You are on page 1of 513

3rd International Conference on

Information and Communication Technologies and Development

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

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

Tirthankar Dasgupta and Anupam Basu

Analyzing Statistical Relationships between Global Indicators through 212

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

Edwin Nyella

Cross Technology Comparison for Information Services in Rural 252

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
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

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

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
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

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

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

John Traxler
University of Wolverhampton, UK

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

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

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

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

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

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
Victoria Stodden Patrick Meier
Berkman Center for Internet and Society Berkman Center for Internet and Society
Harvard Law School Harvard Law School
Cambridge, MA 02138 Cambridge, MA 02138

Abstract—Is the dramatic increase in Internet use associated democracy and regime transitions literature. To be sure, “the
with a commensurate rise in democracy? Few previous studies trouble with the zealots of technology as an instrument of
have drawn on multiple perception-based measures of governance democratic liberation is not that they misconceive technology
to assess the Internet’s effects on the process of democratization.
This paper uses perception-based time series data on “Voice & but that they fail to understand democracy” [2]. ‘In other
Accountability,” “Political Stability,” and “Rule of Law” to pro- words, “it turns out there is no simple general answer to
vide insights into democratic tendency. The results of regression the question: Is the technology democratizing?’ until we have
analysis suggest that the level of “Voice & Accountability” in a made clear what sort of democracy we intend.” We address this
country increases with Internet use, while the level of “Political question first before proceeding with a more detailed literature
Stability” decreases with increasing Internet use. Additionally,
Internet use was found to increase significantly for countries review.
with increasing levels of “Voice & Accountability.” In contrast, Barber’s notion of “strong democracy” comprises the careful
“Rule of Law” was not significantly affected by a country’s level and prudent judgment of citizens who participate in deliber-
of Internet use. Increasing cell phone use did not seem to affect ative, self-governing communities. Schmitter and Karl write
either “Voice & Accountability,” “Political Stability” or “Rule that, “modern political democracy is a system of governance
of Law.” In turn, cell phone use was not affected by any of
these three measures of democratic tendency. When limiting our in which rulers are held accountable for their actions in the
analysis to autocratic regimes, we noted a significant negative public realm by citizens, acting indirectly through the compe-
effect of Internet and cell phone use on “Political Stability” and tition and cooperation of their elected representatives”[3]. The
found that the “Rule of Law” and “Political Stability” metrics two authors emphasize that citizens are the most distinctive
drove ICT adoption. element in democracies. “All regimes have rulers and a public
Index terms—cell phone, democracy, fixed effects model, ICT,
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
I. I NTRODUCTION is usually regarded as an essential for good governance
Does the globalization of the Internet have a democratizing and effective democratic consolidation” [4]. In other words,
effect? The question has already been posed by numerous regular elections are not sufficient. As Zakaria noted, illiberal
studies but these have largely taken the form of qualitative case democracies have free elections but citizens remain cut off
studies and/or large theoretical analyses. In terms of a rigorous, from real power due to the lack of civil liberties [5].
quantitative establishment of the democratization effects of the Huber et al. write that the most basic feature of democracy
Internet, however, the jury is still out [1]. At the heart of is power sharing [6]. They identify three clusters of power
this debate, moreover, lies a more fundamental question about as primarily relevant for the chances of democracy: (1) the
the essence of democracy. In fact, “unless we are clear about balance of power in civil society; (2) the balance of power
what democracy means to us, and what kind of democracy we between state and society; and (3) the transnational balance of
envision, technology is as likely to stunt as to enhance the civic power that shape the first two and constrain political decision-
polity” [2]. The purpose of this paper is to contribute more making. By remaining diverse and independent of the state,
rigorous data-driven analysis to the literature on Internet and political participation by civil society acts as a channel of
democracy since “there is no doubt that rigorous and data- public voice and accountability, and a way of challenging
driven analysis of this relationship will benefit scholars and and checking the unbridled power of authoritarian regimes
policymakers alike” [1]. [4]. The structure of state-society relations is equally relevant
Previous research on the topic of Internet and democracy for democracy. As Huber et al. note, “the power of the state
can be characterized as lacking a serious perusal of the needs to be counterbalanced by the organizational strength of

the civil society to make democracy possible; the state must when Internet access and cell phone use is significantly more
not be so strong and autonomous from all social forces as prevalent and globalized than in the 1990s. If a statistically
to overpower civil society and rule without accountability.” significant relationship between Internet and democracy does
Clearly then a governing body that fails to follow the “rule exist, then it is more likely to manifest itself now and not in the
of law,” should not be considered democratic [3]. These 1990s. Third, we draw on both Internet and cell phone data per
elements of democracy are not sufficient conditions for a stable 100 inhabitants per 181 countries to assess the impact of the
democracy, but they are necessary and indispensable to the information revolution on democratization. We use regression
persistence of democratic governance. analysis to determine whether Internet or cell phone use has
The italicized terms above represent the fundamentals be- had a correlative effect on measures of democratic tendency.
hind the sort of democracy we intend: active citizen par- We also model whether the collection of democratic measures
ticipation, good governance, accountability, power sharing, has had a correlative effect on Internet or cell phone use.
balance of power and rule of law. The few quantitative studies The paper is structured as follows: the first section reviews
that do exist on Internet and democracy tend to aggregate the current debate and literature on Internet and democracy.
these fundamentals of democracy into a single index. Doing The second section explicates the datasets used in this study
so means these lose important information on how these while the third section formalizes the statistical models em-
individual components of democracy may be affected by the ployed in the regression analysis. Section four reviews the
growing prevalence of global Internet access. Furthermore, results and provides an interpretation of the findings. The fifth
past quantitative and qualitative studies tend to focus primarily and final section concludes the study.
on the impact of the Internet on established democracies. They
also focus on the 1990s almost exclusively, a serious limitation II. L ITERATURE R EVIEW
that remains surprisingly understated in the literature. Equally The Internet and democracy literature comprises two distinct
problematic in the current literature is the interchangeable schools of thought each comprising a host of qualitative
use of the terms “Internet” and “information revolution.” The research and some quantitative inquiry. In this section we
terms are purposefully not differentiated on the basis that the review in some detail the qualitative and quantitative literatures
predominant feature of the information society is the spread that have contributed to both schools of thought over the past
of the Internet. While this is true of Western democracies, it ten to fifteen years. In so doing, we compare and contrast the
is certainly not true for the majority of developing, nondemo- main arguments along with the respective findings. As noted
cratic countries, where cell phones are the most widely spread in the introduction, one common shortcoming of the Internet
communication technology after radios [7]. Indeed, the irony is and democracy literature is the tendency to oversimplify our
that “those who might most benefit from the net’s democratic understanding of democracy. The purpose of this literature
and informational potential are least likely to have access to review is thus to redress this gap evident in previous studies.
it, the tools to gain access, or the educational background to The first school of thought is often referred to as the
take advantage of the tools” [2]. more populist school of thought. This strand of the literature
This paper seeks to redress each of these shortcomings. subscribes to the viewpoint that the Internet has democratic
First, since the boundaries of the term “democracy,” and how relevance and impact [8], [9], [10], [11]. According to these
it is measured, is subject to lively debate, we use multiple authors, the Internet will decentralize access to communication
perception-based measures of governance for our dependent and information while increasing citizen access [12]. Best
variables. Governance indicators provide a better set of proxies and Wade write that “the Internet’s collective characteristics
for the sort of democracy we intend as identified above. We (e.g., low cost, multidirectional capability, etc.), helps make
therefore draw on the following three World Bank indicators: this possible.” We first review the qualitative literature that
(1) Voice and Accountability (VA) measuring perceptions comprises this school of thought followed by quantitative
of the extent to which a country’s citizens are able to par- studies.
ticipate in selecting their government, as well as freedom Dahl previously observed that telecommunications tech-
of expression, freedom of association, and a free media; (2) nologies have a key role in making possible the advanced
Rule of Law (RL) measuring perceptions of the extent to democratic country, where policy is firmly anchored in the
which agents have confidence in and abide by the rules of judgment of the “demos” [13]. In his list of the procedural
society, and in particular the quality of contract enforcement, minimal conditions that must be present for modern political
property rights, the police, and the courts, as well as the democracy to exist, Dahl thus argues that citizens should
likelihood of crime and violence; and (3). Political Stability have the right to seek out alternative sources of information.
and Absence of Violence (PS) measuring perceptions of Rheingold has dubbed the Internet as “the great equalizer”
the likelihood that the government will be destabilized or because it can “equalize the balance of power between citizens
overthrown by unconstitutional or violent means, including and power barons” [14]. The idea here “is that the Internet
politically motivated violence and terrorism. These metrics are will serve as a mass audience, and will politicize them in the
drawn from the World Bank Governance Indicators Research process” [12]. Anderson et al. draw on sociological research
Database (see to show that electronic networks lead to the “break-down of
Second, we draw on data from 2000 through 2006, a time status-based social structures” and “increased participation in

discussion, decision-making, and task processes by those who for example). At the international level, the authors make the
typically are political or economically disadvantaged” [15]. bold claim that the Internet has “contributed to the rise of a
Etzioni sees in the Internet the possibility of an advancement more multicentric world structure in which nation-states have
of the state of public affairs through “teledemocracy” [16]. seen their preeminence lessen and non-governmental actors
Other scholars claim that the Internet will “enable a Jeffer- take the stage” [1]. The salient point here is that groups and
sonian revolution” [17]. Grossman argues that a “third epoch individuals can far more efficiently form coalitions of conse-
of democracy is arriving by the hand of technology,” and quence with a range of powerful collectives. As Best and Wade
claims that a “new, hybrid electronic republic’ is now forming rightly note, there are obvious democratic elements to this,
to displace the creation of Montesquieu, Locke, Madison and including the need for “nation-states to provide democratic
their contemporaries” [18], [12]. The practice of “electronic rights to their citizens so as to build legitimacy on the global
democracy,” according to Browning, will differ substantially stage” [1].
from previous renderings of democracy [19]. Perhaps the most We now turn to the quantitative studies that comprise the
provocative claim associated with this more populist school of first school of thought. One of the earlier statistical studies
thought is the one made by Negroponte, who argues that the on this side of the literature was carried out by Kedzie,
nation-state will evaporate as a consequence of the information who provides an account of how information communication
revolution [20]. Snider suggests that citizens need only be technologies contributed to the “third wave” of democracy
potentially informed in order to hold government accountable [26]. Prior to the fall of the Soviet Union and the prolifer-
[21]. ation of new democracies in Eurasia, the mainstream theory
Barber opines that by favoring decentralization, the mul- of democratization held that democracy followed economic
tiplication of choice, and consumer sovereignty, new tech- growth and development [27]. To be sure, one of the few robust
nologies such as the internet or cell phone have already, findings in the literature is that democracy is more likely in
albeit inadvertently, benefited democratic political culture [2]. more developed countries [28], [29], [30]. Longdregan and
According to Barber, “democracy is a form of government Poole have also shown that the most significant predictor of
that depends on information and communication. It is obvious transitions to authoritarianism is poverty [31], [32]. “In short,
then that new technologies of information and communications after 20 years of observation and analysis during the third wave
can be nurturing to democracy. They can challenge passivity, of academic interest in democratization, we can be reasonably
they can enhance information equality, they can overcome certain that a positive relationship between development and
sectarianism and prejudice, and they can facilitate participation democracy exists, though we do not know why” [33].
in deliberative political processes.” Hill and Hughes argue that Kedzie, however, was more interested in testing another
those who subscribe to the populist school have reasons to be potential causal mechanism, the “dictator’s dilemma” hypoth-
optimistic: “If the mere fact that political discourse against esis, which suggests that the globalization of markets places
repressive governments is taking place is a good in itself, then pressure on authoritarian regimes to keep their countries’
the utopians have reason to celebrate [22]. Perhaps the Internet communication borders open. He reasoned that the ensuing
will bring about a wider democratic revolution in the world” massive flow of information would not only allow for “the
[22]. efficient passage of commercial information, but also for more
Bimber is more cautious, arguing that while the Internet is ’democratic’ information” [1]. As Bimber observes, the most
accelerating the process of issue group formation and action important predictions about the Internet’s impact on politics
(in America), the structure of political power has not been amount to “causal claims regarding the effect of information
revolutionized or qualitatively transformed into a new epoch of flow on political participation and the organization of interests”
democracy [12]. According to Bimber, other scholars believe [12]. Other scholars have made related arguments. Webster,
that the Internet has a “transformative potential” because it for example, writes that the Internet has helped to facilitate
facilitates a kind of “one-to-one interaction among citizens a new form of capitalism called “information capitalism” in
and between citizens and government.” Along these lines, which global labor markets require highly flexible workers
some scholars such as Corrado and Firestone write that the who continuously adapt and learn [34]. Regimes that impose
Internet has the potential to promote “unmediated” commu- restrictions on information capitalism forgo the financial re-
nication and thereby decrease citizens’ reliance on officials turns possible by tapping into the information economy [35],
and organizations [23]. In sum, what distinguishes the populist [36], [37].
enthusiasm for the Internet is the “idea that elites and political In his study, Kedzie employs regression analysis to compare
intermediaries will grow less important” [12]. how much of the variation in democracy is explained by both
In the more contemporary, qualitative literature, Steele and traditional predictors of democracy and the strength of Internet
Stein, argue that the Internet amplifies trends in international diffusion by drawing on data from 144 countries [26]. For
relations [24]. Rosenau and Johnson address the impact of the his set of control variables, Kedzie included more traditional
Internet at both the individual and international level [25]. At predictors of democracy including economic development,
the individual level, the authors argue that the Internet can be education, human development and health. He also included
used as a tool to organize collectively to effect social and polit- indicators of pre-Internet information communication tech-
ical change around the world (see, nologies (ICTs). His results suggest that the Internet is indeed

