conference a success.

Thanks to everyone who contributed to ICTD2009 in so many ways, and thank you
for participating! We are confident you will find the following papers, abstracts, and
information not only interesting and useful, but the seeds for further research,
innovation, and developmental impact.

Richard Heeks, University of Manchester M. Bernardine Dias, Carnegie Mellon

Rahul Tongia, Carnegie Mellon University
University Conference Chair
Program Chairs

A Global Empirical Evaluation of New
Communication Technology Use and Democratic
Tendency
Victoria Stodden
Berkman Center for Internet and Society
Harvard Law School
Cambridge, MA 02138
vcs@stanford.edu
Abstract—Is the dramatic increase in Internet use associated
with a commensurate rise in democracy? Few previous studies
have drawn on multiple perception-based measures of governance
to assess the Internet’s effects on the process of democratization.
This paper uses perception-based time series data on “Voice &
Accountability,” “Political Stability,” and “Rule of Law” to pro-
vide insights into democratic tendency. The results of regression
analysis suggest that the level of “Voice & Accountability” in a
country increases with Internet use, while the level of “Political
Stability” decreases with increasing Internet use. Additionally,
Internet use was found to increase significantly for countries
with increasing levels of “Voice & Accountability.” In contrast,
“Rule of Law” was not significantly affected by a country’s level
of Internet use. Increasing cell phone use did not seem to affect
either “Voice & Accountability,” “Political Stability” or “Rule
of Law.” In turn, cell phone use was not affected by any of
these three measures of democratic tendency. When limiting our
analysis to autocratic regimes, we noted a significant negative
effect of Internet and cell phone use on “Political Stability” and
found that the “Rule of Law” and “Political Stability” metrics
drove ICT adoption.
Index terms—cell phone, democracy, fixed effects model, ICT,
internet
I. I NTRODUCTION
Does the globalization of the Internet have a democratizing
effect? The question has already been posed by numerous
studies but these have largely taken the form of qualitative case
studies and/or large theoretical analyses. In terms of a rigorous,
quantitative establishment of the democratization effects of the
Internet, however, the jury is still out [1]. At the heart of
this debate, moreover, lies a more fundamental question about
the essence of democracy. In fact, “unless we are clear about
what democracy means to us, and what kind of democracy we
envision, technology is as likely to stunt as to enhance the civic
polity” [2]. The purpose of this paper is to contribute more
rigorous data-driven analysis to the literature on Internet and
democracy since “there is no doubt that rigorous and data-
driven analysis of this relationship will benefit scholars and
policymakers alike” [1].
Previous research on the topic of Internet and democracy
can be characterized as lacking a serious perusal of the
3
Patrick Meier
Berkman Center for Internet and Society
Harvard Law School
Cambridge, MA 02138
patrick.meier@tufts.edu
democracy and regime transitions literature. To be sure, “the
trouble with the zealots of technology as an instrument of
democratic liberation is not that they misconceive technology
but that they fail to understand democracy” [2]. ‘In other
words, “it turns out there is no simple general answer to
the question: Is the technology democratizing?’ until we have
made clear what sort of democracy we intend.” We address this
question first before proceeding with a more detailed literature
review.
Barber’s notion of “strong democracy” comprises the careful
and prudent judgment of citizens who participate in deliber-
ative, self-governing communities. Schmitter and Karl write
that, “modern political democracy is a system of governance
in which rulers are held accountable for their actions in the
public realm by citizens, acting indirectly through the compe-
tition and cooperation of their elected representatives”[3]. The
two authors emphasize that citizens are the most distinctive
element in democracies. “All regimes have rulers and a public
realm, but only to the extent that they are democratic do
they have citizens” [3]. In contemporary studies of democracy
and particularly in pluralist theory, “a vibrant civil society
is usually regarded as an essential for good governance
and effective democratic consolidation” [4]. In other words,
regular elections are not sufficient. As Zakaria noted, illiberal
democracies have free elections but citizens remain cut off
from real power due to the lack of civil liberties [5].
Huber et al. write that the most basic feature of democracy
is power sharing [6]. They identify three clusters of power
as primarily relevant for the chances of democracy: (1) the
balance of power in civil society; (2) the balance of power
between state and society; and (3) the transnational balance of
power that shape the first two and constrain political decision-
making. By remaining diverse and independent of the state,
political participation by civil society acts as a channel of
public voice and accountability, and a way of challenging
and checking the unbridled power of authoritarian regimes
[4]. The structure of state-society relations is equally relevant
for democracy. As Huber et al. note, “the power of the state
needs to be counterbalanced by the organizational strength of

the civil society to make democracy possible; the state must
not be so strong and autonomous from all social forces as
to overpower civil society and rule without accountability.”
Clearly then a governing body that fails to follow the “rule
of law,” should not be considered democratic [3]. These
elements of democracy are not sufficient conditions for a stable
democracy, but they are necessary and indispensable to the
persistence of democratic governance.
The italicized terms above represent the fundamentals be-
hind the sort of democracy we intend: active citizen par-
ticipation, good governance, accountability, power sharing,
balance of power and rule of law. The few quantitative studies
that do exist on Internet and democracy tend to aggregate
these fundamentals of democracy into a single index. Doing
so means these lose important information on how these
individual components of democracy may be affected by the
growing prevalence of global Internet access. Furthermore,
past quantitative and qualitative studies tend to focus primarily
on the impact of the Internet on established democracies. They
also focus on the 1990s almost exclusively, a serious limitation
that remains surprisingly understated in the literature. Equally
problematic in the current literature is the interchangeable
use of the terms “Internet” and “information revolution.” The
terms are purposefully not differentiated on the basis that the
predominant feature of the information society is the spread
of the Internet. While this is true of Western democracies, it
is certainly not true for the majority of developing, nondemo-
cratic countries, where cell phones are the most widely spread
communication technology after radios [7]. Indeed, the irony is
that “those who might most benefit from the net’s democratic
and informational potential are least likely to have access to
it, the tools to gain access, or the educational background to
take advantage of the tools” [2].
This paper seeks to redress each of these shortcomings.
First, since the boundaries of the term “democracy,” and how
it is measured, is subject to lively debate, we use multiple
perception-based measures of governance for our dependent
variables. Governance indicators provide a better set of proxies
for the sort of democracy we intend as identified above. We
therefore draw on the following three World Bank indicators:
(1) Voice and Accountability (VA) measuring perceptions
of the extent to which a country’s citizens are able to par-
ticipate in selecting their government, as well as freedom
of expression, freedom of association, and a free media; (2)
Rule of Law (RL) measuring perceptions of the extent to
which agents have confidence in and abide by the rules of
society, and in particular the quality of contract enforcement,
property rights, the police, and the courts, as well as the
likelihood of crime and violence; and (3). Political Stability
and Absence of Violence (PS) measuring perceptions of
the likelihood that the government will be destabilized or
overthrown by unconstitutional or violent means, including
politically motivated violence and terrorism. These metrics are
drawn from the World Bank Governance Indicators Research
Database (see http://info.worldbank.org/governance/wgi/).
Second, we draw on data from 2000 through 2006, a time
4
when Internet access and cell phone use is significantly more
prevalent and globalized than in the 1990s. If a statistically
significant relationship between Internet and democracy does
exist, then it is more likely to manifest itself now and not in the
1990s. Third, we draw on both Internet and cell phone data per
100 inhabitants per 181 countries to assess the impact of the
information revolution on democratization. We use regression
analysis to determine whether Internet or cell phone use has
had a correlative effect on measures of democratic tendency.
We also model whether the collection of democratic measures
has had a correlative effect on Internet or cell phone use.
The paper is structured as follows: the first section reviews
the current debate and literature on Internet and democracy.
The second section explicates the datasets used in this study
while the third section formalizes the statistical models em-
ployed in the regression analysis. Section four reviews the
results and provides an interpretation of the findings. The fifth
and final section concludes the study.
II. L ITERATURE R EVIEW
The Internet and democracy literature comprises two distinct
schools of thought each comprising a host of qualitative
research and some quantitative inquiry. In this section we
review in some detail the qualitative and quantitative literatures
that have contributed to both schools of thought over the past
ten to fifteen years. In so doing, we compare and contrast the
main arguments along with the respective findings. As noted
in the introduction, one common shortcoming of the Internet
and democracy literature is the tendency to oversimplify our
understanding of democracy. The purpose of this literature
review is thus to redress this gap evident in previous studies.
The first school of thought is often referred to as the
more populist school of thought. This strand of the literature
subscribes to the viewpoint that the Internet has democratic
relevance and impact [8], [9], [10], [11]. According to these
authors, the Internet will decentralize access to communication
and information while increasing citizen access [12]. Best
and Wade write that “the Internet’s collective characteristics
(e.g., low cost, multidirectional capability, etc.), helps make
this possible.” We first review the qualitative literature that
comprises this school of thought followed by quantitative
studies.
Dahl previously observed that telecommunications tech-
nologies have a key role in making possible the advanced
democratic country, where policy is firmly anchored in the
judgment of the “demos” [13]. In his list of the procedural
minimal conditions that must be present for modern political
democracy to exist, Dahl thus argues that citizens should
have the right to seek out alternative sources of information.
Rheingold has dubbed the Internet as “the great equalizer”
because it can “equalize the balance of power between citizens
and power barons” [14]. The idea here “is that the Internet
will serve as a mass audience, and will politicize them in the
process” [12]. Anderson et al. draw on sociological research
to show that electronic networks lead to the “break-down of
status-based social structures” and “increased participation in

discussion, decision-making, and task processes by those who
typically are political or economically disadvantaged” [15].
Etzioni sees in the Internet the possibility of an advancement
of the state of public affairs through “teledemocracy” [16].
Other scholars claim that the Internet will “enable a Jeffer-
sonian revolution” [17]. Grossman argues that a “third epoch
of democracy is arriving by the hand of technology,” and
claims that a “new, hybrid electronic republic’ is now forming
to displace the creation of Montesquieu, Locke, Madison and
their contemporaries” [18], [12]. The practice of “electronic
democracy,” according to Browning, will differ substantially
from previous renderings of democracy [19]. Perhaps the most
provocative claim associated with this more populist school of
thought is the one made by Negroponte, who argues that the
nation-state will evaporate as a consequence of the information
revolution [20]. Snider suggests that citizens need only be
potentially informed in order to hold government accountable
[21].
Barber opines that by favoring decentralization, the mul-
tiplication of choice, and consumer sovereignty, new tech-
nologies such as the internet or cell phone have already,
albeit inadvertently, benefited democratic political culture [2].
According to Barber, “democracy is a form of government
that depends on information and communication. It is obvious
then that new technologies of information and communications
can be nurturing to democracy. They can challenge passivity,
they can enhance information equality, they can overcome
sectarianism and prejudice, and they can facilitate participation
in deliberative political processes.” Hill and Hughes argue that
those who subscribe to the populist school have reasons to be
optimistic: “If the mere fact that political discourse against
repressive governments is taking place is a good in itself, then
the utopians have reason to celebrate [22]. Perhaps the Internet
will bring about a wider democratic revolution in the world”
[22].
Bimber is more cautious, arguing that while the Internet is
accelerating the process of issue group formation and action
(in America), the structure of political power has not been
revolutionized or qualitatively transformed into a new epoch of
democracy [12]. According to Bimber, other scholars believe
that the Internet has a “transformative potential” because it
facilitates a kind of “one-to-one interaction among citizens
and between citizens and government.” Along these lines,
some scholars such as Corrado and Firestone write that the
Internet has the potential to promote “unmediated” commu-
nication and thereby decrease citizens’ reliance on officials
and organizations [23]. In sum, what distinguishes the populist
enthusiasm for the Internet is the “idea that elites and political
intermediaries will grow less important” [12].
In the more contemporary, qualitative literature, Steele and
Stein, argue that the Internet amplifies trends in international
relations [24]. Rosenau and Johnson address the impact of the
Internet at both the individual and international level [25]. At
the individual level, the authors argue that the Internet can be
used as a tool to organize collectively to effect social and polit-
ical change around the world (see http://www.DigiActive.org,
5
for example). At the international level, the authors make the
bold claim that the Internet has “contributed to the rise of a
more multicentric world structure in which nation-states have
seen their preeminence lessen and non-governmental actors
take the stage” [1]. The salient point here is that groups and
individuals can far more efficiently form coalitions of conse-
quence with a range of powerful collectives. As Best and Wade
rightly note, there are obvious democratic elements to this,
including the need for “nation-states to provide democratic
rights to their citizens so as to build legitimacy on the global
stage” [1].
We now turn to the quantitative studies that comprise the
first school of thought. One of the earlier statistical studies
on this side of the literature was carried out by Kedzie,
who provides an account of how information communication
technologies contributed to the “third wave” of democracy
[26]. Prior to the fall of the Soviet Union and the prolifer-
ation of new democracies in Eurasia, the mainstream theory
of democratization held that democracy followed economic
growth and development [27]. To be sure, one of the few robust
findings in the literature is that democracy is more likely in
more developed countries [28], [29], [30]. Longdregan and
Poole have also shown that the most significant predictor of
transitions to authoritarianism is poverty [31], [32]. “In short,
after 20 years of observation and analysis during the third wave
of academic interest in democratization, we can be reasonably
certain that a positive relationship between development and
democracy exists, though we do not know why” [33].
Kedzie, however, was more interested in testing another
potential causal mechanism, the “dictator’s dilemma” hypoth-
esis, which suggests that the globalization of markets places
pressure on authoritarian regimes to keep their countries’
communication borders open. He reasoned that the ensuing
massive flow of information would not only allow for “the
efficient passage of commercial information, but also for more
’democratic’ information” [1]. As Bimber observes, the most
important predictions about the Internet’s impact on politics
amount to “causal claims regarding the effect of information
flow on political participation and the organization of interests”
[12]. Other scholars have made related arguments. Webster,
for example, writes that the Internet has helped to facilitate
a new form of capitalism called “information capitalism” in
which global labor markets require highly flexible workers
who continuously adapt and learn [34]. Regimes that impose
restrictions on information capitalism forgo the financial re-
turns possible by tapping into the information economy [35],
[36], [37].
In his study, Kedzie employs regression analysis to compare
how much of the variation in democracy is explained by both
traditional predictors of democracy and the strength of Internet
diffusion by drawing on data from 144 countries [26]. For
his set of control variables, Kedzie included more traditional
predictors of democracy including economic development,
education, human development and health. He also included
indicators of pre-Internet information communication tech-
nologies (ICTs). His results suggest that the Internet is indeed

a strong predictor of democracy, more so than traditional
determinants of democracy. In a follow up study, Richards
assessed the relationship between the Internet and physical
integrity [38]. His findings support Kedzie’s. However, the
latter study faces an important limitation since Kedzie’s (rather
simple) longitudinal analysis draws on data from 1993. At
this point during the early 1990s, the Internet was hardly
globalized.
Best and Wade recognize this important short coming in
Kedzie’s study and therefore explore the global effect of the
Internet on democracy over a ten year period, 1992−2002 [1].
They aggregate political and civil rights data from Freedom
House to formulate a democracy index, which serves as their
dependent variable. The number of Internet users per 1, 000
represents their independent variable while the following mea-
sures are used as control variables: GDP per capita, education
and literacy rates, life expectancy, urbanization, prevalence of
non-Internet ICTs. Their analysis shows that a statistically
powerful correlation exists between Internet diffusion and
level of democratization. “The more salient observation to
make, however, is that while economic prevalence and liter-
acy maintain relatively constant correlations with democracy,
the correlation for Internet prevalence gradually strengthens,
almost to the same level as economic prevalence” [1].
The authors suggest that this dynamic reflects the grow-
ing significant relationship between Internet prevalence and
democracy: “perhaps this is an indication that the Internet
has come of age as a correlate of democracy” [1]. Indeed,
they posit that this growth in correlation strength might “be
expected given the positive network externalities, the network
effect’ that is a salient property of the Internet” [1]. However,
the coefficients from the regression analysis reveal that Internet
usage is only able to predict a minimal amount of the variation
in democracy: “to generate one point of democracy, an extra
500 Internet users per 1,000 citizens is needed, or an extra
$5, 882 of GDP per capita is needed” [1]. The scale of democ-
racy runs from 2 to 14. In terms of democracy’s traditional
determinants, GDP was a weak predictor while literacy turned
out to have no significance whatsoever. The other control
variables used were either insignificant or internally correlated.
While Best and Wade’s important contribution to the litera-
ture on Internet and democracy is one of the few contemporary
quantitative studies carried out thus far, their approach does
face a number of important limitations [1]. For one, their
democracy index needs to be unpacked and “its constituent
components, such as freedom of the press, openness of the
electoral process,” for example, tested against traditional de-
terminants of democracy to determine whether one compo-
nent provides more explanatory power than others. Another
limitation of their data is the fact that it extends only to
2002. This should be updated today due to the rapid pace
of ICT diffusion over the past several years. In addition,
several scholars have criticized the Freedom House data with
regards to conceptualization, measurement and aggregation
issues (Munck and Verkuilen, 2002; Rydland et al., 2008).
Furthermore, as discussed subsequently, there is little to no
6
variation in the Freedom House data, which makes meaningful
statistical analysis more difficult.
In contrast to the populist literature, the second school
of thought disputes the majority of claims that exist vis-à-vis
the relationship between Internet and democracy. The counter-
arguments are based on both qualitative and quantitative re-
search. In terms of qualitative research, several scholars argue
that the Internet is merely an extension of the ruling class and
centralized control [39], [40], [41], [34], [42]. According to
Neuman, even if the increase of ICTs had led to an increase
in the motivation to communicate - which he argues has not
happened - then ICTs would have become centralized by
government turning them into social control mechanisms [43].
Scholars who subscribe to this school of thought maintain that
mass media information technologies discourage collective
behavior, ”unless the rise in couch potatoes can be considered
a social movement” [44], [45], [46].
In contrast to Snider’s argument about the mere potential of
citizens being informed acting as a source of accountability, if
power is measured by the potential for “monopoly and control
over information and communication, it is evident that the new
technology can become a dangerous facilitator of tyranny”
[21], [2]. Indeed, while the Internet may enable citizens to
subvert political hierarchy, Barber notes that with increased
participation comes the peril of political and economic surveil-
lance. The populist school of thought is often blind to “how
easily liberating technologies become tools of repression” [2].
Bimber rejects the supposition that the Internet will
have significant effects on public life, point out that “both
theory and empirical evidence cast grave doubt on the
communication-action connection at the core of the populist
theory” [12]. Lippmann argued that the capacity of ICTs to
recreate politics is constrained by human nature, ie. cognitive
processing, and not by the technical properties of the media
themselves [47]. The Internet, then, is no different than other
ICTs even if the new medium differs from previous technolo-
gies in a fundamental way, namely allowing social bonding
to occur asynchronously. In sum, the Internet is “hardly
producing the first dramatic expansion in communication:
telephone, radio, and television also expanded communication
profoundly.” There seems no compelling reason to believe
that the communication capacity of the Net will have such a
dramatically different effect than have other advances in point-
to-point and broadcast communication” [12]. Moreover, Page
argues, new ICTs may very well overcome spatial distance
but his far from sufficient for establishing vibrant forms of
political communication and deliberation [48].
Furthermore, “if democracy is to be understood as delib-
erative and participatory activity on the part of responsible
citizens, it will have to resist the innovative forms of dem-
agoguery that accompany innovative technology and that are
too often overlooked by enthusiasts [2]. Aristotle wrote that
the basis of a democratic state is liberty. Barber adds that a
“free society is free only to the degree that its citizens are
informed and that communication among them is open and
informed [2]. However, recent research and empirical work