a strong predictor of democracy, more so than traditional variation in the Freedom House data, which makes meaningful
determinants of democracy. In a follow up study, Richards statistical analysis more difficult.
assessed the relationship between the Internet and physical In contrast to the populist literature, the second school
integrity [38]. His findings support Kedzie’s. However, the of thought disputes the majority of claims that exist vis-à-vis
latter study faces an important limitation since Kedzie’s (rather the relationship between Internet and democracy. The counter-
simple) longitudinal analysis draws on data from 1993. At arguments are based on both qualitative and quantitative re-
this point during the early 1990s, the Internet was hardly search. In terms of qualitative research, several scholars argue
globalized. that the Internet is merely an extension of the ruling class and
Best and Wade recognize this important short coming in centralized control [39], [40], [41], [34], [42]. According to
Kedzie’s study and therefore explore the global effect of the Neuman, even if the increase of ICTs had led to an increase
Internet on democracy over a ten year period, 1992−2002 [1]. in the motivation to communicate - which he argues has not
They aggregate political and civil rights data from Freedom happened - then ICTs would have become centralized by
House to formulate a democracy index, which serves as their government turning them into social control mechanisms [43].
dependent variable. The number of Internet users per 1, 000 Scholars who subscribe to this school of thought maintain that
represents their independent variable while the following mea- mass media information technologies discourage collective
sures are used as control variables: GDP per capita, education behavior, ”unless the rise in couch potatoes can be considered
and literacy rates, life expectancy, urbanization, prevalence of a social movement” [44], [45], [46].
non-Internet ICTs. Their analysis shows that a statistically In contrast to Snider’s argument about the mere potential of
powerful correlation exists between Internet diffusion and citizens being informed acting as a source of accountability, if
level of democratization. “The more salient observation to power is measured by the potential for “monopoly and control
make, however, is that while economic prevalence and liter- over information and communication, it is evident that the new
acy maintain relatively constant correlations with democracy, technology can become a dangerous facilitator of tyranny”
the correlation for Internet prevalence gradually strengthens, [21], [2]. Indeed, while the Internet may enable citizens to
almost to the same level as economic prevalence” [1]. subvert political hierarchy, Barber notes that with increased
The authors suggest that this dynamic reflects the grow- participation comes the peril of political and economic surveil-
ing significant relationship between Internet prevalence and lance. The populist school of thought is often blind to “how
democracy: “perhaps this is an indication that the Internet easily liberating technologies become tools of repression” [2].
has come of age as a correlate of democracy” [1]. Indeed, Bimber rejects the supposition that the Internet will
they posit that this growth in correlation strength might “be have significant effects on public life, point out that “both
expected given the positive network externalities, the network theory and empirical evidence cast grave doubt on the
effect’ that is a salient property of the Internet” [1]. However, communication-action connection at the core of the populist
the coefficients from the regression analysis reveal that Internet theory” [12]. Lippmann argued that the capacity of ICTs to
usage is only able to predict a minimal amount of the variation recreate politics is constrained by human nature, ie. cognitive
in democracy: “to generate one point of democracy, an extra processing, and not by the technical properties of the media
500 Internet users per 1,000 citizens is needed, or an extra themselves [47]. The Internet, then, is no different than other
$5, 882 of GDP per capita is needed” [1]. The scale of democ- ICTs even if the new medium differs from previous technolo-
racy runs from 2 to 14. In terms of democracy’s traditional gies in a fundamental way, namely allowing social bonding
determinants, GDP was a weak predictor while literacy turned to occur asynchronously. In sum, the Internet is “hardly
out to have no significance whatsoever. The other control producing the first dramatic expansion in communication:
variables used were either insignificant or internally correlated. telephone, radio, and television also expanded communication
While Best and Wade’s important contribution to the litera- profoundly.” There seems no compelling reason to believe
ture on Internet and democracy is one of the few contemporary that the communication capacity of the Net will have such a
quantitative studies carried out thus far, their approach does dramatically different effect than have other advances in point-
face a number of important limitations [1]. For one, their to-point and broadcast communication” [12]. Moreover, Page
democracy index needs to be unpacked and “its constituent argues, new ICTs may very well overcome spatial distance
components, such as freedom of the press, openness of the but his far from sufficient for establishing vibrant forms of
electoral process,” for example, tested against traditional de- political communication and deliberation [48].
terminants of democracy to determine whether one compo- Furthermore, “if democracy is to be understood as delib-
nent provides more explanatory power than others. Another erative and participatory activity on the part of responsible
limitation of their data is the fact that it extends only to citizens, it will have to resist the innovative forms of dem-
2002. This should be updated today due to the rapid pace agoguery that accompany innovative technology and that are
of ICT diffusion over the past several years. In addition, too often overlooked by enthusiasts [2]. Aristotle wrote that
several scholars have criticized the Freedom House data with the basis of a democratic state is liberty. Barber adds that a
regards to conceptualization, measurement and aggregation “free society is free only to the degree that its citizens are
issues (Munck and Verkuilen, 2002; Rydland et al., 2008). informed and that communication among them is open and
Furthermore, as discussed subsequently, there is little to no informed [2]. However, recent research and empirical work

confirms that governments increasingly have the upper hand is global in scope and uses time series data from 1992 to
in controlling and regulating the impact of the information 2002. We seek to build on their work by using data from
revolution [49], [50], [51], [52], [53], [54], [55], [56], [57], 2000 to 2006 and using the World Bank Governance metrics
[58]. As Goldsmith observes, “if governments can raise the as our measures of democratic tendency. Best and Wade
cost of Net transactions, they can regulate the transactions” combined the Freedom House metrics of political rights and
[59]. Beilock and Dimitrova found that countries with lower civil liberties as their measure of democratic tendency. As
Freedom House scores for civil liberties had significantly described subsequently we feel the Freedom House data are
lower Internet usage (even when controlling for economic not well suited to a regression study such as this one.
development) [60].
De Mesquita and George Downs also argue that government A. Our Approach to Measuring Democratic Tendency Using
elites (e.g., in Singapore) have learned to “stifle the bottom-up the World Bank Governance Indicators
democratic potential of the Internet and still promote economic Dahl characterizes a government with power vested in a
growth, contrary to Kedzie’s dictator’s dilemma argument” plurality as follows [13]:
[61]. As Bimber notes, the “central theoretical problem for
1) Control over governmental decisions about policy is
the populist claim is the absence of a clear link between
constitutionally vested in elected officials.
increases in information and increases in popular political
2) Elected officials are chosen and peacefully removed
action” [12] To this end, McLuhan’s old dictum may be wrong:
in relatively frequent, fair and free elections in which
“the medium is not the whole message. Content matters, and
coercion is quite limited.
there is simply no overwhelming reason to believe that a
3) Practically all adults have the right to vote in these
new medium will necessarily enhance the political quality
of communicative content” [12]. In short, “technology need
4) Most adults also have the right to run for the public
not inevitably corrupt democracy, but its potential for benign
offices for which candidates run in these elections.
dominion cannot be ignored” [2].
5) Citizens have an effectively enforced right to freedom
In terms of quantitative analysis, Scheufele and Nisbet’s
of expression, particularly political expression, including
2002 statistical study suggests that the Internet does not
criticism of the officials, the conduct of the government,
increase democracy. “Through linear regression, they find
the prevailing political, economic, and social system, and
that mass media broadcasting (e.g., television, newspapers)
the dominant ideology.
plays a far more effective role than the Internet in promoting
6) They also have access to alternative sources of informa-
democratic citizenship” [1], [62]. Given that an established
tion that are not monopolized by the government or any
body of quantitative research on this topic has yet to material-
other single group.
ize, Scheufele and Nisbet do caution against generalizing the
7) Finally, they have an effectively enforced right to form
results of their study, which focused exclusively on the United
and join autonomous associations, including political
States. Other scholars interested in this line of research have
associations, such as political parties and interest groups,
questioned the supposed direction of causation drummed up
that attempt to influence the government by competing
by the populist school of thought. Using multiple measures
in elections and by other peaceful means.
of regime type, Milner’s statistical analysis demonstrates that,
ceteris paribus, democracies permit much greater online ac- The first four points largely describe procedural aspects of
cess, both in terms of Internet users per capita and Internet a democracy, whereas the last three points delineate the com-
hosts per capita [63]. To this end, the information revolution munication aspect necessary for a well-functioning democratic
may merely be reinforcing pre-existing dynamics. regime. In fact, Diamond goes further and notes that “[s]ome
Milner’s study uses data from 1991 − 2001 to measure the insist on a fairly robust (though still procedural) definition
influence of regime type of adoption of the internet. This study of democracy, like Dahl’s ‘polyarchy.’ By this conception,
attempts to address a slightly different question—whether democracy requires not only free, fair, and competitive elec-
there is a relationship between measures of democracy and tions, but also the freedoms that make them truly meaningful
ICT penetration—but we build on her work by extending the (such as freedom of organization and freedom of expression),
range of years to 2006 (although we begin measurement in alternative sources of information, and institutions to ensure
2000). We adopt a fixed effect model and control for the time that government policies depend on the votes and preferences
component directly in the model. of citizens” [64]. Expanding on Dahl, Diamond has developed
Beilock and Dimitrova develop a model to explain global his own list of characteristics of a democracy [65]:
country differences in Internet use using income, measures of 1) Control of the state and its key decisions and allocations
freedom, region dummies, and development indices [60]. Their lies, in fact as well as in constitutional theory, with
data is a cross section from 2001 and does not take democratic elected officials (and not democratically unaccountable
variables directly into account. actors or foreign powers); in particular, the military is
Best and Wade ask the question closest to that addressed subordinate to the authority of elected civilian officials.
in this paper [1]. They ask whether Internet penetration has 2) Executive power is constrained, constitutionally and in
an effect on the level of democracy in a country. Their study fact, by the autonomous power of other government in-

stitutions (such as an independent judiciary, parliament, respondents, as well as thousands of experts working for the
and other mechanisms of horizontal accountability). private sector, NGOs, and public sector agencies” [66].
3) Not only are electoral outcomes uncertain, with a sig- In points eight through ten, Diamond gives a description
nificant opposition vote and the presumption of party of the role of law in a democracy. The World Bank has a
alternation in government, but no group that adheres to governance metric that expresses some of this: “measuring
constitutional principles is denied the right to form a perceptions of the extent to which agents have confidence in
party and contest elections (even if electoral thresholds and abide by the rules of society, and in particular the quality
and other rules exclude small parties from winning of contract enforcement, property rights, the police, and the
representation in parliament). courts, as well as the likelihood of crime and violence” [66].
4) Cultural, ethnic, religious, and other minority groups This approximates the World Banks “Rule of Law” metric and
(as well as historically disadvantaged majorities) are not we propose it as an empirical measure of Diamond’s points
prohibited (legally or in practice) from expressing their eight through ten.
interests in the political process or from speaking their The World Bank also has a metric measuring political sta-
language or practicing their culture. bility: “perceptions of the likelihood that the government will
5) Beyond parties and elections, citizens have multiple, be destabilized or overthrown by unconstitutional or violent
ongoing channels for expression and representation of means, including politically-motivated violence and terrorism”
their interests and values, including diverse, independent [66]. While not enumerated in either Diamond’s or Dahl’s
associations and movements, which they have the free- lists, Amartya Sen postulated an empirical correlation between
dom to form and join. democratic regimes and political stability [67]. He notes both
6) There are alternative sources of information (including the “political incentives provided by democratic governance”
independent media) to which citizens have (politically) to prevent crises and specifically that the “positive role of
unfettered access. political and civil right applies to the prevention of economic
7) Individuals also have substantial freedom of belief, opin- and social disasters in general” [67]. Thus we investigate the
ion, discussion, speech, publication, assembly, demon- World Banks “Political Stability” metric as another measure
stration, and petition. of democratic tendency.
8) Citizens are politically equal under the law (even though In measuring the relationship between ICT penetration and
they are invariably unequal in their political resources). these democratic variables, it is clear that country wealth is a
9) Individual and group liberties are effectively protected confounding factor that sound be taken into account: wealthier
by an independent, nondiscriminatory judiciary, whose countries are both more likely to be democratic and to be
decisions are enforced and respected by other centers of the heavier users of both the Internet and the cell phone. We
power. gathered gross domestic product (GDP) data from 2000 to
10) The rule of law protects citizens from unjustified deten- 2006 from the International Monetary Fund. The GDP data is
tion, exile, terror, torture, and undue interference in their purchasing parity adjusted to be directly comparable between
personal lives not only by the state but also by organized countries.
nonstate or anti-state forces. Diamond notes that country size is highly correlated with
regime type: “countries with populations under one million are
Like Dahl, Diamond includes procedural aspects of a demo- much more likely to be both democracies and liberal democ-
cratic regime (points one through three) and he enshrines what racies. Two-thirds of these countries are liberal democracies,
he considers essential communication requirements in points while only 30 percent of countries with populations over one
four through seven. These latter points can be characterized as million are. Among the larger 150 countries, only half are
approximating a “freedom of expression” or “political voice” democracies, while 70 percent of the small countries are. The
aspect to democracy. We found the World Bank governance countries with populations over one million are about twice as
metric of “Political Voice and Accountability” to represent likely as small states to have an electoral authoritarian regime
Diamond’s notion well in that it measures “perceptions of the and half again as likely to have a closed authoritarian regime.”
extent to which a country’s citizen’s are able to participate in [64], [65]. Because of this, we included population in our
selecting their government, as well as freedom of expression, models to control for country size. The population data for
freedom of association, and a free media” [66]. Like all the 2000 to 2006 was also gathered from the IMF.1
World Bank Governance metrics, it was built from surveys
and other sources of data within each country. These sources B. Limitations of The Data
give an idea of “freedom of belief, opinion, discussion, speech, The ICT data is gathered from the International Telecom-
publication, assembly, demonstration, and petition” present in munications Union (ITU). The ITU requested the number of
the country, although they do not measure the proliferation Internet and cell phone users from each country. This raises
of channels of communication that Diamond enunciates. The a host of questions about the reliability of the data since it
World Bank governance indicators metrics are based on 35
data sources some of which yield “subjective or perceptions- 1 Both the IMF GDP and population data are available at http://www.imf.
based data” including that from “household and firm survey org/external/pubs/ft/weo/2008/01/weodata/

is self-reported by the country. Perhaps the country has an activity for 40 of the countries most actively engaged in
incentive to under or over-report? It is plausible that some repressing internet activity [68], [69]. Although not as extreme
countries have more reliable data collection mechanisms in as the case of the Internet, cell phone use can be restricted
place than others. With one report per country per year, the by the government as well, and will differ from country to
data are highly granular, but they do seem to follow steady country. Zuckerman gives several examples of government
trends, and steadily upwards in ICT adoption. crackdowns on mobile phone use: Belarus’s reported shutdown
The World Bank did not calculate governance metrics for of their SMS network in March 2006, reports of Ethiopian
2001. We carried out simple linear interpolation to provide a cell phone blocking during the 2005 election protests, and
data point for each country in this year. We made the decision Cambodian blocking of SMS for two days before their 2007
to do this since there are only six time points from which elections [7].
to interpolate, and only one, the year 2000, falls before our North Korea and Cuba were dropped from the study since
missing data. This creates data that are “too smooth” for year official data is not reported for these countries. It is generally
2001 since they have been created from our pre-existing data known that there is very little internet access in Cuba, and little
and this will make our results appear more precise than they to none in North Korea and these are both regimes with little
in fact are. We argue that this effects is minor since the World democracy. If we had been able to include these countries in
Bank data are themselves aggregated from a large number of our estimation of the models, this would likely have bolstered
sources, and thus less subject to noise than using a single our results.2 Details of the data cleaning and amalgamation
source would be. process are on the study website at
Other metrics of the level of democratic rights exist, such as ICTD.
the Freedom House “Freedom in the World” metric. Freedom Our population data was obtained from the International
House carries out an annual global survey of political rights Monetary Fund and contains a sparse number os missing
and civil liberties. We choose not to use this as a measurement values. The IMF has made estimates of their missing data
of democratic tendency for two reasons. We felt that the World to complete the dataset.3
Bank Governance Indicators could be well grounded in the
theory of democracy as measures of democracy. Secondly, the III. E MPIRICAL M ODELING
Freedom House measures have some quantitative limitations. Our data comprises a panel containing N different times
A certain amount of inertia is built into the measurements so series each consisting of T observations. The number of
that it is difficult for a country to move much from one year to countries, N , is 181, and T , the number of years in our study,
the next. For both political rights and civil liberties a country is 7. A fixed effects model of our democratic measures’ effect
is scaled from 0 to 7, giving only 8 possible outcomes for on ICT penetration follows:
a country. Combining these two factors leads to a database
that does not shift very much from year to year. In the years ICTit = β0 + β1 RLit + β2 V Ait + β3 P Sit +
of our study, 2000 to 2006, of the 193 countries surveyed by
Freedom House (after subtracting the 9 with missing values β4 P OPit + β5 GDPit + β6 M Fit +
for both political rights and civil liberties for the entire time γt T D + ξi CD + it
series), 105 had no change in their scores for political rights
and 86 had no changes in their civil rights scores. The average i = 1, . . . , N, t = 1, . . . , T.
variance of those that did exhibit some change from 2000 to
2006 was 0.48 for political rights and 0.32 for civil liberties.
In this paper we model the penetration of Internet or cell
This means the majorities of countries, if they changed at all,
phone usage per 100 inhabitants, ICT , as a function of the
changed by perhaps one point on the 8 point scale.
World Bank democratic measures (Rule of Law, Voice and
Having more years of data, including 2007, would improve
Accountability, and Political Stability)4 , country size, country
our modeling. At the time of this writing, the World Bank
wealth, and the male/female ratio in the country.
Governance metrics were not available for 2007.
ICT is one of “Internet Use” or “Cell Phone Use.”5 GDPit
The IMF estimated some of the population number for some
is the per capita Gross Domestic Product for country i at
of the countries. It is likely this has the effect of providing
year t, adjusted for purchasing power parity.6 P OPit is the
population data that is smoother than it would otherwise be.
Note also that both ICT measures, the Internet and Cell Phone 2 Countries with missing values also tended to support our hypothesis:

use, are measures per 100 inhabitants. We emphasize that this Afghanistan, Bhutan, Comoros, Kiribati, Serbia, St. Kitts and Nevis, Timor-
must be carefully noted in interpretation of the regression Leste, and Tonga. Cuba, Iraq, Montenegro, and North Korea simply did not
furnish enough data for inclusion in the study.
results, since we use population as an independent variable. 3 A precise explanation of their data interpolation procedure was not readily
We also note that even though the ITU collects Internet use available. See
4 Available at
statistics for each country, what it really means to use the
5 The data used in this study is available at
Internet can vary by country due to filtering, censoring, and
other restrictions on access. The OpenNet Intiative at the 6 The data are available at
Berkman Center for Internet and Society monitors the filtering

population of country i at year t and M Fit is the male/female the same test was performed on the World Bank variables, 54,
gender ratio.7 62, and 60 of 181 rejected nonstationarity for Rule of Law,
β0 is the intercept term for country i, and γt T D and ξi CD Voice and Accountability, and Political Stability, respectively
are a time effect and a country effect, respectively. The effect (about one third of the data). Although there some evidence
of time is controlled for by dummy variables: T Dt is 1 for year of stationarity, it appears that the majority of the time series
t and 0 otherwise. Similarly CDi is a dummy variable that is included in this study are nonstationary, and it’s potential
1 for country i and 0 otherwise, controlling for the differences biasing of coefficient estimates is a concern.10 The typical
between countries. Finally, it is a disturbance term with remedy is the difference the data to remove the nonstationarity.
distribution N (0, σi2 ), which we assume to be uncorrelated Running the regressions in the above equation allowed us to
across country cross sections. In this study there are N = 181 carry out tests on the residuals directly to evaluate the level of
countries and T = 7 years. autocorrelation. There are two regressions to be run in this
study, modeling Internet penetration and cell phone usage.
A. Autocorrelation in Panel Data
Typically the Durbin-Watson test with one lag is used to test
A panel regression model of this type is subject to possible for autocorrelation in the structure of the regression residuals
autocorrelation between subsequent observations because of and the regression with Internet use as a dependent variable
the time series components. In a regression model as described was found to have autocorrelation present, and the cell phone
above, it is possible to ’discover’ what Granger and Newbold penetration regression was not, with Durbin-Watson values of
[70] termed “spurious” relationships between the variables. 1.86 and 2.01 respectively.11
That is, tests of significance on estimated coefficients may The Durbin-Watson statistic is created by calculating d =
indicate a significant result, when in fact none is present. (t −t−1 )2
t 2 , where t is the tth residual from the regres-
Granger and Newbold suggest economic time series data may 1 t
sion. It follows that 0 < d < 4. A value of 2 indicates
be especially prone to autocorrelation since they tend to be
no autocorrelation. To test whether the Durbin-Watson test
non-stationary, in that it is not uncommon for the process
statistics could be considered equivalent to 2, the test in the
generating the data to depend on the time it was sampled.
R statistical software package was used [71].
For example, economic time series are commonly subject
The adjusted R2 values were 0.9246 and 0.9199, seeming to
to seasonal or cyclical effects. Time series data that are
fit the Internet regression squarely into Granger and Newbold’s
not stationary will violate the assumptions of least squares
area of caution: a low Durbin-Watson statistic and a high
regression since the variance of the error term will depend
R2 value, and also casts some suspicion on the cell phone
on time and thus introduce a bias into coefficient estimation.8
regression. Granger and Newbold offer that until “a really
Granger and Newbold describe a high R2 value and a low
satisfactory procedure is available, we recommend taking first
Durbin-Watson statistic as warning signs that the estimated
differences of all variables that appear to be highly autocorre-
model may be yielding spurious results.
lated.” (p. 8.) We carried out this operation on both the Internet
Thus it is important to determine whether the time series
and cell phone use regressions because of the high R2 values
data in this study are nonstationary. Our data are sampled
and the nonstationary data in both regressions, even though the
yearly suggesting they may avoid annual cyclical effects.
Durbin-Watson statistic did not suggest autocorrelation among
We analyze 181 countries from 2000 to 2006 (with 2001
the cell phone regression errors. The plots of the residuals
interpolated for the World Bank variables) and thus have 7
for both regressions indicate possible heteroskedasticity. This
values in each time series. As Granger and Newbold mention,
suggests running the following differenced model:
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 ∆ICTit = β00 + β10 ∆RLit + β20 ∆V Ait + β30 ∆P Sit +
assess nonstationarity in each of our time series. Since Internet
and Cell phone use are increasing rapidly for the vast majority β40 ∆P OPit + β50 ∆GDPit + β60 ∆M Fit +
of the countries in our study, we measured the autoregressive
structure of each time series as stationary around a trend line, γt0 T D + ξi0 CD + it
and modeled with one lag. The null hypothesis is that the
data are nonstationary. The test was applied to the ICT and i = 1, . . . , N, t = 2, . . . , T.
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, Running this model for differenced Internet penetration and
meaning that for around 20% of the Internet and cell phone differenced mobile phone use did not improve the Durbin-
penetration time series, nonstationarity is not evident.9 . When
10 Note that a combination of nonstationary time series may in fact be
7 The gender ratio data was obtained from the Census Department’s Inter- stationary. This is termed cointegration.
national Database at 11 The p-values for the Durbin-Watson test are generated via a bootstrapping
8 For a mathematically precise explanation see [70], p. 2.
method and can fluctuate. In this case the p-values were 0.022 for the Internet
9 The tests were performed at the 10% level regression and 0.962 for the cell phone regression.

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

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-

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

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

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

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

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
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
C ELL P HONE P ENETRATION , B OTTOM 20% OF GDP (2006) used globally than the Internet, so this may be a function of
cell technology’s earlier foothold than the Internet’s. As Inter-
Variable Coefficient Standard Error p-value net diffusion catches up to that of cell phone, the democratic
Rule of Law 0.3702 0.4081 0.3660
Voice & Accountability -0.8115 0.3656 0.0281**
metrics may be found to be predictors of this diffusion. This
Political Stability -0.2583 0.1822 0.1585 is not necessarily inconsistent with the populist thread in the
GDP -0.0002 0.0014 0.9053 literature as, according to Bimber, other scholars believe that
POP -0.0309 0.2784 0.9117
Gender Ratio 0.4026 0.3235 0.2155
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
as a function of democratic indicators. Previous research has that “Voice and Accountability” is negatively correlated with
focused on whether ICT use predicts democratic measures. cell phone diffusion. This can be interpreted in favor of the
These results support Bimbers assertions that the structure argument that ICT diffusion can be centralized by government
of political power has not been revolutionized or transformed turning them into social control mechanisms [43]. As the rate
into a new epoch of democracy [12]. Evidence can be found of Voice and Accountability increases, the rate of diffusion
for both the populist thread in the literature and the notion that of ICTs decreases among the poorest and least developed
ICTs may act as an extension of the ruling class. countries. Rates of cell phone use increase globally with
We found a statistically significant positive relationship decreases in political stability, lending further support to the
between the rate of diffusion of the cell phone and the World thesis of ICTs as an extension of ruling class control.
Bank’s “Political Stability” measure capturing perceptions This paradox may be resolved is we consider Page’s view
regarding the likelihood that a government will be destabilized in 1995, that ICT diffusion may be still too nascent and
or overthrown by unconstitutional or violent means. Political insufficient to generate a well functioning system of political
Stability continues to have a significant positive relationship communication and deliberation [48]. It appears we are still
with the rate of cell phone use among the most affluent 20% too early to expect a close relationship between a vibrant
of countries, but that relationship does not hold among the public sphere and ICTs globally. As Bimber states, the “central
wealthiest countries. The “Voice and Accountability” indicator theoretical problem for the populist claim is the absence of a
which measures perceptions of the extent to which citizens clear link between increases in information and increases in
are able to participate in selecting their government, as well popular political action” [12]. This paper provides evidence
as freedom of expression, association and of the media, of the existence of this problem and the need to develop our
was a significant negative predictor of the rate of cell and understanding of this dynamic further.
Internet diffusion among the poorest quintile of countries. The It would be interesting to tie this research more closely to
diffusion of ICT access did not have any significant influence development, in line with Sen’s reasoning that “Developing
on “Rule of Law” while the increasing availability of cell and strengthening a democratic system is an essential compo-
phones were shown to have no influence on any of the three nent of the process of development,” could extend the empiri-
World Bank indicators. cal analysis in a fruitful direction. This might mean specifically
Recall that the Rule of Law metric measures perceptions of testing whether the order in which political and civil rights are
the extent to which individuals have confidence in and abide extended as a country emerges from an autocracy affects the
by the rules of society—in particular the quality of contract rate of development. This could provide a setting in which to
enforcement, property rights, the police, and the courts as well test the “Lee Thesis,” that political rights should be withheld
as the likelihood of crime and violence. This measure was orig- until economic development is achieved.
inally included in the analysis based on Diamond’s research on It may be valuable to explore empirical issues further.
the characteristics of democracy. However, the findings here Modeling the autocorrelation structure with more than one lag
suggest that the increase in “Rule of Law” perceptions does may help reduce autocorrelation. It would also be interesting
not influence ICT penetration, either positively or negatively. to test for cointegration among these variables. Certainly
In other words, perceptions regarding the “Rule of Law” may documented feedback loops exist between our independent
be framed and influenced by factors other than widespread variables, such as GDP and measures of democracy, and taking
ICT use. this explicitly into account may improve the modeling [83],
The populist school of thought believes ICT diffusion will [26]. It is also plausible that feedback loops exist between
decentralize access to communication and information while ICTs and democratic measures and future modeling could
increasing citizen access [12], while Hill and Hughes claim accommodate this. Further research into the modeling aspects
that perhaps the Internet will bring about “a wider demo- could estimate models including variables that control for
cratic revolution in the world” [22]. In our modeling Internet the different manifestations of cell phone and Internet use in
diffusion was not predicted by our measures of democracy, different countries. A more comprehensive model might ex-
implying this revolution has not yet arrived. The rate of cell plore possible non-linearities: whether countries with low ICT
phone diffusion was predicted by higher rates of the “Political adoption rates have different patterns of democratic tendency
Stability” metric. Recall that cell phone are much more widely than those with high adoption rates.