confirms that governments increasingly have the upper hand
in controlling and regulating the impact of the information
revolution [49], [50], [51], [52], [53], [54], [55], [56], [57],
[58]. As Goldsmith observes, “if governments can raise the
cost of Net transactions, they can regulate the transactions”
[59]. Beilock and Dimitrova found that countries with lower
Freedom House scores for civil liberties had significantly
lower Internet usage (even when controlling for economic
development) [60].
De Mesquita and George Downs also argue that government
elites (e.g., in Singapore) have learned to “stifle the bottom-up
democratic potential of the Internet and still promote economic
growth, contrary to Kedzie’s dictator’s dilemma argument”
[61]. As Bimber notes, the “central theoretical problem for
the populist claim is the absence of a clear link between
increases in information and increases in popular political
action” [12] To this end, McLuhan’s old dictum may be wrong:
“the medium is not the whole message. Content matters, and
there is simply no overwhelming reason to believe that a
new medium will necessarily enhance the political quality
of communicative content” [12]. In short, “technology need
not inevitably corrupt democracy, but its potential for benign
dominion cannot be ignored” [2].
In terms of quantitative analysis, Scheufele and Nisbet’s
2002 statistical study suggests that the Internet does not
increase democracy. “Through linear regression, they find
that mass media broadcasting (e.g., television, newspapers)
plays a far more effective role than the Internet in promoting
democratic citizenship” [1], [62]. Given that an established
body of quantitative research on this topic has yet to material-
ize, Scheufele and Nisbet do caution against generalizing the
results of their study, which focused exclusively on the United
States. Other scholars interested in this line of research have
questioned the supposed direction of causation drummed up
by the populist school of thought. Using multiple measures
of regime type, Milner’s statistical analysis demonstrates that,
ceteris paribus, democracies permit much greater online ac-
cess, both in terms of Internet users per capita and Internet
hosts per capita [63]. To this end, the information revolution
may merely be reinforcing pre-existing dynamics.
Milner’s study uses data from 1991 − 2001 to measure the
influence of regime type of adoption of the internet. This study
attempts to address a slightly different question—whether
there is a relationship between measures of democracy and
ICT penetration—but we build on her work by extending the
range of years to 2006 (although we begin measurement in
2000). We adopt a fixed effect model and control for the time
component directly in the model.
Beilock and Dimitrova develop a model to explain global
country differences in Internet use using income, measures of
freedom, region dummies, and development indices [60]. Their
data is a cross section from 2001 and does not take democratic
variables directly into account.
Best and Wade ask the question closest to that addressed
in this paper [1]. They ask whether Internet penetration has
an effect on the level of democracy in a country. Their study
7
is global in scope and uses time series data from 1992 to
2002. We seek to build on their work by using data from
2000 to 2006 and using the World Bank Governance metrics
as our measures of democratic tendency. Best and Wade
combined the Freedom House metrics of political rights and
civil liberties as their measure of democratic tendency. As
described subsequently we feel the Freedom House data are
not well suited to a regression study such as this one.
A. Our Approach to Measuring Democratic Tendency Using
the World Bank Governance Indicators
Dahl characterizes a government with power vested in a
plurality as follows [13]:
1) Control over governmental decisions about policy is
constitutionally vested in elected officials.
2) Elected officials are chosen and peacefully removed
in relatively frequent, fair and free elections in which
coercion is quite limited.
3) Practically all adults have the right to vote in these
elections.
4) Most adults also have the right to run for the public
offices for which candidates run in these elections.
5) Citizens have an effectively enforced right to freedom
of expression, particularly political expression, including
criticism of the officials, the conduct of the government,
the prevailing political, economic, and social system, and
the dominant ideology.
6) They also have access to alternative sources of informa-
tion that are not monopolized by the government or any
other single group.
7) Finally, they have an effectively enforced right to form
and join autonomous associations, including political
associations, such as political parties and interest groups,
that attempt to influence the government by competing
in elections and by other peaceful means.
The first four points largely describe procedural aspects of
a democracy, whereas the last three points delineate the com-
munication aspect necessary for a well-functioning democratic
regime. In fact, Diamond goes further and notes that “[s]ome
insist on a fairly robust (though still procedural) definition
of democracy, like Dahl’s ‘polyarchy.’ By this conception,
democracy requires not only free, fair, and competitive elec-
tions, but also the freedoms that make them truly meaningful
(such as freedom of organization and freedom of expression),
alternative sources of information, and institutions to ensure
that government policies depend on the votes and preferences
of citizens” [64]. Expanding on Dahl, Diamond has developed
his own list of characteristics of a democracy [65]:
1) Control of the state and its key decisions and allocations
lies, in fact as well as in constitutional theory, with
elected officials (and not democratically unaccountable
actors or foreign powers); in particular, the military is
subordinate to the authority of elected civilian officials.
2) Executive power is constrained, constitutionally and in
fact, by the autonomous power of other government in-

stitutions (such as an independent judiciary, parliament,
and other mechanisms of horizontal accountability).
3) Not only are electoral outcomes uncertain, with a sig-
nificant opposition vote and the presumption of party
alternation in government, but no group that adheres to
constitutional principles is denied the right to form a
party and contest elections (even if electoral thresholds
and other rules exclude small parties from winning
representation in parliament).
4) Cultural, ethnic, religious, and other minority groups
(as well as historically disadvantaged majorities) are not
prohibited (legally or in practice) from expressing their
interests in the political process or from speaking their
language or practicing their culture.
5) Beyond parties and elections, citizens have multiple,
ongoing channels for expression and representation of
their interests and values, including diverse, independent
associations and movements, which they have the free-
dom to form and join.
6) There are alternative sources of information (including
independent media) to which citizens have (politically)
unfettered access.
7) Individuals also have substantial freedom of belief, opin-
ion, discussion, speech, publication, assembly, demon-
stration, and petition.
8) Citizens are politically equal under the law (even though
they are invariably unequal in their political resources).
9) Individual and group liberties are effectively protected
by an independent, nondiscriminatory judiciary, whose
decisions are enforced and respected by other centers of
power.
10) The rule of law protects citizens from unjustified deten-
tion, exile, terror, torture, and undue interference in their
personal lives not only by the state but also by organized
nonstate or anti-state forces.
Like Dahl, Diamond includes procedural aspects of a demo-
cratic regime (points one through three) and he enshrines what
he considers essential communication requirements in points
four through seven. These latter points can be characterized as
approximating a “freedom of expression” or “political voice”
aspect to democracy. We found the World Bank governance
metric of “Political Voice and Accountability” to represent
Diamond’s notion well in that it measures “perceptions of the
extent to which a country’s citizen’s are able to participate in
selecting their government, as well as freedom of expression,
freedom of association, and a free media” [66]. Like all the
World Bank Governance metrics, it was built from surveys
and other sources of data within each country. These sources
give an idea of “freedom of belief, opinion, discussion, speech,
publication, assembly, demonstration, and petition” present in
the country, although they do not measure the proliferation
of channels of communication that Diamond enunciates. The
World Bank governance indicators metrics are based on 35
data sources some of which yield “subjective or perceptions-
based data” including that from “household and firm survey
8
respondents, as well as thousands of experts working for the
private sector, NGOs, and public sector agencies” [66].
In points eight through ten, Diamond gives a description
of the role of law in a democracy. The World Bank has a
governance metric that expresses some of this: “measuring
perceptions of the extent to which agents have confidence in
and abide by the rules of society, and in particular the quality
of contract enforcement, property rights, the police, and the
courts, as well as the likelihood of crime and violence” [66].
This approximates the World Banks “Rule of Law” metric and
we propose it as an empirical measure of Diamond’s points
eight through ten.
The World Bank also has a metric measuring political sta-
bility: “perceptions of the likelihood that the government will
be destabilized or overthrown by unconstitutional or violent
means, including politically-motivated violence and terrorism”
[66]. While not enumerated in either Diamond’s or Dahl’s
lists, Amartya Sen postulated an empirical correlation between
democratic regimes and political stability [67]. He notes both
the “political incentives provided by democratic governance”
to prevent crises and specifically that the “positive role of
political and civil right applies to the prevention of economic
and social disasters in general” [67]. Thus we investigate the
World Banks “Political Stability” metric as another measure
of democratic tendency.
In measuring the relationship between ICT penetration and
these democratic variables, it is clear that country wealth is a
confounding factor that sound be taken into account: wealthier
countries are both more likely to be democratic and to be
the heavier users of both the Internet and the cell phone. We
gathered gross domestic product (GDP) data from 2000 to
2006 from the International Monetary Fund. The GDP data is
purchasing parity adjusted to be directly comparable between
countries.
Diamond notes that country size is highly correlated with
regime type: “countries with populations under one million are
much more likely to be both democracies and liberal democ-
racies. Two-thirds of these countries are liberal democracies,
while only 30 percent of countries with populations over one
million are. Among the larger 150 countries, only half are
democracies, while 70 percent of the small countries are. The
countries with populations over one million are about twice as
likely as small states to have an electoral authoritarian regime
and half again as likely to have a closed authoritarian regime.”
[64], [65]. Because of this, we included population in our
models to control for country size. The population data for
2000 to 2006 was also gathered from the IMF.1
B. Limitations of The Data
The ICT data is gathered from the International Telecom-
munications Union (ITU). The ITU requested the number of
Internet and cell phone users from each country. This raises
a host of questions about the reliability of the data since it
1 Both the IMF GDP and population data are available at http://www.imf.
org/external/pubs/ft/weo/2008/01/weodata/

is self-reported by the country. Perhaps the country has an
incentive to under or over-report? It is plausible that some
countries have more reliable data collection mechanisms in
place than others. With one report per country per year, the
data are highly granular, but they do seem to follow steady
trends, and steadily upwards in ICT adoption.
The World Bank did not calculate governance metrics for
2001. We carried out simple linear interpolation to provide a
data point for each country in this year. We made the decision
to do this since there are only six time points from which
to interpolate, and only one, the year 2000, falls before our
missing data. This creates data that are “too smooth” for year
2001 since they have been created from our pre-existing data
and this will make our results appear more precise than they
in fact are. We argue that this effects is minor since the World
Bank data are themselves aggregated from a large number of
sources, and thus less subject to noise than using a single
source would be.
Other metrics of the level of democratic rights exist, such as
the Freedom House “Freedom in the World” metric. Freedom
House carries out an annual global survey of political rights
and civil liberties. We choose not to use this as a measurement
of democratic tendency for two reasons. We felt that the World
Bank Governance Indicators could be well grounded in the
theory of democracy as measures of democracy. Secondly, the
Freedom House measures have some quantitative limitations.
A certain amount of inertia is built into the measurements so
that it is difficult for a country to move much from one year to
the next. For both political rights and civil liberties a country
is scaled from 0 to 7, giving only 8 possible outcomes for
a country. Combining these two factors leads to a database
that does not shift very much from year to year. In the years
of our study, 2000 to 2006, of the 193 countries surveyed by
Freedom House (after subtracting the 9 with missing values
for both political rights and civil liberties for the entire time
series), 105 had no change in their scores for political rights
and 86 had no changes in their civil rights scores. The average
variance of those that did exhibit some change from 2000 to
2006 was 0.48 for political rights and 0.32 for civil liberties.
This means the majorities of countries, if they changed at all,
changed by perhaps one point on the 8 point scale.
Having more years of data, including 2007, would improve
our modeling. At the time of this writing, the World Bank
Governance metrics were not available for 2007.
The IMF estimated some of the population number for some
of the countries. It is likely this has the effect of providing
population data that is smoother than it would otherwise be.
Note also that both ICT measures, the Internet and Cell Phone
use, are measures per 100 inhabitants. We emphasize that this
must be carefully noted in interpretation of the regression
results, since we use population as an independent variable.
We also note that even though the ITU collects Internet use
statistics for each country, what it really means to use the
Internet can vary by country due to filtering, censoring, and
other restrictions on access. The OpenNet Intiative at the
Berkman Center for Internet and Society monitors the filtering
9
activity for 40 of the countries most actively engaged in
repressing internet activity [68], [69]. Although not as extreme
as the case of the Internet, cell phone use can be restricted
by the government as well, and will differ from country to
country. Zuckerman gives several examples of government
crackdowns on mobile phone use: Belarus’s reported shutdown
of their SMS network in March 2006, reports of Ethiopian
cell phone blocking during the 2005 election protests, and
Cambodian blocking of SMS for two days before their 2007
elections [7].
North Korea and Cuba were dropped from the study since
official data is not reported for these countries. It is generally
known that there is very little internet access in Cuba, and little
to none in North Korea and these are both regimes with little
democracy. If we had been able to include these countries in
our estimation of the models, this would likely have bolstered
our results.2 Details of the data cleaning and amalgamation
process are on the study website at http://www.stodden.net/
ICTD.
Our population data was obtained from the International
Monetary Fund and contains a sparse number os missing
values. The IMF has made estimates of their missing data
to complete the dataset.3
III. E MPIRICAL M ODELING
Our data comprises a panel containing N different times
series each consisting of T observations. The number of
countries, N , is 181, and T , the number of years in our study,
is 7. A fixed effects model of our democratic measures’ effect
on ICT penetration follows:
ICTit = β0 + β1 RLit + β2 V Ait + β3 P Sit +
β4 P OPit + β5 GDPit + β6 M Fit +
γt T D + ξi CD + it
i = 1, . . . , N, t = 1, . . . , T.
In this paper we model the penetration of Internet or cell
phone usage per 100 inhabitants, ICT , as a function of the
World Bank democratic measures (Rule of Law, Voice and
Accountability, and Political Stability)4 , country size, country
wealth, and the male/female ratio in the country.
ICT is one of “Internet Use” or “Cell Phone Use.”5 GDPit
is the per capita Gross Domestic Product for country i at
year t, adjusted for purchasing power parity.6 P OPit is the
2 Countries with missing values also tended to support our hypothesis:
Afghanistan, Bhutan, Comoros, Kiribati, Serbia, St. Kitts and Nevis, Timor-
Leste, and Tonga. Cuba, Iraq, Montenegro, and North Korea simply did not
furnish enough data for inclusion in the study.
3 A precise explanation of their data interpolation procedure was not readily
available. See http://www.econstats.com/weo/V023.htm
4 Available at http://info.worldbank.org/governance/wgi/
5 The data used in this study is available at http://www.itu.int/ITU-
D/ict/informationsharing/
6 The data are available at
http://www.imf.org/external/pubs/ft/weo/2008/01/weodata/index.aspx

population of country i at year t and M Fit is the male/female
gender ratio.7
β0 is the intercept term for country i, and γt T D and ξi CD
are a time effect and a country effect, respectively. The effect
of time is controlled for by dummy variables: T Dt is 1 for year
t and 0 otherwise. Similarly CDi is a dummy variable that is
1 for country i and 0 otherwise, controlling for the differences
between countries. Finally, it is a disturbance term with
distribution N (0, σi2 ), which we assume to be uncorrelated
across country cross sections. In this study there are N = 181
countries and T = 7 years.
A. Autocorrelation in Panel Data
A panel regression model of this type is subject to possible
autocorrelation between subsequent observations because of
the time series components. In a regression model as described
above, it is possible to ’discover’ what Granger and Newbold
[70] termed “spurious” relationships between the variables.
That is, tests of significance on estimated coefficients may
indicate a significant result, when in fact none is present.
Granger and Newbold suggest economic time series data may
be especially prone to autocorrelation since they tend to be
non-stationary, in that it is not uncommon for the process
generating the data to depend on the time it was sampled.
For example, economic time series are commonly subject
to seasonal or cyclical effects. Time series data that are
not stationary will violate the assumptions of least squares
regression since the variance of the error term will depend
on time and thus introduce a bias into coefficient estimation.8
Granger and Newbold describe a high R2 value and a low
Durbin-Watson statistic as warning signs that the estimated
model may be yielding spurious results.
Thus it is important to determine whether the time series
data in this study are nonstationary. Our data are sampled
yearly suggesting they may avoid annual cyclical effects.
We analyze 181 countries from 2000 to 2006 (with 2001
interpolated for the World Bank variables) and thus have 7
values in each time series. As Granger and Newbold mention,
finely sampled time series tend to exacerbate the cyclical
effects and thus nonstationarity in the data.
As is typical, the Augmented Dickey-Fuller test was used to
assess nonstationarity in each of our time series. Since Internet
and Cell phone use are increasing rapidly for the vast majority
of the countries in our study, we measured the autoregressive
structure of each time series as stationary around a trend line,
and modeled with one lag. The null hypothesis is that the
data are nonstationary. The test was applied to the ICT and
World Bank data at the country level. For the Internet and
mobile phone data the test rejected the null hypothesis of
nonstationarity for 40 and 35 of 181 countries respectively,
meaning that for around 20% of the Internet and cell phone
penetration time series, nonstationarity is not evident.9 . When
7 The gender ratio data was obtained from the Census Department’s Inter-
national Database at http://www.census.gov/ipc/www/idb/tables.html
8 For a mathematically precise explanation see [70], p. 2.
9 The tests were performed at the 10% level
10
the same test was performed on the World Bank variables, 54,
62, and 60 of 181 rejected nonstationarity for Rule of Law,
Voice and Accountability, and Political Stability, respectively
(about one third of the data). Although there some evidence
of stationarity, it appears that the majority of the time series
included in this study are nonstationary, and it’s potential
biasing of coefficient estimates is a concern.10 The typical
remedy is the difference the data to remove the nonstationarity.
Running the regressions in the above equation allowed us to
carry out tests on the residuals directly to evaluate the level of
autocorrelation. There are two regressions to be run in this
study, modeling Internet penetration and cell phone usage.
Typically the Durbin-Watson test with one lag is used to test
for autocorrelation in the structure of the regression residuals
and the regression with Internet use as a dependent variable
was found to have autocorrelation present, and the cell phone
penetration regression was not, with Durbin-Watson values of
1.86 and 2.01 respectively.11
The Durbin-Watson statistic is created by calculating d =
PT
(t −t−1 )2
t=2P
t 2 , where t is the tth residual from the regres-
1 t
sion. It follows that 0 < d < 4. A value of 2 indicates
no autocorrelation. To test whether the Durbin-Watson test
statistics could be considered equivalent to 2, the test in the
R statistical software package was used [71].
The adjusted R2 values were 0.9246 and 0.9199, seeming to
fit the Internet regression squarely into Granger and Newbold’s
area of caution: a low Durbin-Watson statistic and a high
R2 value, and also casts some suspicion on the cell phone
regression. Granger and Newbold offer that until “a really
satisfactory procedure is available, we recommend taking first
differences of all variables that appear to be highly autocorre-
lated.” (p. 8.) We carried out this operation on both the Internet
and cell phone use regressions because of the high R2 values
and the nonstationary data in both regressions, even though the
Durbin-Watson statistic did not suggest autocorrelation among
the cell phone regression errors. The plots of the residuals
for both regressions indicate possible heteroskedasticity. This
suggests running the following differenced model:
∆ICTit = β00 + β10 ∆RLit + β20 ∆V Ait + β30 ∆P Sit +
β40 ∆P OPit + β50 ∆GDPit + β60 ∆M Fit +
γt0 T D + ξi0 CD + it
i = 1, . . . , N, t = 2, . . . , T.
Running this model for differenced Internet penetration and
differenced mobile phone use did not improve the Durbin-
10 Note that a combination of nonstationary time series may in fact be
stationary. This is termed cointegration.
11 The p-values for the Durbin-Watson test are generated via a bootstrapping
method and can fluctuate. In this case the p-values were 0.022 for the Internet
regression and 0.962 for the cell phone regression.
 11
TABLE I TABLE II
S UMMARY S TATISTICS FOR 2000 DATA S UMMARY S TATISTICS FOR 2006 DATA

Variable Mean Min Max Variable Mean Min Max

Internet Use per 100 7.25 0 45.58 Internet Use per 100 21.66 0.03 92.52
Cell Use per 100 16.02 0 81.73 Cell Use per 100 53.19 0.42 138.06
Rule of Law -0.07 -2.02 1.95 Rule of Law -0.07 -2.00 2.03
Voice & Accountability -0.06 -2.05 1.67 Voice & Accountability -0.07 -2.28 1.72
Political Stability -0.08 -2.73 1.54 Political Stability -0.08 -2.31 1.60
GDP (PPP per capita) 8,998.10 229.36 55,248.25 GDP (PPP per capita) 12,147.70 195.43 76,537.15
POP (millions) 33.20 0.04 1267.43 POP (millions) 35.67 0.051 1314.48
Gender Ratio (M/F) 100.4 85.3 212.3 Gender Ratio (M/F) 100.4 84.3 218.5

TABLE III
R EGRESSION C OEFFICIENT E STIMATES FOR G LOBAL I NTERNET
12
Watson statistics. They became 2.16 and 1.92 respectively. P ENETRATION
The adjusted R2 values were reduced to 0.3686 and 0.4059.
Variable Estimate Standard Error p-value
Both Durbin-Watson statistics reject the null hypothesis of
Rule of Law 0.9018 1.1193 0.4207
no autocorrelation at the 5% level. A common method of Voice & Accountability 0.9122 0.8514 0.2844
controlling for autoregression is using a 2-stage least squares Political Stability -0.3783 0.5345 0.4793
approach [74], [75]. In the first stage, the autoregressive GDP 0.0004 0.0002 0.0972*
POP -0.0488 0.8123 0.9521
structure in the residual is estimated using a model postulating Gender Ratio 0.8212 0.9189 0.3718
that the autocorrelation has a single lag structure, specifically:

i = ρ ∗ i−1 A. Modeling Global ICT Penetration as a Function of Demo-

where i is the ith residual from the initial regression.
The first stage allows us to find an estimate of ρ, ρ̂ using
a least-squares model. In stage 2, ρ̂ is used to remove the
autocorrelation in the variables. Each variable, represented as
V AR in the next equation, is then adjusted to create a new
variable, adjV AR, using the formula:
cratic Tendency
We estimated two panel regressions of ICT penetration with
controls for autocorrelation as discussed in the preceding sec-
tion (differencing and the Durbin-Watson correction). Internet
and cell phone use were modeled as functions of demographic
variables along with control variables:

∆Internetit = β00 + β10 ∆RLit + β20 ∆V Ait + β30 ∆P Sit +

adjV ARi−1 = V ARi − ρ̂V ARi−1
Since we are analyzing panel data with both time and
country dimensions, implementation of the autocorrelation was
applied at the country level using different estimates of ρ for
the Internet and cell phone regressions. Carrying this out on the
differenced data increased the Durbin-Watson statistics to 2.24
and 2.18, and autocorrelation is still detected by this test at the
5% level. The adjusted R2 statistics were 0.4292 and 0.3599.
This was our final model analyzed in the following section.
Since autocorrelation has not been eradicated from the data
we interpret our results cautiously and look for corroboration.
IV. R ESULTS AND F INDINGS
We used the R statistical package to estimate these models
(version 2.7.2) [71]. The complete set of code and data used
in this study can be found at http://www.stodden.net/ICTD.
Tables I and II present summary statistics of the dependent
variables (Internet and cell phone use per 100 country inhab-
itants) and the World Bank metrics for the countries in 2000
and 2006 respectively.
12 Testing the residuals for autocorrelation using the Durbin-Watson test is
the typical procedure, when the sample is large. It is also possible to test for
cointegration: whether the combination of time series is stationary. See [72]
and [73].
β40 ∆P OPit + β50 ∆GDPit + β60 ∆M Fit +
γt0 T D + ξi0 CD + it
i = 1, . . . , N, t = 2, . . . , T.
The regression coefficients for Internet penetration are given
in Table III. The coefficients on the individual country and
time dummy variables are not included for space reasons.13
The most significant coefficient was GDP, and none of the
democratic measures were significant. The positive coefficient
on GDP confirms our intuition that wealthier countries have
higher levels of Internet use.
The same regression was run for cell phone penetration, and
the coefficient estimates are presented in Table IV. Cell phone
use appears not to associated with wealth as Internet use is,
and the male to female gender ratio in the country is strongly
associated with increases in cell phone use, as is the level of
political stability.
The influence of the gender ratio may be driven by outlier
countries: Most countries had a male/female gender ratio of
a little less than 100, implying slightly more females than
males in the population. The coefficient of 4.88 implies that
13 The complete regression results can be found online at http://www.
stodden.net/ICTD.
 12
TABLE IV TABLE VI
R EGRESSION C OEFFICIENT E STIMATES FOR G LOBAL C ELL P HONE C ELL P HONE U SE PER 100 I NHABITANTS FOR H IGHER G ENDER R ATIO
P ENETRATION C OUNTRIES , 2000 AND 2006

Variable Estimate Standard Error p-value Rank Country 2000 Country 2006
Rule of Law -1.4120 2.0333 0.4876 1 United Arab Emirates 43.98 United Arab Emirates 118.51
Voice & Accountability -2.458 1.5946 0.1236 2 Qatar 19.90 Qatar 109.6
Political Stability 2.2823 1.0044 0.0233** 3 Kuwait 21.74 Kuwait 91.49
GDP -0.0002 0.0005 0.7061 4 Oman 6.63 Maldives 87.88
POP 0.1225 1.4327 0.9319 5 Bahrain 30.61 Bahrain 122.88
Gender Ratio 4.8839 1.7342 0.0050*** 6 Saudi Arabia 6.40 Oman 69.59
7 Maldives 2.83 Saudi Arabia 78.05
TABLE V 8 Bhutan 0 Bhutan 9.77
T OP 10 M ALE /F EMALE R ATIOS , 2000 AND 2006 9 Jordan 7.72 Jordan 74.4
10 Djibouti 0.04 Grenada 44.59
Rank Country 2000 Country 2006
1 United Arab Emirates 212.3 United Arab Emirates 218.5
2 Qatar 198.1 Qatar 202.5 confounding factors.
3 Kuwait 150.3 Kuwait 152.3
4 Oman 131.3 Maldives 127.3 The World Bank measure for political stability is also
5 Bahrain 129.9 Bahrain 126.7 significantly positively correlated with increased cell phone
6 Saudi Arabia 125 Oman 124.7 use. This finding suggests that political instability is related
7 Maldives 117.3 Saudi Arabia 120.5
8 Bhutan 112.3 Bhutan 111.0 to the mass diffusion of cell phone usage. In other words,
9 Jordan 109.9 Jordan 110.2 an increase in cell phone availability could increase the
10 Djibouti 107.1 Grenada 108.1 perceived likelihood that the government will be destabilized
or overthrown by unconstitutional or violent means. In their
statistical analysis, Mansfield and Snyder find that the process
as the gender ratio increases by about 5, cell phone usage per of democratization itself is indeed a destabilizing one [76].
100 inhabitants will increase by one phone. Throughout the “Certainly, the virtues of working democratic structures do not
years studied, roughly 7 of the 10 countries with the highest translate into a carefree path to the stabilization of democracy”
male/female ratio each year were located in the middle east, [77]. This finding also supports the arguments presented by
and the ratios at that end of the distribution dwarfed the other Rosenau and Johnson, as well as Shirky, who opine that the
countries’. As displayed in Table VI, it is plausible some of Internet can be used as a tool by civil society to organize
these values are extreme enough to have a large impact on the collectively to effect political change [25], [78].
regression fit, although why this did not occur in the internet
regression is not clear. It is also possible the very high gender B. Modeling ICT Penetration as a Function of Democratic
ratio values represent a ’middle east effect’ since a number Tendency Among the Most and Least Affluent Countries
of those countries are highly represented in the top 10 gender Examining ICT penetration for different strata of wealth
ratio values. This implies that the coefficient on the gender may help isolate effects that are characteristic of those groups.
ratio variable could represent a high growth in cell phone use Since wealth is a driver of investment one would expect
in the middle east. Table VII gives the 2000 and 2006 cell GDP to play a role in the country’s readiness and ability to
phone data for these countries. The coefficient indicates that adopt new communication technologies [79], [80], [81]. As
as the rate of change in the proportion of men increases, so established in the literature we also found increases in GDP
does the rate of change in cell phone use. to be associated with increases in Internet use. We choose
Interestingly, the greater political stability and the lower the to examine ICT penetration in both the top and bottom 20%
perceived threat of violence, the greater cell phone penetration. of countries by 2006 GDP more closely. Our focus on these
This may represent infrastructural stability if associated with groups, in particular the bottom quintile, is driven by Mansfield
political stability and thus a measure of investor’s confidence. and Snyder’s work theorizing the instability of emergent and
It is not clear why this factor would not therefore also be transitional regimes and the existence of the global digital
associated with an increase in Internet use. Perhaps cell phones divide [82].
are easier to proliferate than access to the Internet and so a Table VII lists the countries that fall into each of these
smaller increase in political stability encourages cell phone groups.
increase before Internet increase. As in the previous section we fit a model with an ICT
As shown in Tables I and II, the average cell phone penetration measure as the explanatory variable, and measures
penetration in 2000 was about 16 phones per 100 inhabitants of democratic tendency and controls as independent variables
and in 2006 it was about 53, nearly a three-fold increase. This for a panel regression over years 2000 to 2006. For the
is a high rate of increase but, notably, the countries listed top wealthiest quintile of countries our model did not yield
in Table VI (those with the highest male to female gender statistically significant results for Internet penetration. Table
ratios) had much higher than average growth in cell phone VIII gives the coefficient estimates. Since these countries are
penetration. This is quantified in the significant coefficient in exceptionally wealthy and relatively stable politically it may
the regression in Table IV, while allowing for the included not be a surprise that GDP is not a driver of Internet use, and
 13
TABLE VII TABLE IX
T OP AND B OTTOM 20% OF C OUNTRIES BY GDP IN 2006 I NTERNET P ENETRATION , B OTTOM 20% OF GDP (2006)

Top 20% Countries Bottom 20% Countries Variable Coefficient Standard Error p-value
Qatar Zimbabwe Rule of Law 0.3702 0.4081 0.3660
Luxembourg Congo (Dem. Rep.) Voice & Accountability -0.8115 0.3656 0.0281**
Brunei Darussalam Liberia Political Stability -0.2583 0.1822 0.1585
Norway Burundi GDP -0.0002 0.0014 0.9053
Singapore Guinea-Bissau POP -0.0309 0.2784 0.9117
United States Afghanistan Gender Ratio 0.4026 0.3235 0.2155
Ireland Sierra Leone
Switzerland Niger TABLE X
Hong Kong, China Central African Rep. C ELL P HONE P ENETRATION , T OP 20% OF GDP (2006)
Kuwait Ethiopia
Iceland Malawi
Canada Eritrea Variable Coefficient Standard Error p-value
Netherlands Mozambique Rule of Law 2.354 6.161 0.7029
Austria Togo Voice & Accountability 4.012 4.736 0.3984
Denmark Rwanda Political Stability 5.071 1.887 0.0612*
United Arab Emirates Uganda GDP -0.0008 0.0005 0.1071
Sweden Myanmar POP -2.896 6.483 0.6558
Australia Mali Gender Ratio 2.8300 4.3690 0.5182
Belgium Madagascar
United Kingdom Guinea
Finland Comoros
Germany Tanzania political turmoil and it may be the case that countries with
Japan Nepal higher Voice and Accountability rankings have been reluctant
France Burkina Faso to permit the growth of the Internet in their milieu.
Bahrain Lesotho
Italy Bangladesh Table X gives the coefficient estimates from the panel
Spain Gambia regression for cell phone penetration for the top 20% of
Taiwan, China Haiti wealthiest countries. There is a statistically significant effect in
Greece Zambia
Cyprus Ghana the World Bank metric of Political Stability: greater Political
New Zealand Sao Tomé & Principe Stability is associated with an increase in cell phone use per
Slovenia Benin inhabitant. This result seems intuitive as political stability is
Israel Kenya
Bahamas Senegal historically associated with greater investment in communica-
Korea (Rep.) Cambodia tions infrastructure and is consistent with our earlier regression
Saudi Arabia Chad on cell phone use. This suggests the wealthiest countries may
Czech Republic Côte d’Ivoire
be driving the correlation between political stability and cell
phone penetration. None of the other variables were found
TABLE VIII
I NTERNET P ENETRATION , T OP 20% OF GDP (2006) to have a statistically significant relationship with cell phone
penetration in the wealthiest countries.
Variable Coefficient Standard Error p-value Among the poorest countries, growth in Voice and Ac-
Rule of Law -0.3709 5.4055 0.945 countability had a statistically significant negative effect on
Voice & Accountability 4.771 4.0981 0.246
Political Stability -1.0344 2.2760 0.650 growth in cell phone penetration. Although consistent with
GDP 0.0003 0.0004 0.450 the Internet penetration regression results for this group of
POP 1.6791 5.5724 0.764 countries, the paradox remains as to the increase in per capita
Gender Ratio -1.7979 3.7568 0.633
cell phone use as Voice and Accountability decreases. It is
plausible citizens desire newer forms of ICT when Voice
and Accountability is restricted. Another explanation may be
nor are the measures of democratic tendency, even though our that although not all countries in the bottom quintile are
subsets contains countries with varying levels of autocratic autocratic, a significant proportion are and cell phone use may
control. facilitate the mobilization, organization and coordination of
Among the least wealthy quintile we fit the same panel resistance against autocratic rule. Interestingly, the voice and
regression model as above. Table IX gives the regression coef- accountability metric is not a significant predictor globally, yet
ficient estimates. Our predictors did not yield highly significant is significant for types of ICT among the poorest countries.
coefficient estimates with the exception of the World Bank
voice and Accountability metric. Voice and Accountability V. C ONCLUSIONS AND F UTURE R ESEARCH
is negatively correlated with Internet penetration: implying This paper is the first to our knowledge that uses recent
that when countries notch up in the Voice and Accountability Internet and cell phone use data in an empirical study of
ranking, the use of the Internet increases. This seemingly their relationship to democratic tendency. Previous studies uses
paradoxical finding may be explained when note that our measures of Internet use that ended in 1993 [26], [38]. Best
analysis is restricted the the lowest quintile of country in and Wade’s data reached only to 2002. This paper is also the
wealth. These countries experience disproportionately greater first to the best of our knowledge that measures ICT diffusion

TABLE XI
C ELL P HONE P ENETRATION , B OTTOM 20% OF GDP (2006)
Variable Coefficient Standard Error p-value
Rule of Law 0.3702 0.4081 0.3660
Voice & Accountability -0.8115 0.3656 0.0281**
Political Stability -0.2583 0.1822 0.1585
GDP -0.0002 0.0014 0.9053
POP -0.0309 0.2784 0.9117
Gender Ratio 0.4026 0.3235 0.2155
as a function of democratic indicators. Previous research has
focused on whether ICT use predicts democratic measures.
These results support Bimbers assertions that the structure
of political power has not been revolutionized or transformed
into a new epoch of democracy [12]. Evidence can be found
for both the populist thread in the literature and the notion that
ICTs may act as an extension of the ruling class.
We found a statistically significant positive relationship
between the rate of diffusion of the cell phone and the World
Bank’s “Political Stability” measure capturing perceptions
regarding the likelihood that a government will be destabilized
or overthrown by unconstitutional or violent means. Political
Stability continues to have a significant positive relationship
with the rate of cell phone use among the most affluent 20%
of countries, but that relationship does not hold among the
wealthiest countries. The “Voice and Accountability” indicator
which measures perceptions of the extent to which citizens
are able to participate in selecting their government, as well
as freedom of expression, association and of the media,
was a significant negative predictor of the rate of cell and
Internet diffusion among the poorest quintile of countries. The
diffusion of ICT access did not have any significant influence
on “Rule of Law” while the increasing availability of cell
phones were shown to have no influence on any of the three
World Bank indicators.
Recall that the Rule of Law metric measures perceptions of
the extent to which individuals have confidence in and abide
by the rules of society—in particular the quality of contract
enforcement, property rights, the police, and the courts as well
as the likelihood of crime and violence. This measure was orig-
inally included in the analysis based on Diamond’s research on
the characteristics of democracy. However, the findings here
suggest that the increase in “Rule of Law” perceptions does
not influence ICT penetration, either positively or negatively.
In other words, perceptions regarding the “Rule of Law” may
be framed and influenced by factors other than widespread
ICT use.
The populist school of thought believes ICT diffusion will
decentralize access to communication and information while
increasing citizen access [12], while Hill and Hughes claim
that perhaps the Internet will bring about “a wider demo-
cratic revolution in the world” [22]. In our modeling Internet
diffusion was not predicted by our measures of democracy,
implying this revolution has not yet arrived. The rate of cell
phone diffusion was predicted by higher rates of the “Political
Stability” metric. Recall that cell phone are much more widely
14
used globally than the Internet, so this may be a function of
cell technology’s earlier foothold than the Internet’s. As Inter-
net diffusion catches up to that of cell phone, the democratic
metrics may be found to be predictors of this diffusion. This
is not necessarily inconsistent with the populist thread in the
literature as, according to Bimber, other scholars believe that
the Internet may have a “transformative potential” because
of the “one-to-one interaction among citizens and between
citizens and government” it creates [12].
A paradox is created for the populist school in the finding
that “Voice and Accountability” is negatively correlated with
cell phone diffusion. This can be interpreted in favor of the
argument that ICT diffusion can be centralized by government
turning them into social control mechanisms [43]. As the rate
of Voice and Accountability increases, the rate of diffusion
of ICTs decreases among the poorest and least developed
countries. Rates of cell phone use increase globally with
decreases in political stability, lending further support to the
thesis of ICTs as an extension of ruling class control.
This paradox may be resolved is we consider Page’s view
in 1995, that ICT diffusion may be still too nascent and
insufficient to generate a well functioning system of political
communication and deliberation [48]. It appears we are still
too early to expect a close relationship between a vibrant
public sphere and ICTs globally. As Bimber states, the “central
theoretical problem for the populist claim is the absence of a
clear link between increases in information and increases in
popular political action” [12]. This paper provides evidence
of the existence of this problem and the need to develop our
understanding of this dynamic further.
It would be interesting to tie this research more closely to
development, in line with Sen’s reasoning that “Developing
and strengthening a democratic system is an essential compo-
nent of the process of development,” could extend the empiri-
cal analysis in a fruitful direction. This might mean specifically
testing whether the order in which political and civil rights are
extended as a country emerges from an autocracy affects the
rate of development. This could provide a setting in which to
test the “Lee Thesis,” that political rights should be withheld
until economic development is achieved.
It may be valuable to explore empirical issues further.
Modeling the autocorrelation structure with more than one lag
may help reduce autocorrelation. It would also be interesting
to test for cointegration among these variables. Certainly
documented feedback loops exist between our independent
variables, such as GDP and measures of democracy, and taking
this explicitly into account may improve the modeling [83],
[26]. It is also plausible that feedback loops exist between
ICTs and democratic measures and future modeling could
accommodate this. Further research into the modeling aspects
could estimate models including variables that control for
the different manifestations of cell phone and Internet use in
different countries. A more comprehensive model might ex-
plore possible non-linearities: whether countries with low ICT
adoption rates have different patterns of democratic tendency
than those with high adoption rates.