ACKNOWLEDGMENT [30] R. Burkhard and M. Lewis-Beck, “Comparative democracy: the eco-

nomic development thesis,” American Political Science Review, vol. 88,
The authors would like to thank Michael Best and Colin pp. 903–10, 1994.
Maclay for helpful discussion and the anonymous reviewers [31] J. Longdregan and K. Poole, “Poverty, the coup trap, and the seizure of
for constructive criticism. Sadia Ahsanuddin and Kristen Lovin executive power,” World Politics, vol. 42, pp. 151–83, 1990.
[32] ——, “Does high income promote democracy?” World Politics, vol. 49,
provided outstanding research assistance. pp. 1–30, 1996.
[33] B. Geddes, “What do we know about democratization after twenty
R EFERENCES years?” Annual Review of Political Science, vol. 2, pp. 115–44, 1999.
[1] M. Best and K. Wade, “The internet and democracy: Global catalyst or [34] F. Webster, Globalization, information and change. State University of
democratic dud?” June 2006, sam Nunn School of International Affairs. New York Press, 2002.
[2] B. Barber, “Three scenarios for the future of technology and strong [35] R. Reich, The Work of Nations: Preparing ourselves for 21st century
democracy,” Political Science Quarterly, vol. 113, no. 41573-589, 1999. capitalism. Vintage, 1992.
[3] P. Schmitter and T. Karl, “What democracy is . . . and is not,” Journal [36] T. Friedman, The Lexus and the Olive Tree: Understanding Globaliza-
of Democracy, vol. 2, no. 3, 1991. tion. Anchor, 1999.
[4] P. Norris, “Political protest in fragile states,” July 2006, international [37] L. Waverman, M. Meschi, and M. Fuss, “The impact of telecoms on
Political Science Association World Congress. economic growth in developing countries,” Centre for Economic Policy
[5] F. Zakaria, “The rise of illiberal democracies,” Foreign Affairs, Novem- Research, Tech. Rep., March 2005.
ber 1997. [38] D. Richards, Making the national international: Information technology
[6] Huber, D. Rueschemeyer, and J. Stephens, “The impact of economic and government respect for human rights. Albany, NY: State University
development on democracy,” Journal of Economic Perspectives, vol. 7, of New York Press, 2002.
no. 3, pp. 71–85, 1993. [39] D. Barney, Prometheus Wired: The Hope for Democracy in the Age of
[7] E. Zuckerman, “Draft paper on mobile phones and activism,” 2007. Network Technology. University of Chicago Press, 2000.
[Online]. Available: [40] A. Mazrui and R. Ostergard, Technology transfer in the computer age:
draft-paper-on-mobile-phones-and-activism/ The African experience. State University of New York Press, 2002.
[8] D. Kidd, a new communications commons. Routledge, [41] D. Strienstra, Gender, women’s organizing, and the Internet. Albany,
2003, pp. 47–69. NY: State University of New York Press, 2002.
[9] A. Scott and J. Street, “From media politics to e-protest,” Information, [42] S. Kalathil and T. Boas, Open Networks, Closed Regimes. Carnegie
Communication and Society, vol. 3, no. 2, pp. 215–40, 2000. Endowment for International Peace., 2003.
[10] S. Andrew, “The control revolution: How the internet is putting individ- [43] W. Neuman, The future of the mass audience. Cambridge University
uals in charge and changing the world,” The Century Foundation, Tech. Press, 1991.
Rep., 2000. [44] T. Eyck, “Does information matter? a research note on information
[11] R. Clarke, “Information technology: Weapon of authoritarianism or tool technologies and political protest,” Social Science Journal, vol. 38, pp.
of democracy?” August 1994, iFIP World Congress. 147–160, 2001.
[12] B. Bimber, “The internet and political mobilization - research note on [45] N. Postman, Amusing ourselves to death. New York, NY: Vintage
the 1996 election season,” Social Science Computer Review, vol. 16, Books, 1985.
no. 4, pp. 391–401, 1998. [46] H. Schiller, Culture Inc. Oxford University Press, 1989.
[13] R. Dahl, Democracy and its Critics. New Haven, CT: Yale University, [47] W. Lippmann, Public Opinion. New York, NY: Macmillan, 1934.
1989. [48] Page, Who Deliberates? Mass Media in Modern Democracy. Chicago,
[14] H. Rheingold, Virtual Reality. New York, NY: Summit Books, 1991. IL: University of Chicago Press, 1996.
[15] R. Anderson, T. Bikson, S. Law, and B. Mitchell, “Universal access to [49] G. Cherian, “The limits of singapore’s ‘light touch’ web regulation,”
e-mail: Feasibility and societal implications,” RAND, Tech. Rep., 1995. 2008, politics 2.0 International Conference.
[16] A. Etzioni, The Spirit of Community: Rights, Responsibilities, and the [50] R. Deibert, J. Palfrey, R. Rohozinski, and J. Zittrain, Access Denied:
Communitarian Agenda. New York, NY: Crown Publishers, 1993. The Practice and Policy of Global Internet Filtering. The MIT Press,
[17] J. Pavlik, New Media Technology: Cultural and Commercial Perspective. 2008.
Boston, MA: Allyn and Bacon, 1996. [51] S. Oats, “Internet and democracy in russia,” April 2008, politics 2.0
[18] L. Grossman, The Electronic Republic. New York: Viking, 1995. International Conference.
[19] G. Browning, Electronic Democracy: Using the Internet to Influence [52] R. Singel, “Seeking tighter censorship, repressive states target
Politics. Wilton, CT: Online Inc., 1996. web 2.0 apps,” Wired Magazine, March 2008. [Online]. Available:
[20] N. Negroponte, Being Digital. New York, NY: Vintage, 1995.
[21] J. Snider, “New media: Potential information and democratic account- [53] J. Zittrain, The Future of the Internet—And How to Stop It. Yale
ability: A case study of governmental access community media,” August University Press, 2008.
1996, american Political Science Association Convention. [54] S. Mydans, “Myanmar junta unplugs internet,” The International
[22] K. Hill and J. Hughes, “Computer mediated political communica- Herald Tribune, October 2003. [Online]. Available:
tion: The usenet and political communities,” Political Communication, articles/2007/10/04/asia/04info.php
vol. 14, pp. 3–27, 1999. [55] J. Goldsmith and T. Wu, Who Controls the Internet? Illusions of a
[23] A. Corrado and C. Firestone, Elections in Cyberspace: Toward a Borderless World. Oxford University Press, 2006.
New Era in American Politics, A. Corrado and C. Firestone, Eds. [56] L. Lessig, Code: Version 2.0. Basic Books, 2006.
Washington, DC: Aspen Institute, 1996. [57] D. Drezner, “The internet’s effect on state-society relations,” March
[24] C. Steele and A. Stein, Communications revolutions and international 2006, international Studies Association Convention.
politics. Albany, NY: State University of New York Press, 2002. [58] T. Price, Cyber Activism: Advocacy Groups and the Internet. Wash-
[25] J. Rosenau and D. Johnson, Information technology and turbulence in ington, DC: Foundation for Public Affairs, 2000.
world politics. State University of New York Press, 2002. [59] J. Goldsmith, “Regulation of the internet: Three persistent fallacies,”
[26] C. Kedzie, “Communication and democracy: Coincident revolutions Chicago-Kent Law Review, vol. 73, December 1998.
and the emergent dictator’s dilemma,” RAND, Tech. Rep., 1997. [60] R. Beilock and D. Dimitrova, “An exploratory model of inter-country
[Online]. Available: internet diffusion,” Telecommunications Policy, vol. 27, pp. 237–252,
[27] S. Lipset, “Some social requisites of democracy: Economic development 2003.
and political legitimacy,” American Political Science Review, vol. 53, pp. [61] B. Mesquita and G. Downs, “Development and democracy,” Foreign
69–105, 1959. Affairs, September 2005.
[28] R. Jackman, “On the relation of economic development to democratic [62] D. Scheufele and M. Nisbet, “Being a citizen online: New opportunities
performance,” American Journal of Political Science, vol. 17, pp. 611– and dead ends,” The Harvard International Journal of Press/Politics,
21, 1973. vol. 7, no. 3, pp. 55–75, 2002.
[29] K. Bollen, “Political democracy and the timing of development,” Amer- [63] H. Milner, “The digital divide: The role of political institutions in tech-
ican Social Review, vol. 44, pp. 572–87, 1979. nology diffusion,” August 2003, american Political Science Association.

[64] L. Diamond, “elections without democracy: Thinking about hybrid [76] E. Mansfield and J. Snyder, “Conceptualizing and measuring democ-
regimes,” Journal of Democracy, vol. 13, no. 2, April 2002. racy:evaluating alternative indices,” International Organization, vol. 56,
[65] ——, Developing Democracy: Toward Consolidation. John Hopkins no. 2, pp. 297–337, 2001.
University Press, 1999. [77] H. Barnes, “The contribution of democracy to rebuilding postconflict
[66] D. Kaufmann, A. Kraay, and M. Mastruzzi, “Governance matters vii: societies,” The American Journal of International Law, vol. 95, no. 1,
Aggregate and individual governance indicators 1996-2007,” World pp. 86–101, 2001.
Bank, Tech. Rep., 2007. [78] C. Shirky, Here Comes Everybody: The Power of Organizing without
[67] A. Sen, Development as Freedom. Oxford University Press, 2001. Organizations. Penguin Press, 2008.
[68] “Opennet initiative.” [Online]. Available: [79] C. Argevou, “Ols bias in a nonstationary autoregression,” Information
[69] R. Faris and N. Villeneuve, Access Denied: The Practice and Policy of Technology for Development, vol. 8, no. 1, pp. 15–28, 1998.
Global Internet Filtering. MIT Press, 2008, ch. 1. [80] B. Oyelaran-Oyeyinka and K. Lai, “Internet diffusion in sub-saharn
[70] C. Granger and P. Newbold, “Spurious regressions in econometrics,” africa: A cross-country analysis,” Telecommunication Policy, vol. 8,
Journal of Econometrics, vol. 2, pp. 111–120, 1974. no. 1, pp. 507–527, 1998.
[71] “R.” [Online]. Available: [81] J. Gibbs, K. Kraemer, and J. Dedrick, “Environment and policy factors
[72] C. Granger, “Some properties of time series data and their use in shaping global e-commerce diffusion: A cross country comparison,” The
econometric model specification,” Journal of Econometrics, vol. 16, pp. Information Society, vol. 19, no. 1, pp. 5–18, 1993.
121–130, 1981. [82] N. Kshetri and N. Dholakia, ENCYCLOPEDIA OF INFORMATION
[73] ——, “Co-integration and error correction: Representation, estimation SCIENCE AND TECHNOLOGY, 2nd edition. Idea Group Publishing,
and testing,” Econometrica, vol. 2, pp. 251–176, 1987. 2008.
[74] T. Amemiya, “The nonlinear two-stage least squares estimator,” Journal [83] P. Norris, Digital Divide: Civic Engagement, Information Poverty, and
of Econometrics, vol. 2, pp. 105–110, 1974. the Internet Worldwide. Cambridge University Press, 2001.
[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.