ACKNOWLEDGMENT
The authors would like to thank Michael Best and Colin
Maclay for helpful discussion and the anonymous reviewers
for constructive criticism. Sadia Ahsanuddin and Kristen Lovin
provided outstanding research assistance.
R EFERENCES
[1] M. Best and K. Wade, “The internet and democracy: Global catalyst or
democratic dud?” June 2006, sam Nunn School of International Affairs.
[2] B. Barber, “Three scenarios for the future of technology and strong
democracy,” Political Science Quarterly, vol. 113, no. 41573-589, 1999.
[3] P. Schmitter and T. Karl, “What democracy is . . . and is not,” Journal
of Democracy, vol. 2, no. 3, 1991.
[4] P. Norris, “Political protest in fragile states,” July 2006, international
Political Science Association World Congress.
[5] F. Zakaria, “The rise of illiberal democracies,” Foreign Affairs, Novem-
ber 1997.
[6] Huber, D. Rueschemeyer, and J. Stephens, “The impact of economic
development on democracy,” Journal of Economic Perspectives, vol. 7,
no. 3, pp. 71–85, 1993.
[7] E. Zuckerman, “Draft paper on mobile phones and activism,” 2007.
[Online]. Available: http://www.ethanzuckerman.com/blog/2007/04/09/
draft-paper-on-mobile-phones-and-activism/
[8] D. Kidd, Indymedia.org: a new communications commons. Routledge,
2003, pp. 47–69.
[9] A. Scott and J. Street, “From media politics to e-protest,” Information,
Communication and Society, vol. 3, no. 2, pp. 215–40, 2000.
[10] S. Andrew, “The control revolution: How the internet is putting individ-
uals in charge and changing the world,” The Century Foundation, Tech.
Rep., 2000.
[11] R. Clarke, “Information technology: Weapon of authoritarianism or tool
of democracy?” August 1994, iFIP World Congress.
[12] B. Bimber, “The internet and political mobilization - research note on
the 1996 election season,” Social Science Computer Review, vol. 16,
no. 4, pp. 391–401, 1998.
[13] R. Dahl, Democracy and its Critics. New Haven, CT: Yale University,
1989.
[14] H. Rheingold, Virtual Reality. New York, NY: Summit Books, 1991.
[15] R. Anderson, T. Bikson, S. Law, and B. Mitchell, “Universal access to
e-mail: Feasibility and societal implications,” RAND, Tech. Rep., 1995.
[16] A. Etzioni, The Spirit of Community: Rights, Responsibilities, and the
Communitarian Agenda. New York, NY: Crown Publishers, 1993.
[17] J. Pavlik, New Media Technology: Cultural and Commercial Perspective.
Boston, MA: Allyn and Bacon, 1996.
[18] L. Grossman, The Electronic Republic. New York: Viking, 1995.
[19] G. Browning, Electronic Democracy: Using the Internet to Influence
Politics. Wilton, CT: Online Inc., 1996.
[20] N. Negroponte, Being Digital. New York, NY: Vintage, 1995.
[21] J. Snider, “New media: Potential information and democratic account-
ability: A case study of governmental access community media,” August
1996, american Political Science Association Convention.
[22] K. Hill and J. Hughes, “Computer mediated political communica-
tion: The usenet and political communities,” Political Communication,
vol. 14, pp. 3–27, 1999.
[23] A. Corrado and C. Firestone, Elections in Cyberspace: Toward a
New Era in American Politics, A. Corrado and C. Firestone, Eds.
Washington, DC: Aspen Institute, 1996.
[24] C. Steele and A. Stein, Communications revolutions and international
politics. Albany, NY: State University of New York Press, 2002.
[25] J. Rosenau and D. Johnson, Information technology and turbulence in
world politics. State University of New York Press, 2002.
[26] C. Kedzie, “Communication and democracy: Coincident revolutions
and the emergent dictator’s dilemma,” RAND, Tech. Rep., 1997.
[Online]. Available: http://www.rand.org/publications/RGSD/RGSD127
[27] S. Lipset, “Some social requisites of democracy: Economic development
and political legitimacy,” American Political Science Review, vol. 53, pp.
69–105, 1959.
[28] R. Jackman, “On the relation of economic development to democratic
performance,” American Journal of Political Science, vol. 17, pp. 611–
21, 1973.
[29] K. Bollen, “Political democracy and the timing of development,” Amer-
ican Social Review, vol. 44, pp. 572–87, 1979.
15
[30] R. Burkhard and M. Lewis-Beck, “Comparative democracy: the eco-
nomic development thesis,” American Political Science Review, vol. 88,
pp. 903–10, 1994.
[31] J. Longdregan and K. Poole, “Poverty, the coup trap, and the seizure of
executive power,” World Politics, vol. 42, pp. 151–83, 1990.
[32] ——, “Does high income promote democracy?” World Politics, vol. 49,
pp. 1–30, 1996.
[33] B. Geddes, “What do we know about democratization after twenty
years?” Annual Review of Political Science, vol. 2, pp. 115–44, 1999.
[34] F. Webster, Globalization, information and change. State University of
New York Press, 2002.
[35] R. Reich, The Work of Nations: Preparing ourselves for 21st century
capitalism. Vintage, 1992.
[36] T. Friedman, The Lexus and the Olive Tree: Understanding Globaliza-
tion. Anchor, 1999.
[37] L. Waverman, M. Meschi, and M. Fuss, “The impact of telecoms on
economic growth in developing countries,” Centre for Economic Policy
Research, Tech. Rep., March 2005.
[38] D. Richards, Making the national international: Information technology
and government respect for human rights. Albany, NY: State University
of New York Press, 2002.
[39] D. Barney, Prometheus Wired: The Hope for Democracy in the Age of
Network Technology. University of Chicago Press, 2000.
[40] A. Mazrui and R. Ostergard, Technology transfer in the computer age:
The African experience. State University of New York Press, 2002.
[41] D. Strienstra, Gender, women’s organizing, and the Internet. Albany,
NY: State University of New York Press, 2002.
[42] S. Kalathil and T. Boas, Open Networks, Closed Regimes. Carnegie
Endowment for International Peace., 2003.
[43] W. Neuman, The future of the mass audience. Cambridge University
Press, 1991.
[44] T. Eyck, “Does information matter? a research note on information
technologies and political protest,” Social Science Journal, vol. 38, pp.
147–160, 2001.
[45] N. Postman, Amusing ourselves to death. New York, NY: Vintage
Books, 1985.
[46] H. Schiller, Culture Inc. Oxford University Press, 1989.
[47] W. Lippmann, Public Opinion. New York, NY: Macmillan, 1934.
[48] Page, Who Deliberates? Mass Media in Modern Democracy. Chicago,
IL: University of Chicago Press, 1996.
[49] G. Cherian, “The limits of singapore’s ‘light touch’ web regulation,”
2008, politics 2.0 International Conference.
[50] R. Deibert, J. Palfrey, R. Rohozinski, and J. Zittrain, Access Denied:
The Practice and Policy of Global Internet Filtering. The MIT Press,
2008.
[51] S. Oats, “Internet and democracy in russia,” April 2008, politics 2.0
International Conference.
[52] R. Singel, “Seeking tighter censorship, repressive states target
web 2.0 apps,” Wired Magazine, March 2008. [Online]. Available:
http://blog.wired.com/business/2008/03/etech-what-happ.html
[53] J. Zittrain, The Future of the Internet—And How to Stop It. Yale
University Press, 2008.
[54] S. Mydans, “Myanmar junta unplugs internet,” The International
Herald Tribune, October 2003. [Online]. Available: http://www.iht.com/
articles/2007/10/04/asia/04info.php
[55] J. Goldsmith and T. Wu, Who Controls the Internet? Illusions of a
Borderless World. Oxford University Press, 2006.
[56] L. Lessig, Code: Version 2.0. Basic Books, 2006.
[57] D. Drezner, “The internet’s effect on state-society relations,” March
2006, international Studies Association Convention.
[58] T. Price, Cyber Activism: Advocacy Groups and the Internet. Wash-
ington, DC: Foundation for Public Affairs, 2000.
[59] J. Goldsmith, “Regulation of the internet: Three persistent fallacies,”
Chicago-Kent Law Review, vol. 73, December 1998.
[60] R. Beilock and D. Dimitrova, “An exploratory model of inter-country
internet diffusion,” Telecommunications Policy, vol. 27, pp. 237–252,
2003.
[61] B. Mesquita and G. Downs, “Development and democracy,” Foreign
Affairs, September 2005.
[62] D. Scheufele and M. Nisbet, “Being a citizen online: New opportunities
and dead ends,” The Harvard International Journal of Press/Politics,
vol. 7, no. 3, pp. 55–75, 2002.
[63] H. Milner, “The digital divide: The role of political institutions in tech-
nology diffusion,” August 2003, american Political Science Association.

[64] L. Diamond, “elections without democracy: Thinking about hybrid
regimes,” Journal of Democracy, vol. 13, no. 2, April 2002.
[65] ——, Developing Democracy: Toward Consolidation. John Hopkins
University Press, 1999.
[66] D. Kaufmann, A. Kraay, and M. Mastruzzi, “Governance matters vii:
Aggregate and individual governance indicators 1996-2007,” World
Bank, Tech. Rep., 2007.
[67] A. Sen, Development as Freedom. Oxford University Press, 2001.
[68] “Opennet initiative.” [Online]. Available: http://opennet.net/
[69] R. Faris and N. Villeneuve, Access Denied: The Practice and Policy of
Global Internet Filtering. MIT Press, 2008, ch. 1.
[70] C. Granger and P. Newbold, “Spurious regressions in econometrics,”
Journal of Econometrics, vol. 2, pp. 111–120, 1974.
[71] “R.” [Online]. Available: http://www.r-project.org/
[72] C. Granger, “Some properties of time series data and their use in
econometric model specification,” Journal of Econometrics, vol. 16, pp.
121–130, 1981.
[73] ——, “Co-integration and error correction: Representation, estimation
and testing,” Econometrica, vol. 2, pp. 251–176, 1987.
[74] T. Amemiya, “The nonlinear two-stage least squares estimator,” Journal
of Econometrics, vol. 2, pp. 105–110, 1974.
[75] R. Cumby, J. Huizinga, and M. Obstfeld, “Two-step two-stage least
squares estimation in models with rational expectations,” Journal of
Econometrics, vol. 21, pp. 333–355, 1983.
16
[76] E. Mansfield and J. Snyder, “Conceptualizing and measuring democ-
racy:evaluating alternative indices,” International Organization, vol. 56,
no. 2, pp. 297–337, 2001.
[77] H. Barnes, “The contribution of democracy to rebuilding postconflict
societies,” The American Journal of International Law, vol. 95, no. 1,
pp. 86–101, 2001.
[78] C. Shirky, Here Comes Everybody: The Power of Organizing without
Organizations. Penguin Press, 2008.
[79] C. Argevou, “Ols bias in a nonstationary autoregression,” Information
Technology for Development, vol. 8, no. 1, pp. 15–28, 1998.
[80] B. Oyelaran-Oyeyinka and K. Lai, “Internet diffusion in sub-saharn
africa: A cross-country analysis,” Telecommunication Policy, vol. 8,
no. 1, pp. 507–527, 1998.
[81] J. Gibbs, K. Kraemer, and J. Dedrick, “Environment and policy factors
shaping global e-commerce diffusion: A cross country comparison,” The
Information Society, vol. 19, no. 1, pp. 5–18, 1993.
[82] N. Kshetri and N. Dholakia, ENCYCLOPEDIA OF INFORMATION
SCIENCE AND TECHNOLOGY, 2nd edition. Idea Group Publishing,
2008.
[83] P. Norris, Digital Divide: Civic Engagement, Information Poverty, and
the Internet Worldwide. Cambridge University Press, 2001.

A review of the research on mobile use by micro
and small enterprises (MSEs)
Jonathan Donner and Marcela Escobari
Abstract— The paper offers a systematic review of 14 studies
of the use of mobile telephony by micro and small enterprises
(MSEs) in the developing world, detailing findings about changes
to enterprises’ internal processes and external relationships, and
findings about mobile use vs. traditional landline use. Results
suggest that there is currently more evidence for the benefits of
mobile use accruing mostly (but not exclusively) to existing MSEs
rather than new MSEs, in ways that amplify existing material
and informational flows rather than transform them. The review
presents a more complete picture of mobile use by MSEs than
was previously available to ICTD researchers, and indentifies
priorities for future research, including comparisons of the
impact of mobile use across subsectors of MSEs and assessments
of use of advanced services such as mobile banking and mobile
commerce.
Index Terms—Business Economics, Communication, ICTD,
Mobile Communication, MSE, Social Factors
I. INTRODUCTION
T his paper presents a systematic review of fourteen studies
of the use of mobile telephony by Micro and Small
Enterprises (MSEs) in the developing world [1-14].
The majority of non-agricultural enterprises in the
developing world have ten or fewer employees [15, 16]. These
MSEs employ up to 25% of working-age adults in some
countries [16], and while the contribution of MSEs to
aggregate economic growth remains a matter of debate [17],
their importance to household livelihoods and poverty
alleviation is undeniable. Thus, MSEs are the focus of
programs at many of the world’s largest development
institutions [18].
Since the year 2000, the spread of mobile telephony across
the developing world has raised hopes among policymakers
that MSEs will benefit from easier access to
telecommunications. The successful entrepreneur, suddenly
enabled by his mobile phone, has been given a prominent role
in the global development narrative and become a semi-
regular fixture in both the popular press [19, 20] and
practitioner media [21, 22].
Manuscript version received February 20, 2009.
J. Donner is with Microsoft Research India, jdonner@microsoft.com
M Escobari is with the Center for International Development,
marcela_escobari@harvard.edu
17
Like landlines, mobile phones allow people to communicate
at a distance and exchange information instantaneously. Thus,
there is significant potential for mobile use to increase MSE
productivity. However, since the dynamics underpinning this
potential are nuanced, and since current supporting evidence
is scarce and methodologically heterogeneous [9], it is
important to rigorously examine mobile use by MSEs. For
example, there is a difference between using a mobile to serve
existing customers more effectively, and using it to start a new
business. There is a difference between using a mobile to
check market prices and using it bypass a middleman who
carries goods to market. Popular press and practitioner reports
generally fail to make these distinctions.
Fortunately, a small but methodologically diverse set of
research studies have examined mobile use by MSEs in detail.
This paper offers a systematic review of this existing
literature, identifying known patterns of mobile use, as well as
some important gaps in the research.
The review employs distinct foci. First, it offers an
assessment of how mobile use influences the internal process
of an enterprise, using Porter’s value chain model [23].
Second, it offers a corresponding assessment of how mobile
use influences the network of relationships external to the
enterprise—the value system [23] of producers, traders,
wholesalers, retailers and end-customers. Finally, it explores
two elements unique to mobile communication—the increased
spatial and temporal mobility afforded by wireless devices,
and the resulting blurring of the personal and the professional
spheres—to assess how MSE mobile use differs from landline
use.
II. MICRO AND SMALL ENTERPRISES (MSES)
Working definitions of MSEs vary from country to country
and from researcher to researcher [24]. This analysis defines
an MSE as any non-farm i enterprise, formal or informal, with
less than 50 employees, including sole proprietorships, part-
time businesses, and home-based businesses. The size
thresholds draw on Mead and Leidholm, [16], who note that
the absolute majority of such enterprises in the developing
world are sole proprietorships, and that firms with less than 10
employees substantially outnumber larger enterprises.
A number of factors distinguish the term MSE (micro and
small enterprise) from SME (small and medium enterprise).
The terms MSE and SME are acronyms, each combining two
distinct sizes of enterprises into a single reference. However,

with no commonly accepted definitions of the thresholds
between micro, small, and medium, there are often implicit
conceptual overlaps between the acronyms.
Unlike SMEs, the majority of MSEs are informal
enterprises. Once again, however, there is no universal
standard to determine what makes an enterprise informal vs.
formal [6]. In many cases, even the term entrepreneur may be
a bit of a romantic misnomer. Evidence suggests that among
MSEs, only a small minority of enterprises are poised for
growth; most remain small or struggle to survive, and yield a
low return on labor and capital [17, 25].
Though significantly less growth-oriented and productive
(on average) than SMEs and other larger firms [17], MSEs
share a basic similarity with all enterprises; each combines
investments in capital with some labor (their own, their
families’ or their employees) in the hopes of yielding a
product or service whose market value exceeds the cost of
those inputs. Thus, there has always been a thread in the ICTD
literature that seeks to understand how various technologies
could be used advantageously by MSEs [26-28]. Prior to the
widespread introduction of the mobile into the developing
world, the landline’s importance in this regard was already
clear:
Phones are the information-related technology that
has done the most to reduce costs, increase income
and reduce uncertainty and risk. Phones support the
current reality of informal information systems, they
can help extend social and business networks, and
they clearly substitute for journeys and, in some
cases, for brokers, traders and other business
intermediaries. They therefore work “with the grain”
of informality yet at the same time help to eat into the
problems of insularity that can run alongside. Phones
also meet the priority information needs of this group
of communication rather than processing of
information [27: 18]
The quotation focuses directly on the basic tasks of running
a business—reducing costs, increasing income, managing
risk—and links them to core functions of mediated
communication technologies, particularly the substitution for
journeys. As demonstrated elsewhere, [26] the key is
increased productivity.
III. STUDIES ON MOBILES AND MSES
Recently, studies have emerged that directly address how
MSEs in the developing world are using mobiles rather than
landlines or other ICTs. The studies are not as numerous as
the enthusiasm in the popular press might suggest. They are a
tiny fraction of the total literature on mobile use in the
developing world [29]. They have emerged from different
disciplines, and, as relative contemporaries, often do not cite
each other. This section presents two studies representing
distinct methodologies and conclusions, to provide an
example of the range of available perspectives and to set the
stage for the systematic review.
18
Particularly focused and powerful evidence appears in
Jensen’s [10] research on the fishermen of Kerala. Working
with five-year time series data at three fish markets in coastal
India, Jensen and his team found that “the adoption of mobile
phones by fishermen and wholesalers was associated with a
dramatic reduction in price dispersion, the complete
elimination of waste, and near-perfect adherence to the Law of
One Price. Both consumer and producer welfare increased.”
[10: 879] Soon after the introduction of mobile coverage,
fishermen bought mobiles and accumulated lists of up to 100
buyers in their handsets’ address books; subsequently, while
still at sea, fishermen could call a range of possible landing
points and buyers in order to determine the best price and best
place to sell their catch.
By contrast, Jagun, Heeks, and Whalley’s [9] examination
of the mobile’s role in mediating supply chains in the Nigerian
market for traditional hand-woven ceremonial cloth is broad
in scope. It offers a multidisciplinary literature review, a
conceptual framework articulating effects at multiple levels,
and a detailed case study. They describe “process” benefits to
mobile use, as calls at a distance reduce the time of trades and
replace costly journeys. They also describe “structural”
impacts; finding no disintermediation of traders, but rather an
intensification of their role. Traders are more likely to have
mobiles than the less prosperous weavers in the supply chain,
and thus are better positioned to coordinate with a wider range
of downstream customers and to maintain a more dynamic and
responsive set of relationships with weavers. For example,
weavers previously had to pay cash to get their supplies.
Mobiles give weavers access to credit by enabling calls on
their behalf to fabric vendors by traders, who vouch for the
veracity of weavers’ orders, and promise to cover the costs of
the fabric in advance of the completion of the weavers’ work.
IV. METHODS AND CODING PROTOCOL
Many of the studies of mobile use by MSEs are qualitative,
and do not report statistical findings. Even among quantitative
studies, there is little agreement in terms of dependent and
independent variables under scrutiny. Thus, a statistical meta-
analysis would not be applicable [30]. Similarly, a method
designed specifically for comparing ethnographies, such as
reciprocal translation [31] would be unlikely to bridge
qualitative and quantitative studies.
The analysis draws instead on a systematic review
methodology [32] to aggregate findings across the available
studies. By using a standardized protocol, coding each
individual study for the appearance or absence of certain
assertions, the review assesses and parsimoniously represents
what the research literature, in aggregate, suggests about
mobile use by MSEs. The exercise relies on clearly articulated
eligibility criteria to select studies and on standardized
questions to evaluate them. These two levels of
standardization, agreed upon before the formal review
commenced, separates the exercise from a conventional list-
based or thematic/narrative literature review.

A. Selecting studies
Papers were initially identified by online literature and
database searches (using keyword combinations of mobile,
cellular, microenterprise, MSE, etc.), and by snowball
references from the bibliographies of studies already in hand.
Next, these studies were assessed against a series of eligibility
criteria: to be included, studies had to be specific to both
mobiles and MSEs, report generalizable findings, and contain
detailed primary data about mobile use in everyday
conditions. What started out as a reasonably large body of
studies was trimmed back significantly. In order to provide
additional resources to other researchers, however, this section
lists those excluded papers, along with the rationale for the
decisions.
To be included in the review, papers had to be specifically
focused on mobile phones, which excluded some excellent
research on landlines or payphones and MSEs [28, 33, 34].
The papers also had to be about MSEs, not SMEs. Papers that
did not explicitly include sole proprietors and informal
enterprises were excluded [35-37].
We made a more difficult decision to exclude papers that
were not generalizable to a wide range of MSEs. An
important line of research explores how many individuals earn
livelihoods in the mobile business itself, by selling airtime,
fixing handsets or operating village phones [38-40]. However,
these studies treat mobiles as products and services, rather
than enablers of general business processes.
To fit into the evaluation protocol, papers had to offer
sufficient details around the use of mobiles to illuminate their
role in these business processes. A few surveys that were
otherwise topically correct did not yield information of this
kind [41], or blurred the lines between mobiles and other ICTs
to the point where assertions about mobiles in particular were
difficult to extract [42, 43]. Reviews without new primary
data were excluded [44, 45]. When multiple papers drew on
the same set of data [46, 47], only one paper was retained.
Finally, the review focused on analyses of mobile use in
everyday settings, rather than proposals for or evaluations of
new pilot technologies [48-51] or programmatic interventions
by NGOs [52]. The development of such technologies and
programs is central to the ICTD field, but such initiatives yield
different forms of evidence about mobile use than those that
examine MSEs operating on their own.
This limiting exercise forced a trade-off: the remaining
papers clearly describe some element of the use of mobiles by
MSEs in developing countries, but the population of such
studies is relatively small. Thirteen papers and one book were
retained.
B. Evaluation protocol
The process of developing the evaluation questions was
iterative. It was based mostly on an initial reading of the
documents by the researchers, while also integrating current
narratives in the popular and practitioner literatures. An
original goal was to code studies according to subcategories of
MSEs (to discern differences in mobile use between traders
19
and producers, for example), but it became clear that the
population of existing studies is too small to support that
inquiry.
The final protocol employed three distinct foci. First, it
assessed the impact of mobile use on the internal process of an
enterprise, using Porter’s value chain model [23], Fig. 1. The
value chain comprises the activity inputs into a product or
service: inbound logistics, operations (production), outbound
logistics, marketing and sales, and after-sales service. The
value chain also includes supporting functions: firm
infrastructure, human resources, technology development
(knowledge developed or owned by the enterprise), and
procurement. Together these activities can create customer
value in excess of the costs to provide it, yielding profit.
Porter [23: 168], argues that information and communication
technologies can be used to improve almost any of these
primary and supporting activities. Although the value chain
framework was developed with larger enterprises in mind, it
can be applied to MSEs, since in small firms the same
individual can carry out different business-related activities
during the day. (Indeed, even in larger firms there is often not
a perfect mapping between the activities and functions in the
value chain and distinct people or departments). Both
researchers coded individual papers for mentions of the
mobile’s role in any of the primary or supporting functions.
FIGURE 1: PORTER’S VALUE CHAIN
note: image released to public domain as per
http://en.wikipedia.org/wiki/Image:ValueChain.PNG
The second analysis used another Porter framework, the
value system [23] ii , to offer a corresponding assessment of
how mobile use influences the network of interdependencies
and relationships external to the enterprise, including
producers, traders, wholesalers, retailers and end-customers.
An initial reading of the papers identified four categories of
potential impacts. Some research stresses (a) the increased
availability of information in the network; other studies stress
(b) the entry of new actors, particularly buyers and sellers,
into markets. Both factors tend to increase competition, but do