A review of the research on mobile use by micro

and small enterprises (MSEs)
Jonathan Donner and Marcela Escobari

Like landlines, mobile phones allow people to communicate

Abstract— The paper offers a systematic review of 14 studies at a distance and exchange information instantaneously. Thus,
of the use of mobile telephony by micro and small enterprises there is significant potential for mobile use to increase MSE
(MSEs) in the developing world, detailing findings about changes productivity. However, since the dynamics underpinning this
to enterprises’ internal processes and external relationships, and
findings about mobile use vs. traditional landline use. Results
potential are nuanced, and since current supporting evidence
suggest that there is currently more evidence for the benefits of is scarce and methodologically heterogeneous [9], it is
mobile use accruing mostly (but not exclusively) to existing MSEs important to rigorously examine mobile use by MSEs. For
rather than new MSEs, in ways that amplify existing material example, there is a difference between using a mobile to serve
and informational flows rather than transform them. The review existing customers more effectively, and using it to start a new
presents a more complete picture of mobile use by MSEs than business. There is a difference between using a mobile to
was previously available to ICTD researchers, and indentifies
priorities for future research, including comparisons of the
check market prices and using it bypass a middleman who
impact of mobile use across subsectors of MSEs and assessments carries goods to market. Popular press and practitioner reports
of use of advanced services such as mobile banking and mobile generally fail to make these distinctions.
commerce. 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
Index Terms—Business Economics, Communication, ICTD, literature, identifying known patterns of mobile use, as well as
Mobile Communication, MSE, Social Factors
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].

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].
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,
The majority of non-agricultural enterprises in the wholesalers, retailers and end-customers. Finally, it explores
developing world have ten or fewer employees [15, 16]. These two elements unique to mobile communication—the increased
MSEs employ up to 25% of working-age adults in some spatial and temporal mobility afforded by wireless devices,
countries [16], and while the contribution of MSEs to and the resulting blurring of the personal and the professional
aggregate economic growth remains a matter of debate [17], spheres—to assess how MSE mobile use differs from landline
their importance to household livelihoods and poverty use.
alleviation is undeniable. Thus, MSEs are the focus of
programs at many of the world’s largest development II. MICRO AND SMALL ENTERPRISES (MSES)
institutions [18].
Working definitions of MSEs vary from country to country
Since the year 2000, the spread of mobile telephony across
and from researcher to researcher [24]. This analysis defines
the developing world has raised hopes among policymakers
an MSE as any non-farm i enterprise, formal or informal, with
that MSEs will benefit from easier access to
less than 50 employees, including sole proprietorships, part-
telecommunications. The successful entrepreneur, suddenly
time businesses, and home-based businesses. The size
enabled by his mobile phone, has been given a prominent role
thresholds draw on Mead and Leidholm, [16], who note that
in the global development narrative and become a semi-
the absolute majority of such enterprises in the developing
regular fixture in both the popular press [19, 20] and
world are sole proprietorships, and that firms with less than 10
practitioner media [21, 22].
employees substantially outnumber larger enterprises.
A number of factors distinguish the term MSE (micro and
Manuscript version received February 20, 2009. small enterprise) from SME (small and medium enterprise).
J. Donner is with Microsoft Research India,
M Escobari is with the Center for International Development, 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 Particularly focused and powerful evidence appears in
between micro, small, and medium, there are often implicit Jensen’s [10] research on the fishermen of Kerala. Working
conceptual overlaps between the acronyms. with five-year time series data at three fish markets in coastal
Unlike SMEs, the majority of MSEs are informal India, Jensen and his team found that “the adoption of mobile
enterprises. Once again, however, there is no universal phones by fishermen and wholesalers was associated with a
standard to determine what makes an enterprise informal vs. dramatic reduction in price dispersion, the complete
formal [6]. In many cases, even the term entrepreneur may be elimination of waste, and near-perfect adherence to the Law of
a bit of a romantic misnomer. Evidence suggests that among One Price. Both consumer and producer welfare increased.”
MSEs, only a small minority of enterprises are poised for [10: 879] Soon after the introduction of mobile coverage,
growth; most remain small or struggle to survive, and yield a fishermen bought mobiles and accumulated lists of up to 100
low return on labor and capital [17, 25]. buyers in their handsets’ address books; subsequently, while
Though significantly less growth-oriented and productive still at sea, fishermen could call a range of possible landing
(on average) than SMEs and other larger firms [17], MSEs points and buyers in order to determine the best price and best
share a basic similarity with all enterprises; each combines place to sell their catch.
investments in capital with some labor (their own, their By contrast, Jagun, Heeks, and Whalley’s [9] examination
families’ or their employees) in the hopes of yielding a of the mobile’s role in mediating supply chains in the Nigerian
product or service whose market value exceeds the cost of market for traditional hand-woven ceremonial cloth is broad
those inputs. Thus, there has always been a thread in the ICTD in scope. It offers a multidisciplinary literature review, a
literature that seeks to understand how various technologies conceptual framework articulating effects at multiple levels,
could be used advantageously by MSEs [26-28]. Prior to the and a detailed case study. They describe “process” benefits to
widespread introduction of the mobile into the developing mobile use, as calls at a distance reduce the time of trades and
world, the landline’s importance in this regard was already replace costly journeys. They also describe “structural”
clear: impacts; finding no disintermediation of traders, but rather an
intensification of their role. Traders are more likely to have
Phones are the information-related technology that
mobiles than the less prosperous weavers in the supply chain,
has done the most to reduce costs, increase income
and reduce uncertainty and risk. Phones support the and thus are better positioned to coordinate with a wider range
current reality of informal information systems, they of downstream customers and to maintain a more dynamic and
can help extend social and business networks, and responsive set of relationships with weavers. For example,
they clearly substitute for journeys and, in some weavers previously had to pay cash to get their supplies.
cases, for brokers, traders and other business Mobiles give weavers access to credit by enabling calls on
intermediaries. They therefore work “with the grain” their behalf to fabric vendors by traders, who vouch for the
of informality yet at the same time help to eat into the veracity of weavers’ orders, and promise to cover the costs of
problems of insularity that can run alongside. Phones the fabric in advance of the completion of the weavers’ work.
also meet the priority information needs of this group
of communication rather than processing of IV. METHODS AND CODING PROTOCOL
information [27: 18] Many of the studies of mobile use by MSEs are qualitative,
The quotation focuses directly on the basic tasks of running and do not report statistical findings. Even among quantitative
a business—reducing costs, increasing income, managing studies, there is little agreement in terms of dependent and
risk—and links them to core functions of mediated independent variables under scrutiny. Thus, a statistical meta-
communication technologies, particularly the substitution for analysis would not be applicable [30]. Similarly, a method
journeys. As demonstrated elsewhere, [26] the key is designed specifically for comparing ethnographies, such as
increased productivity. reciprocal translation [31] would be unlikely to bridge
qualitative and quantitative studies.
III. STUDIES ON MOBILES AND MSES The analysis draws instead on a systematic review
Recently, studies have emerged that directly address how methodology [32] to aggregate findings across the available
MSEs in the developing world are using mobiles rather than studies. By using a standardized protocol, coding each
landlines or other ICTs. The studies are not as numerous as individual study for the appearance or absence of certain
the enthusiasm in the popular press might suggest. They are a assertions, the review assesses and parsimoniously represents
tiny fraction of the total literature on mobile use in the what the research literature, in aggregate, suggests about
developing world [29]. They have emerged from different mobile use by MSEs. The exercise relies on clearly articulated
disciplines, and, as relative contemporaries, often do not cite eligibility criteria to select studies and on standardized
each other. This section presents two studies representing questions to evaluate them. These two levels of
distinct methodologies and conclusions, to provide an standardization, agreed upon before the formal review
example of the range of available perspectives and to set the commenced, separates the exercise from a conventional list-
stage for the systematic review. based or thematic/narrative literature review.
A. Selecting studies and producers, for example), but it became clear that the
Papers were initially identified by online literature and population of existing studies is too small to support that
database searches (using keyword combinations of mobile, inquiry.
cellular, microenterprise, MSE, etc.), and by snowball The final protocol employed three distinct foci. First, it
references from the bibliographies of studies already in hand. assessed the impact of mobile use on the internal process of an
Next, these studies were assessed against a series of eligibility enterprise, using Porter’s value chain model [23], Fig. 1. The
criteria: to be included, studies had to be specific to both value chain comprises the activity inputs into a product or
mobiles and MSEs, report generalizable findings, and contain service: inbound logistics, operations (production), outbound
detailed primary data about mobile use in everyday logistics, marketing and sales, and after-sales service. The
conditions. What started out as a reasonably large body of value chain also includes supporting functions: firm
studies was trimmed back significantly. In order to provide infrastructure, human resources, technology development
additional resources to other researchers, however, this section (knowledge developed or owned by the enterprise), and
lists those excluded papers, along with the rationale for the procurement. Together these activities can create customer
decisions. value in excess of the costs to provide it, yielding profit.
To be included in the review, papers had to be specifically Porter [23: 168], argues that information and communication
focused on mobile phones, which excluded some excellent technologies can be used to improve almost any of these
research on landlines or payphones and MSEs [28, 33, 34]. primary and supporting activities. Although the value chain
The papers also had to be about MSEs, not SMEs. Papers that framework was developed with larger enterprises in mind, it
did not explicitly include sole proprietors and informal can be applied to MSEs, since in small firms the same
enterprises were excluded [35-37]. individual can carry out different business-related activities
We made a more difficult decision to exclude papers that during the day. (Indeed, even in larger firms there is often not
were not generalizable to a wide range of MSEs. An a perfect mapping between the activities and functions in the
important line of research explores how many individuals earn value chain and distinct people or departments). Both
livelihoods in the mobile business itself, by selling airtime, researchers coded individual papers for mentions of the
fixing handsets or operating village phones [38-40]. However, mobile’s role in any of the primary or supporting functions.
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 FIGURE 1: PORTER’S VALUE CHAIN
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. note: image released to public domain as per
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 The second analysis used another Porter framework, the
retained. value system [23] ii , to offer a corresponding assessment of
how mobile use influences the network of interdependencies
B. Evaluation protocol and relationships external to the enterprise, including
The process of developing the evaluation questions was producers, traders, wholesalers, retailers and end-customers.
iterative. It was based mostly on an initial reading of the An initial reading of the papers identified four categories of
documents by the researchers, while also integrating current potential impacts. Some research stresses (a) the increased
narratives in the popular and practitioner literatures. An availability of information in the network; other studies stress
original goal was to code studies according to subcategories of (b) the entry of new actors, particularly buyers and sellers,
MSEs (to discern differences in mobile use between traders into markets. Both factors tend to increase competition, but do
so in different ways. One focuses on the actors in the network, V. RESULTS
the other on the information those actors exchange.
Nevertheless, the two changes do not necessarily move in A. Enterprise value chain
tandem; it is one thing to assert that the same set of actors Most studies mention the core processes of marketing and
exchange information at lower cost and higher frequency, sales [1, 2, 4, 6, 7, 9-13]. Analyses ranging from Jensen’s [10]
another to say that markets have expanded. This information- model of searching for the best price for fish to Kamga’s [11]
vs.-actors cut highlights this distinction. Two other categories description of improvements to the local laundry services in
of network impacts can be expressed as assertions that (c) Cote d’Ivoire asserts that mobiles help connect vendors and
mobiles help enterprises cut out middlemen and that (d) buyers, often at a distance and usually at lower cost than an
mobiles help individuals start new businesses. in-person journey. Esselaar et al. [6] report results of a survey
Unlike the value chain analysis (coded for affirmative of SMEs, including 1/3 microenterprises, conducted across 13
mentions only), the value system analysis coded for both countries. “Mobile phones are used more often for keeping in
affirmations and negations of the four potential impacts. After contact with customers and clients” (p 92). This is the highly
the initial reading, we elected to track negations since some of visible, intuitive role of mobiles for small enterprises.
the sources made a point of arguing against one or more of the The picture is sparser for other core processes within the
assertions from the popular and practitioner literature. value chain. Three studies mention inbound and outbound
The third analysis explored how the impact of mobile use logistics [1, 9, 13], particularly Abraham [1], who details how
on small enterprises is or is not different from the impact of fishermen can now use small supply boats (dispatched via
landline telephony on those same kinds of businesses [28, 33, mobile) to stay out fishing longer. Overå [13] describes how
34]. Technological properties of mobile communication make traders in Ghana can time harvests (inbound) and change the
it inherently more prone to adoption by MSEs than landlines: terms of delivery financing (outbound) because of the mobile.
it is cheaper to build towers than lay cable, prepay accounts Operations receives two mentions, again by [1], who notes
have no startup costs, and inexpensive/used handsets are that fishermen use mobiles to coordinate the timing and
readily available. However, in this case the third analysis location of when to drop nets and search for fish. Similarly,
focused narrowly on two differences in use rather than cost or [9] describes how weavers call customers mid-process to
access. revise plans for the garments they are creating.
First, studies were coded for mentions of mobility. Mobility Only studies by Frempong et al. [7] and Molony [11]
is a crucial difference between mobiles and landlines—while mention after-sales service. Molony describes how Tanzanian
landlines connect places to places, mobiles generally connect exporters of carved wood use the mobile to elicit feedback and
people to people, wherever they are and regardless of the time built trust with buyers after (and ideally between) sales.
and situation. This mobility leads to increased individual In terms of crosscutting functions, five studies reference
addressability, and can change how people structure social procurement [1, 6, 7, 9, 13] and address price search by
and economic activity [53-55]. Most relevant to this analysis, buyers of inputs (or by traders). There is little evidence to date
mobility may enable the rise of roaming businesses, just-in- for the mobile’s role in transforming the proprietary
time service and what Townsend [56] has called the “real time technology, infrastructure or HR functions of MSEs, perhaps
city”. And yet some evidence suggests that in the developing because these enterprises are too small to invest in these
world many mobiles are purchased as substitutes for landlines, assets. Although studies outside the review [36, 58] provide
rather than complements to them [57]. anecdotes of small employers giving mobiles to employees,
Second, the nature of the mobile as a portable, personal this infrastructure function is of limited utility for tiny firms
device means it is particularly easy to use for both personal and sole-proprietorships.
and business functions during the same day. Thus, studies of
B. Industry value system
the role of mobiles in the lives of MSE operators are often
different from studies of the role of the device in the The second analysis turns the lens outside the enterprise,
businesses themselves. The analysis coded for studies that towards its location in a network of relationships. The most
explore these social functions. common finding links mobile use to an increase in the flow of
Once the protocol was established, each researcher re-read information between actors in the value system [1, 2, 6, 7, 9-
the papers, coding them in isolation. We then compared our 14]. The two primary sub-themes are more frequent or wide-
codes and resolved any discrepancies through discussion. The ranging exchanges of price information [1, 2, 13], and a more
resulting codes are less prone to reflect the bias of a single generalized discussion of increased communication with
reader. Of the 112 cells on the matrix requiring codes, 16 customers [6, 12, 14]. These findings are reflections of the
required discussion to resolve coding discrepancies between frequent references to marketing and sales and procurement
the two researchers. 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