so in different ways. One focuses on the actors in the network,
the other on the information those actors exchange.
Nevertheless, the two changes do not necessarily move in
tandem; it is one thing to assert that the same set of actors
exchange information at lower cost and higher frequency,
another to say that markets have expanded. This information-
vs.-actors cut highlights this distinction. Two other categories
of network impacts can be expressed as assertions that (c)
mobiles help enterprises cut out middlemen and that (d)
mobiles help individuals start new businesses.
Unlike the value chain analysis (coded for affirmative
mentions only), the value system analysis coded for both
affirmations and negations of the four potential impacts. After
the initial reading, we elected to track negations since some of
the sources made a point of arguing against one or more of the
assertions from the popular and practitioner literature.
The third analysis explored how the impact of mobile use
on small enterprises is or is not different from the impact of
landline telephony on those same kinds of businesses [28, 33,
34]. Technological properties of mobile communication make
it inherently more prone to adoption by MSEs than landlines:
it is cheaper to build towers than lay cable, prepay accounts
have no startup costs, and inexpensive/used handsets are
readily available. However, in this case the third analysis
focused narrowly on two differences in use rather than cost or
access.
First, studies were coded for mentions of mobility. Mobility
is a crucial difference between mobiles and landlines—while
landlines connect places to places, mobiles generally connect
people to people, wherever they are and regardless of the time
and situation. This mobility leads to increased individual
addressability, and can change how people structure social
and economic activity [53-55]. Most relevant to this analysis,
mobility may enable the rise of roaming businesses, just-in-
time service and what Townsend [56] has called the “real time
city”. And yet some evidence suggests that in the developing
world many mobiles are purchased as substitutes for landlines,
rather than complements to them [57].
Second, the nature of the mobile as a portable, personal
device means it is particularly easy to use for both personal
and business functions during the same day. Thus, studies of
the role of mobiles in the lives of MSE operators are often
different from studies of the role of the device in the
businesses themselves. The analysis coded for studies that
explore these social functions.
Once the protocol was established, each researcher re-read
the papers, coding them in isolation. We then compared our
codes and resolved any discrepancies through discussion. The
resulting codes are less prone to reflect the bias of a single
reader. Of the 112 cells on the matrix requiring codes, 16
required discussion to resolve coding discrepancies between
the two researchers.
20
V. RESULTS
A. Enterprise value chain
Most studies mention the core processes of marketing and
sales [1, 2, 4, 6, 7, 9-13]. Analyses ranging from Jensen’s [10]
model of searching for the best price for fish to Kamga’s [11]
description of improvements to the local laundry services in
Cote d’Ivoire asserts that mobiles help connect vendors and
buyers, often at a distance and usually at lower cost than an
in-person journey. Esselaar et al. [6] report results of a survey
of SMEs, including 1/3 microenterprises, conducted across 13
countries. “Mobile phones are used more often for keeping in
contact with customers and clients” (p 92). This is the highly
visible, intuitive role of mobiles for small enterprises.
The picture is sparser for other core processes within the
value chain. Three studies mention inbound and outbound
logistics [1, 9, 13], particularly Abraham [1], who details how
fishermen can now use small supply boats (dispatched via
mobile) to stay out fishing longer. Overå [13] describes how
traders in Ghana can time harvests (inbound) and change the
terms of delivery financing (outbound) because of the mobile.
Operations receives two mentions, again by [1], who notes
that fishermen use mobiles to coordinate the timing and
location of when to drop nets and search for fish. Similarly,
[9] describes how weavers call customers mid-process to
revise plans for the garments they are creating.
Only studies by Frempong et al. [7] and Molony [11]
mention after-sales service. Molony describes how Tanzanian
exporters of carved wood use the mobile to elicit feedback and
built trust with buyers after (and ideally between) sales.
In terms of crosscutting functions, five studies reference
procurement [1, 6, 7, 9, 13] and address price search by
buyers of inputs (or by traders). There is little evidence to date
for the mobile’s role in transforming the proprietary
technology, infrastructure or HR functions of MSEs, perhaps
because these enterprises are too small to invest in these
assets. Although studies outside the review [36, 58] provide
anecdotes of small employers giving mobiles to employees,
this infrastructure function is of limited utility for tiny firms
and sole-proprietorships.
B. Industry value system
The second analysis turns the lens outside the enterprise,
towards its location in a network of relationships. The most
common finding links mobile use to an increase in the flow of
information between actors in the value system [1, 2, 6, 7, 9-
14]. The two primary sub-themes are more frequent or wide-
ranging exchanges of price information [1, 2, 13], and a more
generalized discussion of increased communication with
customers [6, 12, 14]. These findings are reflections of the
frequent references to marketing and sales and procurement
activities in the previous value chain analysis. Reference [9]
mentions an increase in the completeness of the information,
but notes that they saw no increase in quality
 21

TABLE I: SUMMARY OF THE MSE AND MOBILE STUDIES IN THE REVIEW

Enterprise Value Chain Industry Value System Uses

Add Add Bypass Start
Studies Cross-cutting Inform- Buyers/ Middle- busi-
Core Processes Functions ation Sellers men nesses Mobility Social
Inbound &
[1] R. Abraham, “Mobile phones and
Outbound Logistics,
economic development: evidence from the Procurement Yes Yes No -- Yes Yes
Operations,
fishing industry in India,”
Marketing & Sales.
[2] J. C. Aker, “Does digital divide or
provide? The impact of cell phones on grain Marketing & Sales -- Yes Yes -- -- Yes --
markets in Niger”
[3] J. Donner, “Microentrepreneurs and
mobiles: An exploration of the uses of
-- -- -- -- -- -- Yes Yes
mobile phones by small business owners in
Rwanda”
[4] J. Donner, “The use of mobile phones by
microentrepreneurs in Kigali, Rwanda: Marketing & Sales -- -- Yes -- -- -- Yes
Changes to social and business networks”
[5] J. Donner, “Customer acquisition among
small and informal businesses in urban
-- -- -- No -- -- -- Yes
India: Comparing face to face, interpersonal,
and mediated channels”
[6] S. Esselaar, C. Stork, A. Ndiwalana, and
M. Deen-Swarra, “ICT usage and its impact
Marketing & Sales Procurement Yes -- -- -- -- Yes
on profitability of SMEs in 13 African
Countries”
[7] G. Frempong, G. Essegbey, and E.
Tetteh, “Survey on the use of mobile Marketing & Sales;
Procurement Yes Yes -- -- -- --
telephones for micro and small business Service
development: The case of Ghana,”
[8] H. Horst and D. Miller, “The Cell
Phone: An Anthropology of -- -- -- -- -- No -- Yes
Communication.”
[9] A. Jagun, R. Heeks, and J. Whalley, Inbound &
“The Impact of Mobile Telephony on Outbound Logistics,
Procurement Yes Yes No -- -- --
Developing Country Micro-Enterprise: A Operations,
Nigerian Case Study” Marketing & Sales
[10] R. Jensen, “The Digital Provide:
Information (Technology), Market
Marketing & Sales -- Yes Yes -- -- Yes --
Performance, and Welfare in the South
Indian Fisheries Sector”
[11] O. Kamga, “Mobile phone in Cote
Marketing & Sales -- Yes -- -- -- Yes Yes
d'Ivoire: uses and self-fulfillment”
[12] T. Molony, “‘I don't trust the phone; it
always lies’: Trust and information and Marketing & Sales;
-- Yes No No -- -- Yes
communication technologies in Tanzanian Service
micro- and small enterprises”
[13] R. Overå, “Networks, distance, and Inbound Logistics,
trust: Telecommunications Development Outbound Logistics, Procurement Yes Yes -- -- Yes Yes
and changing trading practices in Ghana” Marketing & Sales
[14] J. Samuel, N. Shah, and W.
Hadingham, “Mobile Communications in
-- -- Yes Yes -- Yes Yes Yes
South Africa, Tanzania, and Egypt: Results
from Community and Business Surveys”

While it is one thing to say that mobile use accelerates the
flow of information in existing value systems, it is another to
say that mobile use brings new customers or suppliers into the
market. This is the first element in the systematic review in
which there is some disagreement between the primary
studies. Numerous studies present evidence that mobile use
expands markets by allowing MSEs to reach new customers
[1, 2, 4, 7, 9, 10, 13, 14]. Of the Keralan fishermen, [10: 891]
explains “while almost all sales before mobile phones were
conducted via beach actions, fishermen with phones, often
carrying lists with the numbers of dozens or even hundreds of
potential buyers, would typically call several buyers in
different markets before deciding where to sell their catch”.
Similarly, Aker finds that “Grain traders in markets with cell
phone coverage search over a greater number of markets, have
more contacts and sell in more markets. This underscores the
fact that the primary mechanism by which cell phones affect
market efficiency is a reduction in search costs and hence
transaction costs.” [2: 4-5] In the study, mobile use lowered
price dispersion by 21%, increased profits by 29%.
And yet two other studies specifically argue that the phones
have done little to introduce new buyers. Donner [5] finds that
MSEs in urban India are much more likely to recruit
customers via face-to-face channels, rather than via a phone
call. [12] argues that mobiles help accelerate and strengthen
trusting relationships but only among parties that have already
established a face-to-face bond.
Two other general assertions about the impact of mobile
use on MSE value systems receive less support from the
studies. None of the studies asserts that mobiles help MSEs
bypass middlemen. Indeed three of the papers focus
specifically on middlemen, wholesalers or traders as
enterprises, [2, 9, 13] describing how mobiles allow them to
perform their roles more effectively. Another specifically
emphasizes how producers work with existing middlemen in
their industries, rather than routing around them. Rather than
radically restructuring these marketplaces, Molony argues,
“mobile phones can be seen as a facilitating technology for
existing, trust-based relationships” [12: 78]
Similarly, there is relatively little evidence for the assertion
that mobiles help people start new businesses. Only Samuel et
al. [14] make this case, reporting that among a sample of
MSEs in Egypt and South Africa, 26%-29% of businesses
attributed their start to the availability of the mobile. Taking
the opposite position, Horst and Miller [8: 164] argue that
despite some isolated examples to the contrary (taxi drivers
and musicians), “there is no new spirit of enterprise based on
either the cell phone or the internet” among the Jamaican
households in their study. Nevertheless, they argue that
despite a dearth of new enterprises, the mobile is essential to
the economic survival of those households. By allowing
individuals to leverage broad networks of informal social and
financial support through a process Jamaican mobile users call
“link up,” “the phone is not central to making money, but is
vital to getting money.” [8: 165]
In sum, in value systems where mobile telephony is
22
introduced, there is more evidence for changes in degree
(more information, more customers) than for changes in
structure (new channels, new businesses).
C. On attributes of the mobile vs. the landline
Roughly half of the studies described use cases that take
advantage of mobility. Clearly, fishermen take advantage of
wireless telecommunications [1, 10] to place and receive calls
while on the water. This is not only an advantage for
determining which markets to target, but [1] points out that it
also enables fishermen to feel safer while at sea. Traders [2,
13] use the mobile to be individually addressable wherever
they are. Reference [11] illustrates the responsiveness of
businesses that can serve the customer, 24 hours a day, while
[13] portrays “availability as comparative advantage”, and
argues that this more frequent interaction builds trust between
suppliers and customers.
Given that MSE operators often carry their mobiles
throughout the day and into the evening, a blending of
mediated communication for social and instrumental purposes
often occurs. While some of the papers in the review focus
exclusively on the business functions, others [1, 3-6, 8, 11-14]
illustrate this blurring. Blurring occurs at the aggregate level,
—a survey by Donner [4] found that roughly 1/3 of calls made
by MSE owners in Rwanda were business-related. It also
occurs within individual calls—non-business (“chit-chat”)
exchanges increase trust between clients and customers [12,
13]. Finally [8] describes the “link up” process in Jamaica, in
which individuals retain a roster of numbers of friends, family
and acquaintances that can be tapped periodically for loans or
small cash gift transfers. This process intermingles social and
economic functions of mobile use.
VI. DISCUSSION
This paper offers a systematic review of the current
research on the impact of mobile use on MSEs, applying both
an internal (value chain) and external (value system)
perspective. The review finds a pattern of evidence suggesting
that mobiles increase the information available to MSEs.
Some [2, 10] provide quantitative evidence for how this
information translates into reduced price variability and higher
profits per actor. The current studies suggest mobiles are most
useful for streamlining marketing and sales (downstream) and
procurement (upstream) with existing business contacts. In
some cases, studies suggest that mobile use expands the size
of markets by bringing a larger number of buyers and sellers
into the marketplace. However not all studies found evidence
that new customers were acquired. Far fewer studies present
evidence that mobiles enable the creation of new businesses,
or that mobile use re-organizes value systems to allow
producers to bypass middlemen. Indeed, middlemen are
positioned to take advantage of mobiles themselves.
To summarize, the review of the evidence offered across
the thirteen studies suggests that within the MSE sector,
benefits of mobile use accrue mostly (but not exclusively) to
existing enterprises, in ways that amplify and accelerate

material and informational flows, rather than fundamentally
transform them. This summary does not diminish the positive
utility of mobiles to MSEs, but it places that utility in context
and in doing so echoes assertions by Castells [54] and Harper
[59]. Mobile use by MSEs can be understood as an extension
of the information society, not its restructuring [60].
A second theme in the analysis involves mobiles as
complements or substitutes for landlines. Evidence here
remains mixed; while some case studies highlight enterprises
that take advantage of “availability as comparative advantage”
[13], there is insufficient data to determine whether these are
isolated cases or representations of a more universal
condition. The majority of MSEs may take greater advantage
of place-to-place connectivity, than mobility, but this point
merits further study.
A. Generalization and segmentation
Clearly, mobiles offer distinct benefits to MSEs—
everything from more accurate price information and access to
new customers to better after-sales service and procurement,
from increased responsiveness to the opportunity to build trust
at a distance. However, it is currently difficult to determine
whether the various impacts and benefits enumerated in this
review accrue equally to all MSEs. Looking across a diversity
of MSEs, across nations, industries, and different locations in
value systems, the current research points toward a
multiplicity of intertwined and sometimes contradictory
impacts of mobile use, rather than the universal and rather
uncomplicated benefits which have characterized the popular
rhetoric.
To guide future policy or institutional interventions, it would
be helpful to deploy future research against a set of open
questions: Which kinds of MSEs are gaining the most return
on mobile use? Which kinds (and what proportion) of MSEs
are poised to find new customers and expand their markets,
and not simply be more responsive to the ones they already
have? Which kinds (and what proportion) are unlikely to reap
any benefit from mobiles, or actually may be threatened by
changes in mobile use elsewhere in their value systems?
These questions remain open because most studies to date
have been either sector-specific explorations or broad
aggregate surveys; few studies specifically identify
differences in mobile use or impact among subpopulations of
MSEs. By contrast, recent studies of mobile use by farmers
are identifying factors which differentiate between subgroups,
for example, between growers of perishable and non-
perishable crops [61], by distance from local markets [61], or
according to different levels of infrastructural constraint [62].
An important path for further study would apply similar
comparative analyses to assess and predict the impact of
mobile use by different classes of MSEs.
These are hefty quantitative tasks. Future designs will
require increased attention to the factors that distinguish
subgroups of MSEs as well as careful measurement of
desirable outcomes such as productivity, market participation,
or revenue growth. However, this review helps to identify a
23
range of variables for both the independent and dependant
sides of such analyses. Indeed, a quantification of mobile use
by subpopulations of MSEs may begin to close the gap
between micro-level case studies and research on the impact
of mobiles on macroeconomic growth [63].
B. Enterprises, livelihoods or lives?
The conclusions of this review naturally depend on the
methods and theoretical frameworks employed at the primary
and secondary stages of analysis. Ethnographies such as [11]
and [9] tended to discuss a broader range of uses and impacts
than focused quantitative tests [2, 5, 10]. Similarly, the lenses
chosen for this systematic review highlight some dimensions
of MSE use over others. For example, a livelihoods
framework [64], instead of enterprise-specific frameworks
from Porter [23] would have emphasized different patterns.
The dichotomous treatment of new vs. existing businesses
has its limitations. Reference [9] describes the evolution of
some weavers to “coordinator weavers,” suggesting a
transformation in the structural location and internal processes
of some enterprises, but this was an exception among the
studies, most of which looked only to existing enterprises for
their sample. The methods used by [8] can identify
occupational multiplicity—holding down more than one job at
once—in a way that studies focused specifically on existing
MSEs cannot do. Similarly, [7] is able to assert that few
households start new enterprises thanks to mobile use only
because households rather than MSEs are its primary units of
analysis.
These examples reveal that insights about use of mobiles in
MSEs can come from studies that focus not exclusively on
enterprises, but rather on the individuals who manage them.
Focusing on individuals also allows for increased linkage to
research on social uses of the mobile.
C. New applications on the mobile platform
Studies have documented how mobiles can enable
information search and improve communication between
MSEs, customers and suppliers. However, there has been little
evidence to date that suggests mobiles are being used for
information storage or processing. As was the case with
landlines [27], MSEs value voice calls more than any other
function on the mobile, and use the calls to augment, rather
than replace, face-to-face ties. As [6: 99] explains, mobiles
“cannot be used to track inventory, provide cash flow and
income statements, or even more basically, produce formal
letters, marketing campaigns, or brochures”.
Recently, however, various systems have appeared that go
beyond the voice and peer-to-peer texting functions on the
handset. These make the handset approximate a PC (with
processing happening on the handheld), or as the client in a
client-server model, with primary processing happening
elsewhere on the network. These latter models take advantage
of everything from basic SMS [65] to voice to full-blown
mobile internet browsing experiences. Two of the more
promising applications are distributed marketplaces, such as

Cell Bazaar, Manobi, and Tradenet, and mobile-
banking/mobile-payments initiatives such as Kenya’s popular
M-PESA [66]. In addition, a variety of promising pilots are
underway, such as [48-51], which promise to yield further
functionality, for example, in supply chain management.
Full evaluations of the use and impact of these services on
MSEs are not yet available, and are urgently needed. This
baseline analysis can help clarify which impacts these systems
may be having. Findings that suggest, for example, that
mobile trading platforms help MSEs bypass middlemen,
would be even more impressive given this synthesis’
conclusion that there is limited evidence that basic voice calls
can have this effect. Similarly, findings that suggest MSEs are
using m-banking or m-payments applications to transform
credit relationships or otherwise change the procurement and
sales functions could be assessed more accurately against the
baseline of the voice-based behaviors found so far.
VII. CONCLUSION
This paper has summarized fourteen primary research
studies examining mobile use by MSEs; the research generally
concurs with conventional wisdom—mobile use helps many
MSEs become more productive, particularly but not
exclusively via improvements to sales and marketing and
procurement processes. That said, the review suggests that not
all enterprises will prosper from increased access to
telecommunications, and among those that do, their uses of
mobiles will vary across industries and positions in value
systems. As mentioned above, current evidence suggests that
the benefits of mobile use accrue mostly (but not exclusively)
to existing enterprises, in ways that amplify and accelerate
material and informational flows, rather than fundamentally
transform them.
On balance, MSEs are likely to remain unproductive
relative to larger enterprises [17]. However, the improvements
to productivity associated with mobile use do seem to be
improving the livelihoods of many individuals in the MSE
sector.
The results of this review are helpful to the ICTD research
community in at least three ways. First, by disaggregating and
identifying distinct impacts of mobile use, the review provides
a more nuanced and more accurate representation of the value
of mobile use to MSEs than was previously available. Second,
the review identifies a skew (in both sample and implied
impacts) towards existing enterprises that should be noted by
policymakers who may expect mobiles to create new
businesses and new employment. Finally, the review identifies
two priorities for future research: a segmentation and further
quantification of impacts by subsectors of MSEs, and an
assessment of the use of new non-voice advanced mobile
services (such as mobile banking and mobile marketplaces) by
MSEs
REFERENCES
[1] R. Abraham, “Mobile phones and economic development:
Evidence from the fishing industry in India,” Proceedings of the
24
International Conference on Information and Communication
Technologies and Development, (ICTD 2006). Berkeley, CA:
IEEE, pp. 48-56, 2006.
[2] J. C. Aker, “Does digital divide or provide? The impact of cell
phones on grain markets in Niger,” BREAD Working Papers. vol.
177, Durham, NC: Bureau for Research and Economic Analysis of
Development, 2008.
[3] J. Donner, “Microentrepreneurs and mobiles: An exploration of the
uses of mobile phones by small business owners in Rwanda,”
Information Technologies and International Development, vol. 2,
pp. 1-21, 2004.
[4] ___, “The use of mobile phones by microentrepreneurs in Kigali,
Rwanda: Changes to social and business networks” Information
Technologies and International Development, vol. 3, pp. 3-19,
2006.
[5] ___, “Customer acquisition among small and informal businesses
in urban India: Comparing face to face, interpersonal, and
mediated channels,” The Electronic Journal of Information
Systems in Developing Countries, vol. 32, pp. 1-16, 2007.
[6] S. Esselaar, C. Stork, A. Ndiwalana, and M. Deen-Swarra, “ICT
usage and its impact on profitability of SMEs in 13 African
countries,” Information Technologies and International
Development, vol. 4, pp. 87-100, 2007.
[7] G. Frempong, G. Essegbey, and E. Tetteh, “Survey on the use of
mobile telephones for micro and small business development: The
case of Ghana,” Institutes of the Council for Scientific and
Industrial Research, Science and Technology Policy Research
(STEPRI), Accra, 2007.
[8] H. Horst and D. Miller, The cell phone: An anthropology of
communication. Oxford: Berg, 2006.
[9] A. Jagun, R. Heeks, and J. Whalley, “The impact of mobile
telephony on developing country micro-enterprise: A Nigerian case
study,” Information Technologies and International Development,
vol. 4, pp. 47-65, 2008.
[10] R. Jensen, “The digital provide: Information (technology), market
performance, and welfare in the South Indian fisheries sector,”
Quarterly Journal of Economics, vol. 122, pp. 879-924, 2007.
[11] O. Kamga, “Mobile phone in Cote d'Ivoire: Uses and self-
fulfillment,” Proceedings of the International Conference on
Information and Communication Technologies and Development,
(ICTD 2006). Berkeley, CA, pp. 184-192, 2006.
[12] T. Molony, “‘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, vol. 3, pp. 67-83, 2006.
[13] R. Overå, “Networks, distance, and trust: Telecommunications
development and changing trading practices in Ghana,” World
Development, vol. 34, pp. 1301-1315, 2006.
[14] J. Samuel, N. Shah, and W. Hadingham, “Mobile communications
in South Africa, Tanzania, and Egypt: Results from community and
business surveys,” in Africa: the impact of mobile phones. vol.
2007, Vodafone, Ed. London: Moving the Debate Forward: The
Vodafone Policy Paper Series #3, 2005, pp. 44-52.
[15] International Labor Organization, “Key indicators of the labour
market,” International Labor Organization, Geneva 2003.
[16] D. C. Mead and C. Leidholm, “The dynamics of micro and small
enterprises in developing countries,” World Development, vol. 26,
pp. 61-74, 1998.
[17] R. la Porta and A. Shleifer, “The unofficial economy and economic
development,” Brookings Institute, Washington, DC 2008.
[18] D. Snodgrass, “Inventory and analysis of donor-sponsored MSE
development programs,” Development Alternatives Inc.,
Washington, DC 2005.
[19] Economist, “Calling an end to poverty,” The Economist. vol. 376,
2005, pp. 51-52.
[20] S. LaFraniere, “Cellphones catapult rural Africa to 21st century,”
in The New York Times New York, p. 1, 2005.
[21] USAID, “Using cellular phones in Uganda for rural income
generation and more,” Issue 7 ed. vol. 2006: DOT-COM Alliance
E-Newsletter, 2004.
[22] A. Bhavnani, R. Won-Wai Chiu, S. Janakiram, and P. Silarszky,
“The role of mobile phones in sustainable rural poverty reduction,”

World Bank Global Information and Communications Department,
Washington, DC 15 June 2008.
[23] M. E. Porter, Competitive advantage: Creating and sustaining
superior performance. New York: Free Press, 1985.
[24] D. C. Mead and C. Morrison, “The informal sector elephant,”
World Development, vol. 24, pp. 1611-1619, 1998.
[25] R. Duncombe and R. Heeks, “Enterprise development and
information and communications in developing countries:
Supporting ‘ICT flyers’,” Manchester: IDPM, University of
Manchester, 2001.
[26] R. J. Saunders, J. J. Warford, and B. Wellenieus,
Telecommunications and economic development, 2 ed. Baltimore,
MD: Johns Hopkins University Press, 1994.
[27] R. Duncombe and R. Heeks, “Information and communication
technologies and small enterprise in Africa: Findings from
Botswana." vol. 2006 Manchester, UK: Institute for Development
Policy and Management, University of Manchester, 1999.
[28] ___, “Enterprise across the digital divide: Information systems and
rural microenterprise in Botswana,” Journal of International
Development, vol. 14, pp. 61-74, 2002.
[29] J. Donner, “Research approaches to mobile use in the developing
world: A review of the literature,” The Information Society, vol.
24, pp. 140-159, 2008.
[30] R. J. Light and D. B. Pillemer, Summing up: The science of
reviewing research. Cambridge, MA: Harvard University Press,
1984.
[31] G. W. Noblit and R. D. Hare, Meta-ethnography: Synthesizing
qualitative studies. Newbury Park, CA: Sage, 1988.
[32] J. H. Littell, J. Corcoran, and V. Pillai, Systematic reviews and
meta-analysis. New York: Oxford University Press, 2008.
[33] D. Souter, N. Scott, C. Garforth, R. Jain, O. Mascararenhas, and K.
McKemey, “The economic impact of telecommunications on rural
livelihoods and poverty reduction: A study or rural communities in
India (Gujarat), Mozambique, and Tanzania,” Commonwealth
Telecommunications Organisation for UK Department for
International Development, 2005.
[34] R. Duncombe and R. Heeks, “Information, ICTs and small
enterprise: Findings from Botswana,” Small Scale Enterprises in
Developing and Transitional Economies, H. Katrak and R.
Strange, Eds. New York: Palgrave, 2002, pp. 285-304.
[35] J. O. Adeoti and A. I. Adeoti, “Easing the burden of fixed
telephone lines on small-scale entrepreneurs in Nigeria: GSM lines
to the rescue,” Telematics and Informatics, vol. 25, pp. 1-18, 2008.
[36] R. A. Boadi, R. Boateng, R. Hinson, and R. A. Opoku,
“Preliminary insights into m-commerce adoption in Ghana,”
Information Development, vol. 23, pp. 253-265, 2007.
[37] W. Mei and L. H. Yun, “Perpetual contact for business: Mobile
phones and Fujian entrepreneurs,” Proceedings of ANZCA08:
Power and Place, Wellington, NZ, 2008.
[38] S. Aminuzzaman, H. Baldersheim, and I. Jamil, “Talking back:
Empowerment and mobile phones in rural Bangladesh: A study of
the village pay phone of Grameen bank,” Contemporary South
Asia, vol. 12, pp. 327-348, 2003.
[39] H. Ilahiane and J. Sherry, “Joutia: Street vendor entrepreneurship
and the informal economy of information and communication
technologies in morocco,” The Journal of North African Studies,
vol. 13, pp. 243-255, 2008.
[40] J. Lugo and T. Sampson, “E-informality in Venezuela: The 'other
path' of technology,” Bulletin of Latin American Research, vol. 27,
pp. 102-118, 2008.
[41] B. F. Chogi, “The impact of mobile phone technologies on medium
and small enterprises/jua kali (MSEs),” Proceedings of CPRsouth2
Chennai, 2007.
[42] R. O. Opiyo and O. A. K'Akumu, “ICT application in the informal
sector: The case of the Kariokor market MSE cluster in Nairobi,”
Urban Forum, vol. 17, pp. 241-261, 2006.
[43] E. D. Moyi, “Networks, information and small enterprises: New
technologies and the ambiguity of empowerment,” Information
Technology for Development, vol. 10, pp. 221-232, 2003.
[44] K. O. Davis, Cosmas. , “ICTs as appropriate technologies for
African development” IFC/FT First Annual Essay Competition:
Business & Development: The Private Path to Prosperity,
25
Washington, DC: Financial Times Limited / International Finance
Corporation (IFC), pp. 43-54, 2007.
[45] A. T. Rashid and L. Elder, “Mobile phones and development: An
analysis of IDRC-supported projects,” Electronic Journal of
Information Systems in Developing Countries, vol. 36, pp. 1-16,
2009.
[46] T. Molony, “Running out of credit: The limitations of mobile
telephony in a Tanzanian agricultural marketing system,” Journal
of Modern African Studies, vol. 46, pp. 1-22, 2008.
[47] ___, “The role of mobile phones in Tanzania’s informal
construction sector: The case of Dar es Salaam” Urban Forum, vol.
19, pp. 175-186, June 2008.
[48] P. S. Javid and T. S. Parikh, “Augmenting rural supply chains with
a location-enhanced mobile information system,” Proceedings of
the International Conference on Information and Communication
Technologies and Development, (ICTD 2006). Berkeley, CA pp.
110-119, 2006.
[49] A. Kumar, N. Rajput, S. Agarwal, D. Chakraborty, and A. A.
Nanavati, “Organizing the unorganized - employing IT to empower
the under-privileged,” Proceedings of the 17th international
conference on World Wide Web, Beijing, pp. 935-944, 2008.
[50] D. Chakraborty, K. Dasgupta, S. Mittal, A. Misra, A. Gupta, E.
Newmark, and C. L. Oberle, “Businessfinder: Harnessing presence
to enable live yellow pages for small, medium and micro mobile
businesses,” IEEE Communications Magazine, vol. 45, pp. 144-
151, 2007.
[51] S. Biswas and S. Roy, “A shared ICT infrastructure for Indian
SME clusters,” in 2nd IEEE/ACM International Conference on
Information and Communication Technologies and Development
(ICTD2007), Bangalore, pp. 221-230, 2007.
[52] Food India, “Inter-city marketing network for women micro-
entrepreneurs using cell phone,” Food India 2002.
[53] J. E. Katz and M. Aakhus, “Perpetual contact: Mobile
communication, private talk, public performance,” Cambridge,
UK: Cambridge University Press, 2002.
[54] M. Castells, J. L. Qiu, M. Fernández-Ardèvol, and A. Sey, Mobile
communication and society: A global perspective (information
revolution and global politics). Cambridge, MA: MIT Press, 2007.
[55] C. Licoppe, “'connected' presence: The emergence of a new
repertoire for managing social relationships in a changing
communication technoscape,” Environment and Planning D:
Society and Space, vol. 22, pp. 135-156, 2004.
[56] A. M. Townsend, “Life in the real-time city: Mobile telephones
and urban metabolism,” Journal of Urban Technology, vol. 7, pp.
85-104, 2000.
[57] J. Hamilton, “Are main lines and mobile phones substitutes or
complements? Evidence from Africa,” Telecommunications Policy,
vol. 27, pp. 109-133, 2003.
[58] J. Donner, “The social and economic implications of mobile
telephony in Rwanda: An ownership/access typology,” in Thumb
culture: The meaning of mobile phones for society, P. Glotz, S.
Bertschi, and C. Locke, Eds. Bielefeld, Germany: Transcript
Verlag, 2005, pp. 37-52.
[59] R. Harper, “Are mobiles good or bad for society?,” in Mobile
democracy: Essays on society, self and politics, K. Nyíri, Ed.
Budapest, Hungary: Passagen Verlag, pp. 185-214, 2003.
[60] J. Donner, “Shrinking fourth world? Mobiles, development, and
inclusion,” in Handbook of mobile communication studies, J. Katz,
Ed. Cambridge, MA: MIT Press,2008, pp. 29-42.
[61] M. Muto and T. Yamano, “The impact of mobile phone coverage
expansion on market participation: Panel data evidence from
Uganda,” Japan Bank for International Cooperation, Tokyo 2008.
[62] S. Gandhi, S. Mittal, and G. Tripathi, “The impact of mobiles on
agricultural productivity,” in India: The Impact of Mobile Phones,
The Vodafone Policy Paper Series #9 London: Vodafone, 2009,
pp. 21-33.
[63] L. Waverman, M. Meschi, and M. Fuss, “The impact of telecoms
on economic growth in developing nations." vol. 2007 London:
Moving the Debate Forward: The Vodafone Policy Paper Series
#3, 2005, pp. 10-23.
[64] R. Duncombe, “Using the livelihoods framework to analyze ICT
applications for poverty reduction through microenterprise,”

Information Technologies and International Development, vol. 3,
pp. 81-100, 2007.
[65] R. Veeraraghavan, N. Yasodhar, and K. Toyama, “Warana
unwired: Mobile phones replacing PCs in a rural sugarcane
cooperative,” Proceedings of the 2nd IEEE/ACM International
26
Conference on Information and Communication Technologies and
Development (ICTD2007), Bangalore, pp. 89-98, 2007.
[66] O. Morawczynski and G. Miscione, “Examining trust in mobile
banking transactions in Kenya: The case of m-PESA in Kenya,” in
IFIP WG 9.4-University of Pretoria Joint Workshop, Pretoria,
South Africa, 2008

i
The analysis includes two studies of fishermen but excludes
farmers
ii
The term ‘value system’ is sometimes used interchangeably
with ‘industry value chain’. This review uses the Porter
nomenclature to distinguish between the intra- and extra-
enterprise systems

An Evaluation of the use of ICT within Primary
Education in Malawi
David Hollow and Paola Masperi
Abstract— The paper demonstrates how appropriate technology,
when combined with quality curriculum-based content, has the
potential to have a positive impact on primary education within
developing countries. It documents an evaluation of portable
learning technology from the Ministry of Education, Science and
Technology in Malawi, providing theoretical background and
educational context before detailing the methodology and
findings of the study. Significant impacts from the introduction
of the program were increased school attendance, reduced drop-
out rates and improved student and teacher enthusiasm.
Alongside this, the audio and video material and interactive
learning techniques offered potential pedagogical benefits in
combining learner-centered and outcome-based activities with
continuous assessment, helping children to retain information
more effectively and record higher test scores. Major challenges
were also identified regarding implementation and sustainability.
These centered around the need for rigorous teacher training,
classroom integration, appropriate deployment, maintenance,
sustained impact and overall cost-effectiveness. In closing, the
paper emphasizes the need for such programs to be driven by
educational concerns and recognizes the similar challenges faced
in many related initiatives.
Index Terms— Education, Evaluation, Malawi, Technology
I. INTRODUCTION
I N setting the context for the subsequent evaluation the
paper begins with an overview of two distinct spheres,
firstly the current education context in Malawi and secondly
the potential role of technology within education. These two
seemingly disparate themes are brought together through the
case study of the Interactive Learning Program. An overview
of the program is provided, followed by an explanation of the
methodological approach employed in the monitoring and
evaluation exercise. The findings are then categorized into
four sections of analysis which inform the concluding
comments and recommendations.
The Interactive Learning Program in Malawi comes under
Manuscript received September 21, 2008.
David Hollow is a PhD candidate with the ICT4D Collective in the
Geography Department at Royal Holloway, University of London. e-mail:
d.m.hollow@rhul.ac.uk
Paola Masperi is an independent development consultant, based in London.
e-mail: paola@mayamiko.org
27
the authority of the Ministry of Education, Science and
Technology (MoEST). Funding for the initiative was provided
by a British company who also developed the educational
software for the portable devices.
II. CONTEXT
A. Education in Malawi
Free primary education was introduced in Malawi in 1994 as
a result of the commitment made by the country at the World
Conference on Education for All (EFA) in 1990. Widespread
agreement exists regarding the vital place of education within
poverty reduction efforts, capacity building and growth
strategies of developing countries. In Malawi, numerous
policy, budgetary and multilevel commitments have
contributed to significant progress in the delivery of primary
education since 1994. This includes enrolment figures rising
from 1.9 million to 3.2 million, the construction of 1,000 new
classrooms and roll out of the new curriculum to 5,500
schools [1]. However, Malawi is one of the poorest countries
in the world, with a GNI per capita of $230. Life expectancy
at birth is 48 years and 63% of the population live on less than
$2 per day [2]. Within such a context of extreme poverty, the
rapid advancements in access to free schooling have put the
national education system under considerable strain.
The increase in enrolment has led to significant pressure
upon primary schools in Malawi, most notably in regard to
increasing class sizes, a lack of fully qualified teachers,
limited teaching materials and inadequate infrastructure. It is
estimated that an extra 8,000 teachers are currently required to
meet the MoEST target pupil to teacher ratio of 1:60 [3]. In
addition, and despite national net enrolment levels of 91%,
national drop out rates remain high with only 44% of those
that enroll in Standard 1 completing Standard 5 and less than
30% reaching Standard 8, the final year of primary education
[4]. Linked to this and demonstrating the ongoing challenge in
regard to teaching capabilities and educational outcomes is a
declining performance in national examinations with a failure
rate of over 40% [1].
The importance of ensuring good quality education
alongside improved access [5] has been long recognized in
Malawi and in 1999 it was decided that a comprehensive

reform of the curriculum was required due to the
developments that had occurred since 1994 [3]. The relevance
of topics such as democracy, human rights, gender and
HIV/AIDS were increasingly acknowledged and these have
been systematically incorporated into the new primary school
curriculum in the form of life skills education.
B. Technology in Education
There is significant attention focused upon the potential for
ICT to assist in leapfrogging educational challenges within
developing nations [6] [7] [8], with much enthusiasm for a
possible technology-enabled ‘breakthrough in learning’ [9].
However, much debate surrounds the question of whether the
infusion of technology into education has actually instigated
more than incremental changes to the field. Proponents assert
that the last decade have resulted in the emergence of a new
landscape for education across the globe [10] with technology
positively effecting student motivation [11] and, when
initiatives are implemented with fidelity, leading to a
significant increase in learning [12]. This is challenged by
those who argue that achieving structural technological
change in schools takes much longer than anticipated [13]
with no substantial evidence that the introduction of
technology has yet caused any fundamental change in either a
developed or developing context [14]. Wagner [15] notes a
wide variety of outcomes from ICT for education projects in
developing countries, with significant negative impacts
including the reinforcing of dependencies, imposition without
community involvement and collapse due to lack of funding
or political commitment.
Within this contested environment, the range of technologies
theoretically available for deployment within developing
country educational contexts is rapidly expanding, with a
transition towards increasingly portable, powerful and
adaptable tools [16]. The new technologies available have
potential to mark a significant transition beyond
conceptualizing e-learning through conventional static
computer laboratories [17], emphasising instead the place of
anytime, anywhere computing [16] and mobile learning [18].
Such a shift creates opportunity for major changes in the
application of educational technology, with key potential
benefits noted as increased enthusiasm, cooperation,
communication and student ownership [19].
Despite the potential benefits of learning technologies [20]
it is important to recognize that not all solutions which have
contributed to educational advancements in the developed
world can simply be transferred to a developing world
context. In order for the potential of ICTs to be realized in
improving the quality of education it is important they are
applied with cultural understanding, local knowledge and
sensitivity [11] [21]. Indeed, ‘effective use of ICTs must be
tied to the needs of developing countries and challenge the
one size fits all approach of many programmes’ [19;7]. A
recognition of this fact emphasises the need for primary focus
on appropriate software development linked to curriculum-
28
based content and sustained classroom integration.
III. INTERACTIVE LEARNING PROGRAMME
Having outlined the overall context of education within
Malawi and the theoretical role of technology within
education, attention now turns to the Interactive Learning
Program. It considers whether the initiative constitutes a
suitable application of portable technology to support the
provision of basic education within the country. To begin, the
specifics of the program are explained, including the
background, technology and pedagogical rationale.
A. Background to Program
The Interactive Learning Program was introduced in 2006
and following initial positive feedback the MoEST requested,
in early 2007, that the initiative be scaled up to incorporate 50
primary schools. The schools that the MoEST selected for
participation ranged from Karonga District in the north
through to the southernmost districts of Mulanje and
Phalombe. A total of 520 custom-made handheld interactive
learning aids were distributed to the participating schools.
This totaled ten devices for each school, except for two test
schools which received 20 devices for the purposes of targeted
monitoring and evaluation.
B. Technology Utilized
The device in question is a handheld interactive learning aid
(from herein referred to as the learning machine or gadget as
named by the participating children), slightly larger than a
mobile phone and able to play video and audio through either
a loudspeaker or headphones. Positioned next to the screen are
a selection of buttons which can be pressed in response to
questions asked in the lesson being watched. The device has
an in-built rechargeable lithium battery with power for
between four and six hours of continuous play. A total of 25
lessons in Chichewa and 40 in English are preinstalled and
stored on two gigabytes of internal Flash memory, leaving
additional room for newly developed lessons to be
incorporated. The lessons are designed for use by children in
Standard 3, 4 and 5, each lasting approximately 20 minutes
and covering a range of curriculum based subjects such as
General Studies, Social Studies, Science, Mathematics,
Geography, Life Skills and English.
With less than 5% of primary schools across Malawi
estimated as having access to a reliable supply of mains
electricity [3] the initiative was dependent upon solar charging
systems. Each of the project schools were equipped with a 20-
Watt solar panel connected to a deep-cycle sealed battery,
generating enough energy to power 20 of the learning
machines. Teachers were given training regarding appropriate
charging and usage, and instructed to leave the devices to
recharge overnight once every two days.