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

The analysis includes two studies of fishermen but excludes
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

the authority of the Ministry of Education, Science and

Abstract— The paper demonstrates how appropriate technology, Technology (MoEST). Funding for the initiative was provided
when combined with quality curriculum-based content, has the by a British company who also developed the educational
potential to have a positive impact on primary education within
software for the portable devices.
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 II. CONTEXT
findings of the study. Significant impacts from the introduction
of the program were increased school attendance, reduced drop- A. Education in Malawi
out rates and improved student and teacher enthusiasm. Free primary education was introduced in Malawi in 1994 as
Alongside this, the audio and video material and interactive
learning techniques offered potential pedagogical benefits in
a result of the commitment made by the country at the World
combining learner-centered and outcome-based activities with Conference on Education for All (EFA) in 1990. Widespread
continuous assessment, helping children to retain information agreement exists regarding the vital place of education within
more effectively and record higher test scores. Major challenges poverty reduction efforts, capacity building and growth
were also identified regarding implementation and sustainability. strategies of developing countries. In Malawi, numerous
These centered around the need for rigorous teacher training, policy, budgetary and multilevel commitments have
classroom integration, appropriate deployment, maintenance,
sustained impact and overall cost-effectiveness. In closing, the
contributed to significant progress in the delivery of primary
paper emphasizes the need for such programs to be driven by education since 1994. This includes enrolment figures rising
educational concerns and recognizes the similar challenges faced from 1.9 million to 3.2 million, the construction of 1,000 new
in many related initiatives. classrooms and roll out of the new curriculum to 5,500
schools [1]. However, Malawi is one of the poorest countries
Index Terms— Education, Evaluation, Malawi, Technology 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
I. INTRODUCTION rapid advancements in access to free schooling have put the