C. Pedagogy
The digital content on the learning machines is designed in
order to enable and promote interactive and outcome-based
learning, linked to the Malawi national curriculum and
actively supporting the government Primary Curriculum
Assessment Reform [22]. The lessons are designed to be user
friendly, with the audio-visual content accessible to illiterate
or semi-literate users. Instructions from lip synchronized
cartoon characters explain the functions of the various buttons
on the learning machine. Once a lesson title has been selected
the methodology becomes interactive and learner-centered.
Having listened to brief teaching points the learners are tested
on their understanding of the information supplied through the
posing of multiple-choice questions answered by pressing
buttons on the device. At the end of the lesson learners are
given opportunity to test what they have learnt through
undertaking an overall quiz of ten questions where they are
congratulated according to the score attained. This outcome-
based approach provides incentive for the children to work
conscientiously and allows the teacher to monitor progress.
The decision to include lessons in both Chichewa and
English was taken on the basis that early learning content is
most effective when communicated in the vernacular language
[23] [24]. The dual language approach equips children for the
challenging transition from Chichewa to English as the
official medium of school instruction which takes place in
Standard 5. The most conducive learning environment for
using the devices is with groups of between four and six
students. In this context team work, group participation and
the development of leadership skills can each be encouraged.
IV. METHODOLOGICAL APPROACH
In regard to the overall context of technology within
education, Kozma [25;21] concludes after assessing a wide
variety of studies that there is ‘no consistent relationship
between the mere availability or use of ICT and student
learning’. Beyond such generic assertions, there are
significant knowledge gaps remaining regarding what kind of
initiative works and what does not. As InfoDev [11;5]
summarize, ‘despite thousands of impact studies, the impact of
ICT use on student achievement remains difficult to measure
and open to much reasonable debate’. In light of such
observations all partners in the Interactive Learning Program
initiative recognized the necessity for a comprehensive
monitoring and evaluation exercise, the key objectives of
which were to;
• assess the impact of the program on primary education
• assess the feasibility of the program in each test school
• identify program weaknesses for future refinement
• develop both teacher and organizational capacity
• provide feedback to partners regarding suitability of scale-
up and sustainability
A common theme from ICT-enhanced education programs
within developing countries is the marginalized place of
29
monitoring and evaluation. Where undertaken, there is often
strong emphasis placed on hardware-based input indicators
[25] rather than more complex learning outcomes. In order to
avoid this simplification and demonstrate an alterative
approach, a rigorous monitoring, evaluation and impact
assessment structure was built into the program throughout
implementation. This involved three field visits from the
monitoring and evaluation team in September 2007,
November 2007 and March 2008.
In addition to being marginalized, monitoring and evaluation
of ICT for education programs has rarely embraced creative
processes of assessment [26]. Recognizing this, a systems-
based approach was utilized in designing the methodology
[27] [28] [29]. This involved adopting a mixed- and multi-
method approach which focused on process and aimed to
engage with multiple stakeholders so as to gain a plurality of
perspectives regarding program impact. It also ensured that
data gathered was not solely anecdotal but credible,
dependable and confirmable [30]. Within this overall aim for
increased creativity and rigor, the research remained subject to
the standard complexities and constraints of limited time,
budget, data and personnel [31] [32].
It light of the constraints it was decided by the MoEST that
five representative test schools would serve as a sample of the
50 participating in the program. The schools selected by the
MoEST were Chin’gombe, Mwatibu, Mthentera, Mbinzi and
Dzenza. None of these schools had any form of ICT
incorporated into the curriculum prior to the introduction of
the program. A total of 15 days were spent in these schools
due to a recognition of the benefits of prioritizing classroom-
based research [33]. A combination of qualitative and
quantitative methods were utilized and these are now outlined
alongside the rationale for each.
A total of 15 lessons in which the learning machines were
being used were observed throughout the process [34].
Teachers were asked to conduct the lesson as they would
normally, without altering classroom arrangements. Observing
usage of the technology in a normal environment provided an
ideal foundation for the subsequent methods, giving
opportunity to discuss successes and suggest ways to work
around challenges with the children, teachers and head-
teachers.
Some 15 group interviews were held with children in the
five schools, in order to hear what difference the program had
made to their lives. The children were asked what they liked
and disliked and improvements they would make for the
future. The interviews followed a semi-structured, guided
approach [35] [36] facilitated by a teacher and conducted in
Chichewa. Participants were selected at random from
Standard 3 and 4 classes, were aged between 7 and 15, and
had equal gender representation.
Following these group interviews, more detailed
conversations were held with individual children in order to
learn about particular experiences or perspectives that they
had expressed. Such unstructured and story-based
conversation was useful in identifying unanticipated program

impacts [37]. In addition to this, a three stage individual
interview was conducted with the head-teacher of each school
and also with four related officials from the MoEST. These
interviews provided opportunity to obtain input regarding the
program feasibility, impact and future direction.
Conducting focus groups was considered to be the most
appropriate method to gain detailed feedback from teachers
regarding their view of the program and the difference it had
made to their lives, including positive and negative impacts
and potential improvements. A total of 15 focus groups were
conducted in this safe environment [38] which facilitated
robust dialogue [39] but ensured no one felt pressurized to
share personal experiences [40].
Building on these, the learning octagon was a research tool
developed specifically for the program in order to stimulate
more detailed discussion between teachers in the focus
groups. The octagon providing a pictorial representation to
assess the strengths and weaknesses of eight different
dimensions to the program by drawing on a combination of
the Octagon tool [41] and Most Significant Change approach
[42] [43]. This enabled the teachers to collaborate on creating
a visual representation of the impacts of the initiative.
In addition to these methods, and in order to enable on-
going monitoring on a weekly basis, one teacher from each of
the five schools was selected to complete a diary documenting
their experiences of using the learning machines in their
lessons. The diaries were semi-structured and designed to
provide a continuous record of particular strengths,
weaknesses and challenges encountered.
Baseline tests were conducted at stage one and three of the
research. The objective of the test was to provide a
quantitative assessment of the impact of the program on the
attainment of the children regarding both curriculum and life-
skills. In each of the five test schools there were 12 children
randomly selected from Standard 3 and 4, providing a total of
60 children, 30 boys and 30 girls. Five additional schools
were selected to act as a control group with an additional 60
children tested.
Alongside focusing on the test schools, evaluation
questionnaires and equipment feedback forms were distributed
to all 50 participating schools in order to gain a broader
understanding of program impact. The questionnaires gave
opportunity for feedback regarding patterns of usage and
challenges encountered in implementation. The equipment
feedback forms enabled the documentation of technical
problems encountered throughout the program.
Each of the methods outlined above enabled a detailed
perspective to be developed regarding the strengths and
weaknesses of the program. The monitoring and evaluation
methodology employed was participatory throughout [44] and
engaged with children, teachers, headmasters, community
leaders, government officials and civil society representatives.
Each school was visited at least three times throughout the six
month research period and this allowed for a progressive
approach and the development of good relationships with a
number of stakeholders.
30
V. ANALYSIS
The monitoring and evaluation exercise was focused on
assessing three interconnected spheres which cut across the
five previously identified research objectives. These were
impact upon students, impact upon teachers and effectiveness
of the technology. Analysis of each sphere is based on the data
gathered using the range of methodological approaches
previously outlined. The analysis is illustrated throughout with
quotations from children, teachers, head teachers and MoEST
officials.
A. Impact upon Students
In assessing the overall impact upon students the study
considered the specific impact on four main areas of student
attendance, enthusiasm, attainment in curriculum subjects and
attainment in life skills subjects.
A significant and universally agreed impact of the program
was the increase in school attendance. Teachers reported a
large increase in class sizes with fewer children absent than
prior to the intervention. One student from Standard 4
explained why this was the case:
Before the gadgets more pupils were absenting themselves
from classes but now we encourage our fellow pupils to come
to school and tell them, today if you absent yourself, you will
miss using the gadgets. I used to absent myself 50% of the
time before the gadgets came, now I come to school everyday.
This is a significant achievement within a national context of
high drop-out rates and absenteeism. For any educational
intervention to succeed it is necessary for learners to be
attending school on a regular basis and the motivating
influence of appropriate technology is clearly a contributing
factor in achieving this desired end. Despite strong anecdotal
evidence, assessing the statistical significance of the change in
attendance was hampered by a lack of daily attendance
records maintained in each school, meaning it was impossible
to track exact attendance patterns prior to the current term.
Children in the group interviews reported that they were
enthusiastic about coming to school now that they were using
the learning machines. Several students reported having
shared the experience with family and community members
who actively encouraged them not to miss out on the
opportunity to learn with the new technology. As one student
from Standard 3 explained:
They [parents] said to me that you have to work hard and
make sure you do not run away from lessons when you are
using the computer. I used to run away from class but now I
have changed my behavior … because I am attracted to the
gadgets and if I miss the chance to use it, it will never come
again.
Other children emphasized what they had learnt through the
lessons and how this related to what they hoped to do in the
future. One student from Standard 5 reported the impact of a
particular lesson regarding counting:

Now I know how to add and subtract – I could see the
pictures and I was able to subtract. This makes a foundation
for me to be able to work in a bank – this is what I want to do
in the future.
The learning machines are perceived by the children as a
mixture between mobile phones, games devices and video
players. This makes them attractive to the children, who
quickly become familiar with them and enjoy the process of
being congratulated by the device for answering questions
correctly. Observed learners appeared to enjoy working in
groups and taking turns in pressing the buttons. Teachers also
noted a similar impact and emphasized the improved listening
skills of the students:
They are able to explain things now – and the gadgets really
help them with listening skills – if they do not listen then they
miss what has been said and they cannot answer the question
– and it makes them be fully attentive.
Increased attentiveness in class and greater motivation to
attend school also had an effect on the attainment of the
children in both curriculum and life skills based lessons. Both
students and teachers reported that the use of audio and video,
as well as the continuous assessment quizzes, increased
retention and affected attainment. One teacher reflected on
tests completed the previous week:
We had the mid-term tests last week and more pupils did
well than before – when I asked them why they said it was
because of the gadgets – ‘we just remember what we have
learnt on them’. The increase in attainment has gone up by
about 30% on average from what they normally achieve.
Despite such assertions it was difficult to assess quantitative
impact on student attainment due to the short period of time
since the beginning of the intervention. A comparison of
scores between the baseline and second test demonstrated that
certain sectors of the curriculum had considerably more
correct answers after using the learning machines. However,
the majority of students had only completed 15% of available
lessons at the time of the second test and the overall
improvement in curriculum attainment was limited to 1.5%.
Similarly, improved attainment in life skills was difficult to
quantify as test scores related to knowledge gained rather than
necessarily to behavioral and lifestyle change. The overall
impact on life skills attainment in the baseline test was a 3%
improvement. Again, the questions which were answered
correctly indicated an increased understanding of the content
for those lessons which the teachers had chosen to use most
regularly.
Although quantitative change proved difficult to assess,
many students interviewed were able to recall accurate
information on the life skills lessons they had used and
expressed satisfaction to have learnt skills useful to their daily
lives. Learners explained how they had put into practice what
had been taught in the lessons, indicating potential behavioral
change:
[my favorite lesson was…] Preventing malaria – because it
affects many children in the village. I learnt that we must
sleep under treated bed nets and must not play with stagnant
31
water. I did not know this before the gadgets.
Children were also able to talk about culturally sensitive
subjects such as HIV/AIDS and explain what they had learnt
from the lessons. Teachers stressed the value of the interactive
lessons in a cultural context where such discussions are often
considered taboo. The children requested that new content be
added to the devices so that they could learn about a greater
variety of topics. The most commonly requested subjects were
Agriculture, Mathematics, English, Religious Education,
Science and Technology, Physical Education and Music.
Similarly, the teachers requested that new content would
remain linked with the curriculum reform in order to provide
them with support in this transition.
The impact of the initiative on the students is intrinsically
linked to the way in which the teachers are affected, and this
is now considered.
B. Impact upon Teachers
In assessing the overall impact upon teachers the study
considered two factors, teacher enthusiasm and teacher
workload.
The majority of teachers were enthusiastic about the
introduction of the program into their school and several
commented that they felt honored to have been selected as a
school for the pilot. They were pleased that MoEST officials
were visiting their school and showing an interest in their
work. They were keen to learn about technology and could see
the positive effects on themselves and the learners. They also
noted how the introduction of the learning machines had
helped them in altering their teaching style, adopting new and
innovative approaches:
We love it – before the gadgets it was just talk and write for
some subjects – but now the kids can see the pictures. Some of
the children can be sitting doing one thing while others are
doing the gadgets.
Young trainee teachers particularly enjoyed using the
technology and often engaged in using it without
apprehension. This confidence had a clear effect on the
manner in which the children approached the devices and
made use of them in the classroom. The enthusiasm of the
students also affected the teachers:
…because if the learners are enthusiastic then it makes us
enthusiastic. If the gadgets help the students to learn then the
teachers have to be happy that the students are doing well.
However, this enthusiasm was not felt universally. Mature
teachers often showed less interest in the program, were
fearful of using the technology and reluctant to let the children
use it independently. In one of the test schools after six
months of usage there were only three teachers still using the
learning machines. All the other teachers in the school had
become uninterested in the program, considering the
additional workload to be too much of a burden.
The teachers reported a variety of different experiences
regarding the impact of the program on their workload. This
was dependent upon the manner of implementation in each

school and the degree to which the learning machines were
incorporated within the lesson schedule. The overall feeling
from the teachers was that the introduction of the gadgets had
increased the workload in certain areas and decreased it in
others:
There are two things, lesson preparation - for this it might
take more effort - and lesson delivery - for this it might take
less time. It has made us do more work – we have to guide the
students. And there are not enough gadgets – so if we had
enough gadgets then it would be ok and then it would not
increase our workload.
Since the intervention was a pilot project the majority of
schools chose to use the gadgets after the school day had
finished so as not to interfere with the daily timetable. This
meant that the teachers were required to stay in school for an
extra hour every day to facilitate the lessons. Most were
willing to do this but some expressed reluctance due to pre-
existing after-school commitments. It was also commented in
two schools that the initiative resulted in more preparation
work for the teachers because they needed to become fully
familiar with the devices themselves before making use of
them in lessons with the children. In the two schools where
the teachers had chosen to incorporate the gadgets into the
school day then a decrease in workload was reported. In these
cases the gadgets helped with effective facilitation, the
management of large classes and provided support in
preparing teaching and learning materials for the lessons.
In light of this, the program would benefit from the
development and distribution of a pedagogical guide for
teachers regarding the effective integration of the learning
machines into the learning routine of the children. Teachers
should also be encouraged to utilize the full range of lessons
that are available on the devices. This would complement the
development of a schedule for teachers to use the machines
during school time, giving clear explanation how it can most
appropriately support the curriculum.
The increased ability of the children to memorize what had
been taught was also recognized to make the job easier by one
teacher from Standard 4:
Gadgets have simplified the work for the teachers – when
the child sees something he remembers – when he just hears
he forgets. When they see it, it stays in their mind.
Others noted that the impact of the learning machines in
encouraging so many children to return to school had caused
class sizes to expand significantly, resulting in more work
attempting to manage the children. A common concluding
comment from teachers was that the workload would decrease
significantly if there was a lower pupil to gadget ratio and if
the lessons on the gadgets were fully compatible with the
syllabus and incorporated into the curriculum. The program
would therefore benefit from an increased allocation of 20
learning machines per school. This would facilitate a more
concentrated impact and allow a greater number of children in
smaller groups to benefit from the content on the devices.
In light of the challenges presented, it is clear that regular
training sessions for the teachers would assist in maintaining
32
the momentum of the program and improving teacher
motivation. The limited budget of the MoEST makes it a
constant challenge to deliver adequate training, especially in
remote areas. The teachers requested opportunity to ask
questions and provide feedback, requiring effective
communication channels between the stakeholders. As
highlighted by one official:
If you come back frequently then the schools will know you
are coming and they will keep using the gadgets but if they
think no one is coming to visit them then they will forget to use
them. This is because when they are monitored they will feel
that somebody is appreciating their job – they are happy that
they can be involved and they see that somebody cares.
The feedback from teachers demonstrated that sustained
monitoring from MoEST officials constituted a significant
capacity building process. The impact was especially
pronounced in marginalized rural areas which were
inaccessible and rarely visited. In light of this is it clear that a
continuing focus on teacher training which both develops
skills and instills value is intrinsic to the success of this and
any ICT for education initiative. The ability of the teachers to
utilize technology with confidence and adapt it to the specific
needs of the classroom is a central determining factor.
Without this integration any initiative remains an appealing
add-on but does not have significant effect on the culture of
the classroom, pedagogy, or rationale for learning.
C. Effectiveness of the Technology
The success of the initiative, whilst not determined by the
technology, was dependent upon it operating effectively
within each school. Usability, durability and the reliability of
the devices and charging solution were the three areas
assessed in the study.
The lesson observations demonstrated that the majority of
children were able to operate the learning machines
independently, understanding the purpose of the buttons and
completing quizzes without supervision. The process of lesson
selection occasionally required guidance from teachers but
this was rarely necessary because the most capable children
were seen to take a lead in the learning groups, demonstrating
considerable confidence with the devices. However, a
significant factor limiting usability was the low number of
devices per classroom and subsequent high number of
students per group, on occasion as many as ten learners. This
was not conducive to learning as the machine is designed for
small groups, with small screen and quiet audio output. In
such situations, dominant children were observed bringing the
machine to their ear in order to hear the lesson, meaning other
children were unable to engage and became easily distracted.
This was confirmed by a participating official from the
MoEST who noted:
Bringing such technology is very important - it is only
that at the moment the gadgets are very few – there needs to
be more. It is important that you have the right number of
gadgets for the number of children because it is difficult for so

many children to share.
Due to the nature of such a pilot program it was anticipated
that there would be significant problems regarding device
durability. However, after six months of usage only 8% had
developed problems, the majority being due to software issues
easily resolved through reformatting. Two of the devices
stopped working due to the on/off switch breaking, these were
repaired in situ and left fully functioning. A major concern in
this respect was that teachers often assumed they had broken
the devices when all that was actually required was for them
to be charged fully.
The solar panels operated effectively in each of the five test
schools throughout the program with all of them maintaining
adequate power to charge the devices as often as necessary.
Several program schools outside the five included in the test
did report problems including two faulty batteries, one loose
connection and one stolen panel. In addition to these, one
school adopted the practice of disconnecting the battery from
the panel to remove it for safe keeping during the night and
this prevented the battery from charging fully.
VI. CONCLUSIONS
The analysis of impact on students, teachers and
technological effectiveness has demonstrated the educational
potential of the Interactive Learning Program. The
introduction of the devices had significant positive impact
upon school attendance and levels of enthusiasm. Also
noteworthy was the increased value placed on teachers and the
role of education within the community. However, alongside
this, considerable challenges were encountered including the
negative impact on teacher workload, lack of classroom
integration, significant infrastructural constraints and the
necessity for further teacher training. The future success,
sustainability and potential for project scalability is dependent
upon engaging with each of these identified challenges. The
observations made in the analysis speak directly to the
program efficacy but are also applicable lessons for ICT for
education initiatives more widely. The introduction of
technology into education systems in the developing world is
a complex procedure with considerable potential for failure
and therefore it is vital that rigorous monitoring and
evaluation is incorporated throughout all such initiatives.
The price of portable educational technology is constantly
decreasing, making initiatives such as the Malawi Interactive
Learning Program increasingly cost effective. The September
2008 price of each learning machine was circa US$55 and as
each device can be used by several children the per child cost
is considerably lower than other widely publicized solutions.
Significant future potential also lies in the opportunity for
integrating the educational content with new generation
mobile phones and other handheld devices becoming
progressively more available across the region. Linked to this,
increased ability to send and receive data in standard formats
(for example Macromedia Flash, xml, mp3 or mp4) will allow
33
for the development of user interaction. Market influences on
the development of mobile technology also ensure sustained
improvements in hardware capabilities and the subsequent
emergence of new educational possibilities.
However, within all such initiatives, it is necessary that the
application of technology is conceptualized as a tool in
facilitating the overall aim of catalyzing a more fully effective
approach to education. In reaching this goal many of the
challenges documented remain the same regardless of the
presence of technology.
It is not overly surprising that the introduction of ICT into
primary schools in Malawi caused a dramatic increase in
attendance figures. The significant question for ongoing
research remains one of long term impact on approach to
schooling, once initial enthusiasm surrounding the program
has subsided. Whilst a significant achievement, increasing
school attendance figures is not the most significant long-term
challenge facing education in Africa [5]. Instead, it should
constitute an initial building block which can be capitalized on
through adopting an integrated approach to developing the
capacity of the education system. One dimension to this will
be the effective utilization of curriculum enhancing
technology, such as the Interactive Learning Program. For full
benefits to be realized, such programs require transition from
their position of peripheral curiosity to one of being integrated
as a sustainable teaching tool. Future research is required to
assess the degree to which this is occurring in Malawi.
A variety of different solutions are required to address the
challenge of providing equitable access to good quality
education in the developing world. The increasing availability
and decreasing cost of portable devices ensures that they are
likely to play a prominent role in the future, however this
should not lead to such technologies being presented as an
educational panacea. For programs to be of maximum
educational benefit in Malawi, as in many such initiatives, the
critical issues for consideration and action remain pedagogy,
classroom integration and teacher training.
REFERENCES
[1] MoEST, Link for Education Governance. Education Sector Performance
Analysis. Lilongwe, Malawi. 2007.
[2] World Bank, Data and statistics: Malawi. 2006.
www.web.worldbank.org Accessed 21/08/08
[3] MoEST, Primary Curriculum and Assessment Reform (PCAR).
Implementation Plan 2007-2010, Lilongwe, Malawi. 2008.
[4] UNESCO, Institute for Statistics, country data 2008.
www.uis.unesco.org Accessed 21/08/08
[5] EFA, Global Monitoring Report: The quality imperative. 2005.
www.portal.unesco.org/education Accessed 21/08/08
[6] D. Pye and J. Stephenson, Using ICT to Increase the Effectiveness of
Community-based, Non-formal Education for Rural People in Sub-
Saharan Africa: the CERP Project Final Report (DFID Educational
Paper 50). London: Department for International Development 2003.
[7] V. Tinio, ICT in Education. E-Primers for UNDP, New York 2003.
[8] J. Leach, DEEP IMPACT: an investigation of the use of information and
communication technologies for teacher education in the global south.
Researching the Issues, 58. DFID 2005.
[9] D. Wagner, Monitoring and evaluation of ICT for Education: An
Introduction. Chapter 1 in: D. Wagner et. al. 2005. Monitoring and
 34
Evaluation of ICT in Education Projects – A Handbook for Developing [35] J. Burgess, The Art of Interviewing. In: A. Roger and H. A. Viles, (eds)
Countries, InfoDev 1-19, 2005. The Student Companion to Geography. Blackwell: Oxford, 242-249,
[10] D. Keats, The genesis and emergence of Education 3.0 in higher 2003.
education: the potential for Africa. First Monday 12, 3, 2007. [36] M. Q. Patton, Qualitative Research and Evaluation Methods. California:
[11] InfoDev, Knowledge Maps: ICTs in Education. World Bank 2005. Sage, 2002.
[12] C. Fadel and C. Lemke, Technology in Schools: What the research says. [37] R. Chambers, Participatory Workshops – a sourcebook of 21 sets of
Metiri Group, Commissioned by Cisco Systems 2006. ideas and activities. United Kingdom: Earthscan, 2002.
[13] L. Cuban, Oversold and Underused – Computers in the Classroom. [38] G. Valentine, Being seen and heard? The ethical complexities of
Harvard: USA 2001. working with children and young people at home and at school. Ethics,
[14] T. James and J. Miller, Developing a Monitoring and Evaluation Plan for Place and Environment. 2, 141-155, 1999.
ICT in Education. Chapter 4 in: D. Wagner et. al. 2005. Monitoring and [39] D. Stewart and P. Shamdasani, Focus Groups: Theory and Practice.
Evaluation of ICT in Education Projects – A Handbook for Developing Newbury Park: Sage, 1990.
Countries, InfoDev. 57-77 2005. [40] P, Bhattacharjee, Stepping Stones – A participatory tool to integrate
[15] D. Wagner, B. Day, J. S. Sun, Recommendations for a pro-poor ICT4D gender into HIV/AIDS work. Development in Practice, 10, 5, 691-694,
non-formal education policy. Final Report for, Imfundo: Partnership for 2000.
IT in Education, DFID, 2004. [41] Sida. The Octagon. A tool for the assessment of strengths and
[16] M. Sharples, Disruptive Devices: personal technologies and education. weaknesses in NGOs, 2002.
Educational Technology Research Paper Series 11, The University of [42] P. Sigsgaard, Monitoring without indicators: an ongoing testing of the
Birmingham, 2000. MSC approach. Evaluation Journal of Australasia. 2, 1, 2002.
[17] H. U. Hoppe, R. Joiner, M. Milrad and M. Sharples, Wireless and [43] J. Dart & R. J. Davies, A dialogical story-based evaluation tool: the most
Mobile Technologies in Education. Journal of Computer Assisted significant change technique, American Journal of Evaluation. 23, 2,
Learning, Guest editorial, 19, 3, 255-259, 2003. 137-155, 2003.
[18] F. Lehner and H. Nosekabel, The Role of Mobile Devices In E-Learning [44] M. Huberman, The Many Modes of Participatory Evaluation. Chapter 7
-First Experiences With A Wireless E-Learning Environment. WMTE in J.B. Cousins and L. M. Earl (eds) 1995. Participatory Evaluation in
103-106, 2002. Education. Studies in Evaluation use and organisational learning.
[19] J. Leach, T. Power, R. Thomas, X. Fadani and A. Mbebe, 4D London: Routledge, 103-114, 1995.
Technologies: appropriating handheld computers to serve the needs of
teachers and learners in rural African settings. Centre for Research and
Development in Teacher Education, Open University: UK, 2003.
[20] C. Pontefract, Learning to Share, Insights Education 1 Missing the
Connection – using ICTs in Education, February 2003
[21] R. Dhanarajan, Learning Technologies, where is the challenge?
Education, Communication, Information 1, 1 Spring 2002
[22] MoEST, Primary Curriculum and Assessment Reform (PCAR).
Literature Review of Primary Education in Malawi, Lilongwe, Malawi,
2002.
[23] G. Bunyi, Rethinking the place of African indigenous languages in
African education. International Journal of Educational Development 19,
4-5, 337-350, 1999.
[24] EFA, Global monitoring report: Literacy for Life, 2006.
www.portal.unesco.org/education Accessed 21/08/08
[25] R. Kozma and D. Wagner, Core Indicators for Monitoring and
Evaluation Studies for ICT in Education. Chapter 3 in: D. Wagner et al.
2005. Monitoring and Evaluation of ICT in Education Projects – A
Handbook for Developing Countries, InfoDev. 35-57, 2005.
[26] J. Taylor and S. Soal, Measurement in Development Practice: From the
Mundane to the Transformational. Chapter 5 in L. Earle (ed) 2004.
Creativity and Constraint: Grassroots Monitoring and Evaluation and the
International Aid Arena. INTRAC Policy Series 18, 95-110, 2004.
[27] P. Morgan, Measuring the Development of Capacity: Is this Still a Good
Idea? Chapter 2 in L. Earle (ed) 2004. ‘Creativity and Constraint:
Grassroots Monitoring and Evaluation and the International Aid Arena’.
INTRAC Policy Series 18, 47-62, 2004.
[28] D. Watson, Monitoring and evaluation of capacity and capacity
development. ECDPM, Discussion paper 58B. April 2006
[29] D. Cabrera, A Theory of Systems Evaluation. Systems evaluation and
evaluation systems whitepaper series. New York: Ithaca, Cornell
University, 2006.
[30] J. Baxter and J. Eyles, Evaluating qualitative research in social
geography: establishing ‘rigour’ in interview analysis. Transactions of
the Institute of British Geographers, 22, 4, 505-525, 1997.
[31] M. Bamberger, J. Rugh and L. Mabry, Real World Evaluation. Working
Under Budget, Time, Data and Political Constraints. London: Sage,
2006.
[32] D. Gray, Doing Research in the Real World. California: Sage 2004.
[33] M. C. O’Sullivan, What is happening in the classroom? A common-
sense approach to improving the quality of primary education in
developing countries. Teacher Development, 9, 3, 301-314, 2005.
[34] J. W. Creswell, Educational Research: planning, conducting, and
evaluating quantitative and qualitative research. New Jersey: Pearson
Education, 2005.

Claim Mobile: Engaging Conflicting Stakeholder
Requirements in Healthcare in Uganda
Melissa R. Ho, Emmanuel K. Owusu, and Paul M. Aoki
Abstract—Claim Mobile is a platform designed to support a
project that subsidizes healthcare by reimbursing health service
providers in Uganda for treatment of patients with sexually
transmitted infections. As with many development projects, the
Uganda Output-Based Aid (OBA) project involves a number of
stakeholders: the service providers, the project implementers,
the financiers, and the Ugandan government. Design of an
appropriate solution requires meeting the various and conflicting
requirements of all of these stakeholders. In this paper we detail
the rapid design and testing of a pilot implementation of a
mobile and web-based system for processing claims forms, based
on two prior field visits to Uganda. Based on a comparative
device study, semi-structured interviews, health clinic surveys,
and a brief deployment, we affirm the selection of the mobile
phone as a platform from the health clinic perspective, and
further suggest that effective design for development requires
more than addressing requirements of the the “users” of the
mobile phones but also all the other stakeholders involved, who
may have conflicting requirements.
Index Terms—mobile phone, ICTD, health, participatory de-
sign, Africa, HCI
I. I NTRODUCTION
Mobile phones are frequently touted as being the appro-
priate and sustainable platform for rural healthcare in Africa.
They are relatively cheap, durable, consume less power than
laptops and desktops, and incorporate a battery that makes
them more amenable to use in places with intermittent or no
power. Commonly proposed uses are for data collection [1],
[2] and decision support for rural health workers [3], [4]. Some
projects also use mobile devices as a platform for information
dissemation as well as data gathering [5]. However, these are
all generally “closed loop” systems in which researchers are
able to control all aspects of the system design and operation,
focusing their research primarily on the rural health workers
that will be using the mobile phones.
Other applications have even more potential for large-scale
impact. In the agricultural sector, we have observed how
the introduction of transparent market prices and subsequent
hiring of “middlemen” to purchase from farmers has reduced
Manuscript received October 1, 2009. This work was supported by the
Blum Center for Developing Economies and the U.S. National Science
Foundation Summer Undergraduate Program in Engineering Research at
Berkeley (SUPERB) under Grant No. 0453604.
Melissa R. Ho is with the School of Information at the University
of California, Berkeley, 94720, USA. (phone: +256 777 723 786; email:
mho@ischool.berkeley.edu).
Emmanuel K. Owusu is with the Computer Engineering Department at
Iowa State University, 50011, USA. (email: kwame@iastate.edu).
Paul M. Aoki is with Intel Research, Berkeley, CA, 94704, USA (email:
aoki@acm.org)
35
fraud and transformed supply-chain management for the E-
choupal project [6]. While health information is critical to the
improvement of healthcare in developing regions, financing
healthcare also remains a significant unsolved problem. Can
we take lessons from e-Choupal and apply them in the
healthcare sector? The design of usable, reliable, and fraud-
resistant tools for management of these aid flows is an area
with potential for very significant impact.
However, in the case of healthcare, the financial models are
very different from commercial markets – financing of health-
care typically comes through transnational aid agencies like
the World Bank and International Monetary Fund (IMF), and is
implemented by non-governmental organizations (NGOs) and
the local government. Since the NGOs are typically experts in
health, not technology, data processing is often outsourced to
third-party information technology (IT) vendors. Relationships
between the vendors, the NGOs, the local governments, and
the transnational aid agencies are not always smooth - and
limitations in communications infrastructure means that the
information flows between them are scattered at best.
In this paper we suggest that the “closed loop model”
generally used by researchers in deployments of mobile health
applications does not map onto the financial and political
realities of the mainstream of healthcare provision in Africa,
and limits the ability of pilot programs to increase their scale
and impact. We describe an innovative, IT-based, NGO-run
healthcare access program in Uganda, and our experiences
designing and deploying Claim Mobile, a mobile-phone based
system intended to address inefficiencies and help the program
scale to additional districts. We argue that in addition to
addressing the needs of the primary users in the system, the
health workers, our design must consider the requirements,
motivations and concerns of the other stakeholders: the IT
vendors, the NGOs, the government, and the aid agencies.
Our designs must consider the larger order ramifications of
how we may positively and negatively impact both the “users”
who will be generating the data, and the entities that will be
engaged in managing and using the information in the resulting
database. Just as the e-Choupal project assimilated the mid-
dlemen by hiring them as kiosk operators, we propose that we
can design applications structured to accommodate conflicting
stakeholder requirements, while also alleviating information
inequalities resulting from limitations in the system prior to
the introduction of the information technology.

Fig. 1. A HealthyLife voucher. The ‘M’ in the top left is a note written
by the health service provider to indicate that the first client voucher on the
left was given to a male client, and that therefore the partner vouchers on the
right should be given to a female client.
II. BACKGROUND
Providing effective health care in poor countries is an
essential component to economic development and poverty
reduction. Unfortunately donors supporting this endeavor often
find that resources given are not matched by desired gains.
The output-based aid (OBA) model of financing seeks to ad-
dress this by paying healthcare providers directly for services
rendered instead of paying for the service provision up front.
However, OBA program management is information intensive,
necessitating much paperwork to track and reimburse payment
claims. Smartphones (mobile phones with advanced features
such as the ability to run third-party software) have the
potential to alleviate this burden. In collaboration with a local
NGO and their partnering IT vendor, we have proposed to
deploy a number of smartphones for use in an OBA project
based in Western Uganda, with dual goals of reducing claim
processing time and improving communication between the
health care providers and the OBA management agency.
The project is managed by the local branch of a multi-
national NGO and a for-profit health insurance company, in
collaboration with the Ugandan Ministry of Health (MoH)
and Ministry of Finance (MoF). The project is primarily
funded by an aid agency based in Europe, with additional
funding for the expansion coming from a separate transna-
tional funding agency. Together, they run a voucher program
called HealthyLife, which treats sexually transmitted infec-
tions (STIs), reimbursing providers for the diagnosis and full
course of treatment only after the patient is seen. This program
was implemented in response to the high burden of sexually
transmitted infections in Uganda, and began in July 2006
in four districts of southwestern Uganda: Mbarara, Ibanda,
Kirihura and Isingiro (See Table I).
Patients buy treatment vouchers in pairs, one for the client
and a second one for the client’s sexual partner (See Figure 1).
Each voucher is good for one consultation (generally including
a lab test to diagnose the STI) and three follow-up visits.
During the consultation, the provider completes a claim form
recording the client’s demographics, the examination and
laboratory results, a diagnosis and details of the course of
treatment prescribed (See Figure 5). Completed claims forms
are sent to the voucher management office in the city of
Mbarara, the main urban center of Western Uganda. Forms
36
Mbarara, Uganda
• HIV prevalence: 10% of adult population (15-49
years)
• Syphilis prevalence: about 5-7% of adult population
• 1 in 4 households had at least one phone.
• 39% reported STI symptoms
• only 1/3 sought care
• 54% of respondents who sought any STI treatment
reported using private clinics.
TABLE I
S OME BACKGROUND STI STATISTICS ABOUT M BARARA , U GANDA [7].
can take two weeks or more to move from the providers office
to the management agency. The current data management
system requires all claims to be submitted on paper forms
to the management agency. At least another two to four
weeks are spent reviewing each claim, cleaning data from
improperly-completed forms, and verifying that the service
took place among suspect claims. Two months or more can
go by before the provider is reimbursed for service provision.
In Uganda, private providers traditionally operate on a fee-
for-service model, receive prompt payment, and do not have
a large operating margin. In many cases, payment is provided
prior to service. Delays in payment result in delays in pro-
curement of replacement prescriptions and medical supplies,
often leading to a temporary hiatus in service. Encouraging
provider involvement in the OBA program requires a great
deal of confidence on the part of the providers to participate. If
a system to shorten claims processing could be devised, more
providers could join the scheme and more patients could be
provided the life-saving STI treatment voucher subsidy.
The remainder of this paper details the system we are cur-
rently piloting, in which claims are submitted via Internet from
a mobile phone directly to all the parties in the management
agency. In addition to describing our user studies and how this
has informed the design of the system, we discuss the problem
of negotiating conflicting stakeholder requirements. We find
that in projects with multiple stakeholders, the introduction of
a system may disrupt balances of power, particularly around
the flow of information and money. As a result, the design of
this system, in order to secure positive support from all parties
involved, must carefully balance stakeholder incentives.
III. M ETHODS
The research described here involved an iterative process of
field research and prototyping. The fieldwork and deployments
have been done over the course of three visits to Uganda:
an initial two-week visit in Summer 2007 to establish a
relationship with the project, in which we also conducted a
survey of the clinics in the program; a followup visit for
three weeks in November 2007; and a five-week pre-pilot
deployment in August-September 2008. During all three visits
we conducted semi-structured interviews with the various
stakeholders, and directly observed claims form entry and
processing. When given permission, we did audio and video
recording of interviews and user study activities. In all, we

have approximately 30 hours of audio, and have done detailed
interviews in seven of the 12 participating clinics (in addition
to the initial survey of all of the clinics), as well as intensive
observation in two clinics, a rural, high-claim-volume clinic
with very little exposure to computers, and an urban low-
claim-volume clinic with its own computers. The last visit
entailed a comparative user study as well as deployment of
the mobile phones in the latter two clinics.
A. Clinic Surveys
The clinic survey was conducted in conjunction with a
larger survey of available infrastructure at contracted clinics.
We asked 14 questions, assessing familiarity with computers
and mobile phones, but primarily gathering feedback from the
health clinics on the claims process (e.g., how long it takes
them to fill out the paper forms, and what their priorities
might be for improvement of the process). We also collected
various documents from the management agencies regarding
the performance of each clinic, including all available financial
reports on processed claims, and in November, we returned to
seven of the clinics to do in-depth surveys and to follow-up
on the survey findings.
B. Rapid Development and Pre-pilot Deployment
Initial prototyping occurred in early 2008, and we returned
to Uganda in Summer 2008 to do a three-stage pilot deploy-
ment, first testing the functionality of our software, second
reviewing the proposed claims process with the management
agencies, and finally taking the phones to the health clinics to
test the mobile phone interface in the field. During this time
we also conducted another round of semi-structured interviews
to gather information on changes in the claims submission
process (for example, claims processing had moved from
Mbarara to the national capital, Kampala). We did iterative
development based on feedback from the various stakeholders,
trying out features as they were suggested, and developing
new tools as seemed merited by findings in our interviews.
To gain a more in-depth understanding of health clinic life,
we stayed overnight for three days in the rural health clinic,
thereby supplementing the the semi-structured interviews with
direct observation of actual practice.
The primary purpose of this last field visit was to con-
duct a pre-pilot demonstration, using the mobile phones to
submit actual claim data to the management agency, have it
reviewed, and have the management agency provide feedback
to the health clinics via the mobile phones. We simulated
the proposed process, physically following the claims forms
from the time the patient comes into the health clinic, through
the preparation of the claims summary forms, physically
transporting the forms to the management agency where we
observed the claims approval, and data entry into the existing
database. We simultaneously had the service providers submit
the claims form via Claim Mobile, enabling the management
agency to provide feedback to the service providers through
the system. The pre-pilot is still operational, with mobile
phones remaining in the two clinics, and the full pilot will
37
Fig. 2. This diagram illustrates the flow of money and information between
selected stakeholders in the OBA project, both currently, and how it will be
once the Claim Mobile system is fully deployed.
be conducted in May 2009, with mobile phone-based claims
processing expanded to 8-10 additional clinics.
IV. S TAKEHOLDERS
In this section we describe the funding, management and
service provider organizations to which we alluded in Sec-
tion II. Fundamentally, all stakeholders want to improve STI
treatment and reduce the prevalence of STIs. Each stakeholder
also has a financial interest in the success of the overall project
- staying afloat for the health service providers, and staying
within the aid agencies’ target budgets for the management
agency partners. The discussion highlights the ways in which
the various stakeholders have competing as well as common
interests. While we will detail several stakeholders in this
section, there are six key stakeholders: the aid agencies who
fund the OBA project, the financial management agency
(FMA) which receives the funds from the government and
disburses them, their program management office (PMO) in
Mbarara which runs the program and interacts directly with
the health service providers, the technical management agency
(TMA) that manages the claims processing, the health service
providers (HSPs), and us, the mobile platform developers
(MPDs). Figure 2 illustrates some of the relationships between
these entities which we will describe in detail in the remainder
of this section, based on qualitative fieldwork and document
analysis.
A. Aid Agencies
As the funder of the HealthyLife STI treatment program,
the involvement of the European aid agency is more than
apparent. Their role in the management of the program is
more supervisory – a consultant goes to Uganda at irregular
intervals to help with planning of the program, and they do
some monitoring. They also have commissioned another non-
profit, affiliated with a North American university, to conduct
an evaluation of the program. Ultimately, however, they control
the flow of money to the financial management agency, which
then pays the IT vendor to handle the technical aspects of the
operation.
In the past year, the European aid agency has worked with
an additional transnational aid agency to fund the expansion

of the project into additional districts. While they may not
have a direct impact on the information processes in the
project itself, the funders’ internal actions have direct impact
on the project as a whole. In one example, a delay in
payment to the European aid agency resulted in a delay in
payment to the two management agencies. As a result the IT
38
for the PMO, and one for the TMA. However, the copy that
remains in the PMO does not have the voucher number, a
critical piece of information, and with stacks of hundreds of
claims per month, the information is not in a format actually
accessible to the program office until the TMA sends back