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
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
the potential role of technology within education. These two increasing class sizes, a lack of fully qualified teachers,
seemingly disparate themes are brought together through the limited teaching materials and inadequate infrastructure. It is
case study of the Interactive Learning Program. An overview estimated that an extra 8,000 teachers are currently required to
of the program is provided, followed by an explanation of the meet the MoEST target pupil to teacher ratio of 1:60 [3]. In
methodological approach employed in the monitoring and addition, and despite national net enrolment levels of 91%,
evaluation exercise. The findings are then categorized into national drop out rates remain high with only 44% of those
four sections of analysis which inform the concluding that enroll in Standard 1 completing Standard 5 and less than
comments and recommendations. 30% reaching Standard 8, the final year of primary education
The Interactive Learning Program in Malawi comes under [4]. Linked to this and demonstrating the ongoing challenge in
regard to teaching capabilities and educational outcomes is a
Manuscript received September 21, 2008. declining performance in national examinations with a failure
David Hollow is a PhD candidate with the ICT4D Collective in the rate of over 40% [1].
Geography Department at Royal Holloway, University of London. e-mail: The importance of ensuring good quality education
Paola Masperi is an independent development consultant, based in London. alongside improved access [5] has been long recognized in
e-mail: Malawi and in 1999 it was decided that a comprehensive
reform of the curriculum was required due to the based content and sustained classroom integration.
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 III. INTERACTIVE LEARNING PROGRAMME
been systematically incorporated into the new primary school Having outlined the overall context of education within
curriculum in the form of life skills education. Malawi and the theoretical role of technology within
education, attention now turns to the Interactive Learning
B. Technology in Education Program. It considers whether the initiative constitutes a
suitable application of portable technology to support the
There is significant attention focused upon the potential for
provision of basic education within the country. To begin, the
ICT to assist in leapfrogging educational challenges within
specifics of the program are explained, including the
developing nations [6] [7] [8], with much enthusiasm for a
background, technology and pedagogical rationale.
possible technology-enabled ‘breakthrough in learning’ [9].
However, much debate surrounds the question of whether the
infusion of technology into education has actually instigated A. Background to Program
more than incremental changes to the field. Proponents assert The Interactive Learning Program was introduced in 2006
that the last decade have resulted in the emergence of a new and following initial positive feedback the MoEST requested,
landscape for education across the globe [10] with technology in early 2007, that the initiative be scaled up to incorporate 50
positively effecting student motivation [11] and, when primary schools. The schools that the MoEST selected for
initiatives are implemented with fidelity, leading to a participation ranged from Karonga District in the north
significant increase in learning [12]. This is challenged by through to the southernmost districts of Mulanje and
those who argue that achieving structural technological Phalombe. A total of 520 custom-made handheld interactive
change in schools takes much longer than anticipated [13] learning aids were distributed to the participating schools.
with no substantial evidence that the introduction of This totaled ten devices for each school, except for two test
technology has yet caused any fundamental change in either a schools which received 20 devices for the purposes of targeted
developed or developing context [14]. Wagner [15] notes a monitoring and evaluation.
wide variety of outcomes from ICT for education projects in
developing countries, with significant negative impacts
including the reinforcing of dependencies, imposition without B. Technology Utilized
community involvement and collapse due to lack of funding The device in question is a handheld interactive learning aid
or political commitment. (from herein referred to as the learning machine or gadget as
Within this contested environment, the range of technologies named by the participating children), slightly larger than a
theoretically available for deployment within developing mobile phone and able to play video and audio through either
country educational contexts is rapidly expanding, with a a loudspeaker or headphones. Positioned next to the screen are
transition towards increasingly portable, powerful and a selection of buttons which can be pressed in response to
adaptable tools [16]. The new technologies available have questions asked in the lesson being watched. The device has
potential to mark a significant transition beyond an in-built rechargeable lithium battery with power for
conceptualizing e-learning through conventional static between four and six hours of continuous play. A total of 25
computer laboratories [17], emphasising instead the place of lessons in Chichewa and 40 in English are preinstalled and
anytime, anywhere computing [16] and mobile learning [18]. stored on two gigabytes of internal Flash memory, leaving
Such a shift creates opportunity for major changes in the additional room for newly developed lessons to be
application of educational technology, with key potential incorporated. The lessons are designed for use by children in
benefits noted as increased enthusiasm, cooperation, Standard 3, 4 and 5, each lasting approximately 20 minutes
communication and student ownership [19]. and covering a range of curriculum based subjects such as
Despite the potential benefits of learning technologies [20] General Studies, Social Studies, Science, Mathematics,
it is important to recognize that not all solutions which have Geography, Life Skills and English.
contributed to educational advancements in the developed With less than 5% of primary schools across Malawi
world can simply be transferred to a developing world estimated as having access to a reliable supply of mains
context. In order for the potential of ICTs to be realized in electricity [3] the initiative was dependent upon solar charging
improving the quality of education it is important they are systems. Each of the project schools were equipped with a 20-
applied with cultural understanding, local knowledge and Watt solar panel connected to a deep-cycle sealed battery,
sensitivity [11] [21]. Indeed, ‘effective use of ICTs must be generating enough energy to power 20 of the learning
tied to the needs of developing countries and challenge the machines. Teachers were given training regarding appropriate
one size fits all approach of many programmes’ [19;7]. A charging and usage, and instructed to leave the devices to
recognition of this fact emphasises the need for primary focus recharge overnight once every two days.
on appropriate software development linked to curriculum-
C. Pedagogy monitoring and evaluation. Where undertaken, there is often
The digital content on the learning machines is designed in strong emphasis placed on hardware-based input indicators
order to enable and promote interactive and outcome-based [25] rather than more complex learning outcomes. In order to
learning, linked to the Malawi national curriculum and avoid this simplification and demonstrate an alterative
actively supporting the government Primary Curriculum approach, a rigorous monitoring, evaluation and impact
Assessment Reform [22]. The lessons are designed to be user assessment structure was built into the program throughout
friendly, with the audio-visual content accessible to illiterate implementation. This involved three field visits from the
or semi-literate users. Instructions from lip synchronized monitoring and evaluation team in September 2007,
cartoon characters explain the functions of the various buttons November 2007 and March 2008.
on the learning machine. Once a lesson title has been selected In addition to being marginalized, monitoring and evaluation
the methodology becomes interactive and learner-centered. of ICT for education programs has rarely embraced creative
Having listened to brief teaching points the learners are tested processes of assessment [26]. Recognizing this, a systems-
on their understanding of the information supplied through the based approach was utilized in designing the methodology
posing of multiple-choice questions answered by pressing [27] [28] [29]. This involved adopting a mixed- and multi-
buttons on the device. At the end of the lesson learners are method approach which focused on process and aimed to
given opportunity to test what they have learnt through engage with multiple stakeholders so as to gain a plurality of
undertaking an overall quiz of ten questions where they are perspectives regarding program impact. It also ensured that
congratulated according to the score attained. This outcome- data gathered was not solely anecdotal but credible,
based approach provides incentive for the children to work dependable and confirmable [30]. Within this overall aim for
conscientiously and allows the teacher to monitor progress. increased creativity and rigor, the research remained subject to
The decision to include lessons in both Chichewa and the standard complexities and constraints of limited time,
English was taken on the basis that early learning content is budget, data and personnel [31] [32].
most effective when communicated in the vernacular language It light of the constraints it was decided by the MoEST that
[23] [24]. The dual language approach equips children for the five representative test schools would serve as a sample of the
challenging transition from Chichewa to English as the 50 participating in the program. The schools selected by the
official medium of school instruction which takes place in MoEST were Chin’gombe, Mwatibu, Mthentera, Mbinzi and
Standard 5. The most conducive learning environment for Dzenza. None of these schools had any form of ICT
using the devices is with groups of between four and six incorporated into the curriculum prior to the introduction of
students. In this context team work, group participation and the program. A total of 15 days were spent in these schools
the development of leadership skills can each be encouraged. 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
IV. METHODOLOGICAL APPROACH alongside the rationale for each.
A total of 15 lessons in which the learning machines were
In regard to the overall context of technology within
being used were observed throughout the process [34].
education, Kozma [25;21] concludes after assessing a wide
Teachers were asked to conduct the lesson as they would
variety of studies that there is ‘no consistent relationship
normally, without altering classroom arrangements. Observing
between the mere availability or use of ICT and student
usage of the technology in a normal environment provided an
learning’. Beyond such generic assertions, there are
ideal foundation for the subsequent methods, giving
significant knowledge gaps remaining regarding what kind of
opportunity to discuss successes and suggest ways to work
initiative works and what does not. As InfoDev [11;5]
around challenges with the children, teachers and head-
summarize, ‘despite thousands of impact studies, the impact of
ICT use on student achievement remains difficult to measure
Some 15 group interviews were held with children in the
and open to much reasonable debate’. In light of such
five schools, in order to hear what difference the program had
observations all partners in the Interactive Learning Program
made to their lives. The children were asked what they liked
initiative recognized the necessity for a comprehensive
and disliked and improvements they would make for the
monitoring and evaluation exercise, the key objectives of
future. The interviews followed a semi-structured, guided
which were to;
approach [35] [36] facilitated by a teacher and conducted in
• assess the impact of the program on primary education
Chichewa. Participants were selected at random from
• assess the feasibility of the program in each test school
Standard 3 and 4 classes, were aged between 7 and 15, and
• identify program weaknesses for future refinement
had equal gender representation.
• develop both teacher and organizational capacity
Following these group interviews, more detailed
• provide feedback to partners regarding suitability of scale-
conversations were held with individual children in order to
up and sustainability
learn about particular experiences or perspectives that they
A common theme from ICT-enhanced education programs
had expressed. Such unstructured and story-based
within developing countries is the marginalized place of
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 V. ANALYSIS
interviews provided opportunity to obtain input regarding the The monitoring and evaluation exercise was focused on
program feasibility, impact and future direction. assessing three interconnected spheres which cut across the
Conducting focus groups was considered to be the most five previously identified research objectives. These were
appropriate method to gain detailed feedback from teachers impact upon students, impact upon teachers and effectiveness
regarding their view of the program and the difference it had of the technology. Analysis of each sphere is based on the data
made to their lives, including positive and negative impacts gathered using the range of methodological approaches
and potential improvements. A total of 15 focus groups were previously outlined. The analysis is illustrated throughout with
conducted in this safe environment [38] which facilitated quotations from children, teachers, head teachers and MoEST
robust dialogue [39] but ensured no one felt pressurized to officials.
share personal experiences [40].
Building on these, the learning octagon was a research tool
developed specifically for the program in order to stimulate A. Impact upon Students
more detailed discussion between teachers in the focus In assessing the overall impact upon students the study
groups. The octagon providing a pictorial representation to considered the specific impact on four main areas of student
assess the strengths and weaknesses of eight different attendance, enthusiasm, attainment in curriculum subjects and
dimensions to the program by drawing on a combination of attainment in life skills subjects.
the Octagon tool [41] and Most Significant Change approach A significant and universally agreed impact of the program
[42] [43]. This enabled the teachers to collaborate on creating was the increase in school attendance. Teachers reported a
a visual representation of the impacts of the initiative. large increase in class sizes with fewer children absent than
In addition to these methods, and in order to enable on- prior to the intervention. One student from Standard 4
going monitoring on a weekly basis, one teacher from each of explained why this was the case:
the five schools was selected to complete a diary documenting Before the gadgets more pupils were absenting themselves
their experiences of using the learning machines in their from classes but now we encourage our fellow pupils to come
lessons. The diaries were semi-structured and designed to to school and tell them, today if you absent yourself, you will
provide a continuous record of particular strengths, miss using the gadgets. I used to absent myself 50% of the
weaknesses and challenges encountered. time before the gadgets came, now I come to school everyday.
Baseline tests were conducted at stage one and three of the This is a significant achievement within a national context of
research. The objective of the test was to provide a high drop-out rates and absenteeism. For any educational
quantitative assessment of the impact of the program on the intervention to succeed it is necessary for learners to be
attainment of the children regarding both curriculum and life- attending school on a regular basis and the motivating
skills. In each of the five test schools there were 12 children influence of appropriate technology is clearly a contributing
randomly selected from Standard 3 and 4, providing a total of factor in achieving this desired end. Despite strong anecdotal
60 children, 30 boys and 30 girls. Five additional schools evidence, assessing the statistical significance of the change in
were selected to act as a control group with an additional 60 attendance was hampered by a lack of daily attendance
children tested. records maintained in each school, meaning it was impossible
Alongside focusing on the test schools, evaluation to track exact attendance patterns prior to the current term.
questionnaires and equipment feedback forms were distributed Children in the group interviews reported that they were
to all 50 participating schools in order to gain a broader enthusiastic about coming to school now that they were using
understanding of program impact. The questionnaires gave the learning machines. Several students reported having
opportunity for feedback regarding patterns of usage and shared the experience with family and community members
challenges encountered in implementation. The equipment who actively encouraged them not to miss out on the
feedback forms enabled the documentation of technical opportunity to learn with the new technology. As one student
problems encountered throughout the program. from Standard 3 explained:
Each of the methods outlined above enabled a detailed They [parents] said to me that you have to work hard and
perspective to be developed regarding the strengths and make sure you do not run away from lessons when you are
weaknesses of the program. The monitoring and evaluation using the computer. I used to run away from class but now I
methodology employed was participatory throughout [44] and have changed my behavior … because I am attracted to the
engaged with children, teachers, headmasters, community gadgets and if I miss the chance to use it, it will never come
leaders, government officials and civil society representatives. again.
Each school was visited at least three times throughout the six Other children emphasized what they had learnt through the
month research period and this allowed for a progressive lessons and how this related to what they hoped to do in the
approach and the development of good relationships with a future. One student from Standard 5 reported the impact of a
number of stakeholders. particular lesson regarding counting:
Now I know how to add and subtract – I could see the water. I did not know this before the gadgets.
pictures and I was able to subtract. This makes a foundation Children were also able to talk about culturally sensitive
for me to be able to work in a bank – this is what I want to do subjects such as HIV/AIDS and explain what they had learnt
in the future. from the lessons. Teachers stressed the value of the interactive
The learning machines are perceived by the children as a lessons in a cultural context where such discussions are often
mixture between mobile phones, games devices and video considered taboo. The children requested that new content be
players. This makes them attractive to the children, who added to the devices so that they could learn about a greater
quickly become familiar with them and enjoy the process of variety of topics. The most commonly requested subjects were
being congratulated by the device for answering questions Agriculture, Mathematics, English, Religious Education,
correctly. Observed learners appeared to enjoy working in Science and Technology, Physical Education and Music.
groups and taking turns in pressing the buttons. Teachers also Similarly, the teachers requested that new content would
noted a similar impact and emphasized the improved listening remain linked with the curriculum reform in order to provide
skills of the students: them with support in this transition.
They are able to explain things now – and the gadgets really The impact of the initiative on the students is intrinsically
help them with listening skills – if they do not listen then they linked to the way in which the teachers are affected, and this
miss what has been said and they cannot answer the question is now considered.
– and it makes them be fully attentive.
Increased attentiveness in class and greater motivation to
B. Impact upon Teachers
attend school also had an effect on the attainment of the
children in both curriculum and life skills based lessons. Both In assessing the overall impact upon teachers the study
students and teachers reported that the use of audio and video, considered two factors, teacher enthusiasm and teacher
as well as the continuous assessment quizzes, increased workload.
retention and affected attainment. One teacher reflected on The majority of teachers were enthusiastic about the
tests completed the previous week: introduction of the program into their school and several
We had the mid-term tests last week and more pupils did commented that they felt honored to have been selected as a
well than before – when I asked them why they said it was school for the pilot. They were pleased that MoEST officials
because of the gadgets – ‘we just remember what we have were visiting their school and showing an interest in their
learnt on them’. The increase in attainment has gone up by work. They were keen to learn about technology and could see
about 30% on average from what they normally achieve. the positive effects on themselves and the learners. They also
Despite such assertions it was difficult to assess quantitative noted how the introduction of the learning machines had
impact on student attainment due to the short period of time helped them in altering their teaching style, adopting new and
since the beginning of the intervention. A comparison of innovative approaches:
scores between the baseline and second test demonstrated that We love it – before the gadgets it was just talk and write for
certain sectors of the curriculum had considerably more some subjects – but now the kids can see the pictures. Some of
correct answers after using the learning machines. However, the children can be sitting doing one thing while others are
the majority of students had only completed 15% of available doing the gadgets.
lessons at the time of the second test and the overall Young trainee teachers particularly enjoyed using the
improvement in curriculum attainment was limited to 1.5%. technology and often engaged in using it without
Similarly, improved attainment in life skills was difficult to apprehension. This confidence had a clear effect on the
quantify as test scores related to knowledge gained rather than manner in which the children approached the devices and
necessarily to behavioral and lifestyle change. The overall made use of them in the classroom. The enthusiasm of the
impact on life skills attainment in the baseline test was a 3% students also affected the teachers:
improvement. Again, the questions which were answered …because if the learners are enthusiastic then it makes us
correctly indicated an increased understanding of the content enthusiastic. If the gadgets help the students to learn then the
for those lessons which the teachers had chosen to use most teachers have to be happy that the students are doing well.
regularly. However, this enthusiasm was not felt universally. Mature
Although quantitative change proved difficult to assess, teachers often showed less interest in the program, were
many students interviewed were able to recall accurate fearful of using the technology and reluctant to let the children
information on the life skills lessons they had used and use it independently. In one of the test schools after six
expressed satisfaction to have learnt skills useful to their daily months of usage there were only three teachers still using the
lives. Learners explained how they had put into practice what learning machines. All the other teachers in the school had
had been taught in the lessons, indicating potential behavioral become uninterested in the program, considering the
change: additional workload to be too much of a burden.
[my favorite lesson was…] Preventing malaria – because it The teachers reported a variety of different experiences
affects many children in the village. I learnt that we must regarding the impact of the program on their workload. This
sleep under treated bed nets and must not play with stagnant was dependent upon the manner of implementation in each
school and the degree to which the learning machines were the momentum of the program and improving teacher
incorporated within the lesson schedule. The overall feeling motivation. The limited budget of the MoEST makes it a
from the teachers was that the introduction of the gadgets had constant challenge to deliver adequate training, especially in
increased the workload in certain areas and decreased it in remote areas. The teachers requested opportunity to ask
others: questions and provide feedback, requiring effective
There are two things, lesson preparation - for this it might communication channels between the stakeholders. As
take more effort - and lesson delivery - for this it might take highlighted by one official:
less time. It has made us do more work – we have to guide the If you come back frequently then the schools will know you
students. And there are not enough gadgets – so if we had are coming and they will keep using the gadgets but if they
enough gadgets then it would be ok and then it would not think no one is coming to visit them then they will forget to use
increase our workload. them. This is because when they are monitored they will feel
Since the intervention was a pilot project the majority of that somebody is appreciating their job – they are happy that
schools chose to use the gadgets after the school day had they can be involved and they see that somebody cares.
finished so as not to interfere with the daily timetable. This The feedback from teachers demonstrated that sustained
meant that the teachers were required to stay in school for an monitoring from MoEST officials constituted a significant
extra hour every day to facilitate the lessons. Most were capacity building process. The impact was especially
willing to do this but some expressed reluctance due to pre- pronounced in marginalized rural areas which were
existing after-school commitments. It was also commented in inaccessible and rarely visited. In light of this is it clear that a
two schools that the initiative resulted in more preparation continuing focus on teacher training which both develops
work for the teachers because they needed to become fully skills and instills value is intrinsic to the success of this and
familiar with the devices themselves before making use of any ICT for education initiative. The ability of the teachers to
them in lessons with the children. In the two schools where utilize technology with confidence and adapt it to the specific
the teachers had chosen to incorporate the gadgets into the needs of the classroom is a central determining factor.
school day then a decrease in workload was reported. In these Without this integration any initiative remains an appealing
cases the gadgets helped with effective facilitation, the add-on but does not have significant effect on the culture of
management of large classes and provided support in the classroom, pedagogy, or rationale for learning.
preparing teaching and learning materials for the lessons.
In light of this, the program would benefit from the
C. Effectiveness of the Technology
development and distribution of a pedagogical guide for
teachers regarding the effective integration of the learning The success of the initiative, whilst not determined by the
machines into the learning routine of the children. Teachers technology, was dependent upon it operating effectively
should also be encouraged to utilize the full range of lessons within each school. Usability, durability and the reliability of
that are available on the devices. This would complement the the devices and charging solution were the three areas
development of a schedule for teachers to use the machines assessed in the study.
during school time, giving clear explanation how it can most The lesson observations demonstrated that the majority of
appropriately support the curriculum. children were able to operate the learning machines
The increased ability of the children to memorize what had independently, understanding the purpose of the buttons and
been taught was also recognized to make the job easier by one completing quizzes without supervision. The process of lesson
teacher from Standard 4: selection occasionally required guidance from teachers but
Gadgets have simplified the work for the teachers – when this was rarely necessary because the most capable children
the child sees something he remembers – when he just hears were seen to take a lead in the learning groups, demonstrating
he forgets. When they see it, it stays in their mind. considerable confidence with the devices. However, a
Others noted that the impact of the learning machines in significant factor limiting usability was the low number of
encouraging so many children to return to school had caused devices per classroom and subsequent high number of
class sizes to expand significantly, resulting in more work students per group, on occasion as many as ten learners. This
attempting to manage the children. A common concluding was not conducive to learning as the machine is designed for
comment from teachers was that the workload would decrease small groups, with small screen and quiet audio output. In
significantly if there was a lower pupil to gadget ratio and if such situations, dominant children were observed bringing the
the lessons on the gadgets were fully compatible with the machine to their ear in order to hear the lesson, meaning other
syllabus and incorporated into the curriculum. The program children were unable to engage and became easily distracted.
would therefore benefit from an increased allocation of 20 This was confirmed by a participating official from the
learning machines per school. This would facilitate a more MoEST who noted:
concentrated impact and allow a greater number of children in Bringing such technology is very important - it is only
smaller groups to benefit from the content on the devices. that at the moment the gadgets are very few – there needs to
In light of the challenges presented, it is clear that regular be more. It is important that you have the right number of
training sessions for the teachers would assist in maintaining gadgets for the number of children because it is difficult for so
many children to share. for the development of user interaction. Market influences on
Due to the nature of such a pilot program it was anticipated the development of mobile technology also ensure sustained
that there would be significant problems regarding device improvements in hardware capabilities and the subsequent
durability. However, after six months of usage only 8% had emergence of new educational possibilities.
developed problems, the majority being due to software issues However, within all such initiatives, it is necessary that the
easily resolved through reformatting. Two of the devices application of technology is conceptualized as a tool in
stopped working due to the on/off switch breaking, these were facilitating the overall aim of catalyzing a more fully effective
repaired in situ and left fully functioning. A major concern in approach to education. In reaching this goal many of the
this respect was that teachers often assumed they had broken challenges documented remain the same regardless of the
the devices when all that was actually required was for them presence of technology.
to be charged fully. It is not overly surprising that the introduction of ICT into
The solar panels operated effectively in each of the five test primary schools in Malawi caused a dramatic increase in
schools throughout the program with all of them maintaining attendance figures. The significant question for ongoing
adequate power to charge the devices as often as necessary. research remains one of long term impact on approach to
Several program schools outside the five included in the test schooling, once initial enthusiasm surrounding the program
did report problems including two faulty batteries, one loose has subsided. Whilst a significant achievement, increasing
connection and one stolen panel. In addition to these, one school attendance figures is not the most significant long-term
school adopted the practice of disconnecting the battery from challenge facing education in Africa [5]. Instead, it should
the panel to remove it for safe keeping during the night and constitute an initial building block which can be capitalized on
this prevented the battery from charging fully. 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
VI. CONCLUSIONS technology, such as the Interactive Learning Program. For full
The analysis of impact on students, teachers and benefits to be realized, such programs require transition from
technological effectiveness has demonstrated the educational their position of peripheral curiosity to one of being integrated
potential of the Interactive Learning Program. The as a sustainable teaching tool. Future research is required to
introduction of the devices had significant positive impact assess the degree to which this is occurring in Malawi.
upon school attendance and levels of enthusiasm. Also A variety of different solutions are required to address the
noteworthy was the increased value placed on teachers and the challenge of providing equitable access to good quality
role of education within the community. However, alongside education in the developing world. The increasing availability
this, considerable challenges were encountered including the and decreasing cost of portable devices ensures that they are
negative impact on teacher workload, lack of classroom likely to play a prominent role in the future, however this
integration, significant infrastructural constraints and the should not lead to such technologies being presented as an
necessity for further teacher training. The future success, educational panacea. For programs to be of maximum
sustainability and potential for project scalability is dependent educational benefit in Malawi, as in many such initiatives, the
upon engaging with each of these identified challenges. The critical issues for consideration and action remain pedagogy,
observations made in the analysis speak directly to the classroom integration and teacher training.
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 REFERENCES
a complex procedure with considerable potential for failure [1] MoEST, Link for Education Governance. Education Sector Performance
Analysis. Lilongwe, Malawi. 2007.
and therefore it is vital that rigorous monitoring and [2] World Bank, Data and statistics: Malawi. 2006.
evaluation is incorporated throughout all such initiatives. Accessed 21/08/08
The price of portable educational technology is constantly [3] MoEST, Primary Curriculum and Assessment Reform (PCAR).
Implementation Plan 2007-2010, Lilongwe, Malawi. 2008.
decreasing, making initiatives such as the Malawi Interactive
[4] UNESCO, Institute for Statistics, country data 2008.
Learning Program increasingly cost effective. The September Accessed 21/08/08
2008 price of each learning machine was circa US$55 and as [5] EFA, Global Monitoring Report: The quality imperative. 2005.
each device can be used by several children the per child cost Accessed 21/08/08
[6] D. Pye and J. Stephenson, Using ICT to Increase the Effectiveness of
is considerably lower than other widely publicized solutions. Community-based, Non-formal Education for Rural People in Sub-
Significant future potential also lies in the opportunity for Saharan Africa: the CERP Project Final Report (DFID Educational
integrating the educational content with new generation Paper 50). London: Department for International Development 2003.
[7] V. Tinio, ICT in Education. E-Primers for UNDP, New York 2003.
mobile phones and other handheld devices becoming [8] J. Leach, DEEP IMPACT: an investigation of the use of information and
progressively more available across the region. Linked to this, communication technologies for teacher education in the global south.
increased ability to send and receive data in standard formats Researching the Issues, 58. DFID 2005.
[9] D. Wagner, Monitoring and evaluation of ICT for Education: An
(for example Macromedia Flash, xml, mp3 or mp4) will allow Introduction. Chapter 1 in: D. Wagner et. al. 2005. Monitoring and
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,
[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. 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,
[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 fraud and transformed supply-chain management for the E-
project that subsidizes healthcare by reimbursing health service choupal project [6]. While health information is critical to the
providers in Uganda for treatment of patients with sexually improvement of healthcare in developing regions, financing
transmitted infections. As with many development projects, the
Uganda Output-Based Aid (OBA) project involves a number of healthcare also remains a significant unsolved problem. Can
stakeholders: the service providers, the project implementers, we take lessons from e-Choupal and apply them in the
the financiers, and the Ugandan government. Design of an healthcare sector? The design of usable, reliable, and fraud-
appropriate solution requires meeting the various and conflicting resistant tools for management of these aid flows is an area
requirements of all of these stakeholders. In this paper we detail with potential for very significant impact.
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 However, in the case of healthcare, the financial models are
device study, semi-structured interviews, health clinic surveys,
and a brief deployment, we affirm the selection of the mobile very different from commercial markets – financing of health-
phone as a platform from the health clinic perspective, and care typically comes through transnational aid agencies like
further suggest that effective design for development requires the World Bank and International Monetary Fund (IMF), and is
more than addressing requirements of the the “users” of the implemented by non-governmental organizations (NGOs) and
mobile phones but also all the other stakeholders involved, who the local government. Since the NGOs are typically experts in
may have conflicting requirements.
Index Terms—mobile phone, ICTD, health, participatory de-
health, not technology, data processing is often outsourced to
sign, Africa, HCI 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
Mobile phones are frequently touted as being the appro- information flows between them are scattered at best.
priate and sustainable platform for rural healthcare in Africa.
They are relatively cheap, durable, consume less power than
In this paper we suggest that the “closed loop model”
laptops and desktops, and incorporate a battery that makes
generally used by researchers in deployments of mobile health
them more amenable to use in places with intermittent or no
applications does not map onto the financial and political
power. Commonly proposed uses are for data collection [1],
realities of the mainstream of healthcare provision in Africa,
[2] and decision support for rural health workers [3], [4]. Some
and limits the ability of pilot programs to increase their scale
projects also use mobile devices as a platform for information
and impact. We describe an innovative, IT-based, NGO-run
dissemation as well as data gathering [5]. However, these are
healthcare access program in Uganda, and our experiences
all generally “closed loop” systems in which researchers are
designing and deploying Claim Mobile, a mobile-phone based
able to control all aspects of the system design and operation,
system intended to address inefficiencies and help the program
focusing their research primarily on the rural health workers
scale to additional districts. We argue that in addition to
that will be using the mobile phones.
addressing the needs of the primary users in the system, the
Other applications have even more potential for large-scale health workers, our design must consider the requirements,
impact. In the agricultural sector, we have observed how motivations and concerns of the other stakeholders: the IT
the introduction of transparent market prices and subsequent vendors, the NGOs, the government, and the aid agencies.
hiring of “middlemen” to purchase from farmers has reduced Our designs must consider the larger order ramifications of
how we may positively and negatively impact both the “users”
Manuscript received October 1, 2009. This work was supported by the
Blum Center for Developing Economies and the U.S. National Science who will be generating the data, and the entities that will be
Foundation Summer Undergraduate Program in Engineering Research at engaged in managing and using the information in the resulting
Berkeley (SUPERB) under Grant No. 0453604. database. Just as the e-Choupal project assimilated the mid-
Melissa R. Ho is with the School of Information at the University
of California, Berkeley, 94720, USA. (phone: +256 777 723 786; email: dlemen by hiring them as kiosk operators, we propose that we can design applications structured to accommodate conflicting
Emmanuel K. Owusu is with the Computer Engineering Department at stakeholder requirements, while also alleviating information
Iowa State University, 50011, USA. (email:
Paul M. Aoki is with Intel Research, Berkeley, CA, 94704, USA (email: inequalities resulting from limitations in the system prior to the introduction of the information technology.

Mbarara, Uganda
• HIV prevalence: 10% of adult population (15-49
• 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.
Fig. 1. A HealthyLife voucher. The ‘M’ in the top left is a note written TABLE I
by the health service provider to indicate that the first client voucher on the S OME BACKGROUND STI STATISTICS ABOUT M BARARA , U GANDA [7].
left was given to a male client, and that therefore the partner vouchers on the
right should be given to a female client.

can take two weeks or more to move from the providers office
II. BACKGROUND to the management agency. The current data management
system requires all claims to be submitted on paper forms
Providing effective health care in poor countries is an to the management agency. At least another two to four
essential component to economic development and poverty weeks are spent reviewing each claim, cleaning data from
reduction. Unfortunately donors supporting this endeavor often improperly-completed forms, and verifying that the service
find that resources given are not matched by desired gains. took place among suspect claims. Two months or more can
The output-based aid (OBA) model of financing seeks to ad- go by before the provider is reimbursed for service provision.
dress this by paying healthcare providers directly for services In Uganda, private providers traditionally operate on a fee-
rendered instead of paying for the service provision up front. for-service model, receive prompt payment, and do not have
However, OBA program management is information intensive, a large operating margin. In many cases, payment is provided
necessitating much paperwork to track and reimburse payment prior to service. Delays in payment result in delays in pro-
claims. Smartphones (mobile phones with advanced features curement of replacement prescriptions and medical supplies,
such as the ability to run third-party software) have the often leading to a temporary hiatus in service. Encouraging
potential to alleviate this burden. In collaboration with a local provider involvement in the OBA program requires a great
NGO and their partnering IT vendor, we have proposed to deal of confidence on the part of the providers to participate. If
deploy a number of smartphones for use in an OBA project a system to shorten claims processing could be devised, more
based in Western Uganda, with dual goals of reducing claim providers could join the scheme and more patients could be
processing time and improving communication between the provided the life-saving STI treatment voucher subsidy.
health care providers and the OBA management agency. The remainder of this paper details the system we are cur-
The project is managed by the local branch of a multi- rently piloting, in which claims are submitted via Internet from
national NGO and a for-profit health insurance company, in a mobile phone directly to all the parties in the management
collaboration with the Ugandan Ministry of Health (MoH) agency. In addition to describing our user studies and how this
and Ministry of Finance (MoF). The project is primarily has informed the design of the system, we discuss the problem
funded by an aid agency based in Europe, with additional of negotiating conflicting stakeholder requirements. We find
funding for the expansion coming from a separate transna- that in projects with multiple stakeholders, the introduction of
tional funding agency. Together, they run a voucher program a system may disrupt balances of power, particularly around
called HealthyLife, which treats sexually transmitted infec- the flow of information and money. As a result, the design of
tions (STIs), reimbursing providers for the diagnosis and full this system, in order to secure positive support from all parties
course of treatment only after the patient is seen. This program involved, must carefully balance stakeholder incentives.
was implemented in response to the high burden of sexually
transmitted infections in Uganda, and began in July 2006 III. M ETHODS
in four districts of southwestern Uganda: Mbarara, Ibanda, The research described here involved an iterative process of
Kirihura and Isingiro (See Table I). field research and prototyping. The fieldwork and deployments
Patients buy treatment vouchers in pairs, one for the client have been done over the course of three visits to Uganda:
and a second one for the client’s sexual partner (See Figure 1). an initial two-week visit in Summer 2007 to establish a
Each voucher is good for one consultation (generally including relationship with the project, in which we also conducted a
a lab test to diagnose the STI) and three follow-up visits. survey of the clinics in the program; a followup visit for
During the consultation, the provider completes a claim form three weeks in November 2007; and a five-week pre-pilot
recording the client’s demographics, the examination and deployment in August-September 2008. During all three visits
laboratory results, a diagnosis and details of the course of we conducted semi-structured interviews with the various
treatment prescribed (See Figure 5). Completed claims forms stakeholders, and directly observed claims form entry and
are sent to the voucher management office in the city of processing. When given permission, we did audio and video
Mbarara, the main urban center of Western Uganda. Forms 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. 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
We asked 14 questions, assessing familiarity with computers once the Claim Mobile system is fully deployed.
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 be conducted in May 2009, with mobile phone-based claims
might be for improvement of the process). We also collected processing expanded to 8-10 additional clinics.
various documents from the management agencies regarding
the performance of each clinic, including all available financial IV. S TAKEHOLDERS
reports on processed claims, and in November, we returned to In this section we describe the funding, management and
seven of the clinics to do in-depth surveys and to follow-up service provider organizations to which we alluded in Sec-
on the survey findings. tion II. Fundamentally, all stakeholders want to improve STI
treatment and reduce the prevalence of STIs. Each stakeholder
B. Rapid Development and Pre-pilot Deployment also has a financial interest in the success of the overall project
Initial prototyping occurred in early 2008, and we returned - staying afloat for the health service providers, and staying
to Uganda in Summer 2008 to do a three-stage pilot deploy- within the aid agencies’ target budgets for the management
ment, first testing the functionality of our software, second agency partners. The discussion highlights the ways in which
reviewing the proposed claims process with the management the various stakeholders have competing as well as common
agencies, and finally taking the phones to the health clinics to interests. While we will detail several stakeholders in this
test the mobile phone interface in the field. During this time section, there are six key stakeholders: the aid agencies who
we also conducted another round of semi-structured interviews fund the OBA project, the financial management agency
to gather information on changes in the claims submission (FMA) which receives the funds from the government and
process (for example, claims processing had moved from disburses them, their program management office (PMO) in
Mbarara to the national capital, Kampala). We did iterative Mbarara which runs the program and interacts directly with
development based on feedback from the various stakeholders, the health service providers, the technical management agency
trying out features as they were suggested, and developing (TMA) that manages the claims processing, the health service
new tools as seemed merited by findings in our interviews. providers (HSPs), and us, the mobile platform developers
To gain a more in-depth understanding of health clinic life, (MPDs). Figure 2 illustrates some of the relationships between
we stayed overnight for three days in the rural health clinic, these entities which we will describe in detail in the remainder
thereby supplementing the the semi-structured interviews with of this section, based on qualitative fieldwork and document
direct observation of actual practice. analysis.
The primary purpose of this last field visit was to con-
duct a pre-pilot demonstration, using the mobile phones to A. Aid Agencies
submit actual claim data to the management agency, have it As the funder of the HealthyLife STI treatment program,
reviewed, and have the management agency provide feedback the involvement of the European aid agency is more than
to the health clinics via the mobile phones. We simulated apparent. Their role in the management of the program is
the proposed process, physically following the claims forms more supervisory – a consultant goes to Uganda at irregular
from the time the patient comes into the health clinic, through intervals to help with planning of the program, and they do
the preparation of the claims summary forms, physically some monitoring. They also have commissioned another non-
transporting the forms to the management agency where we profit, affiliated with a North American university, to conduct
observed the claims approval, and data entry into the existing an evaluation of the program. Ultimately, however, they control
database. We simultaneously had the service providers submit the flow of money to the financial management agency, which
the claims form via Claim Mobile, enabling the management then pays the IT vendor to handle the technical aspects of the
agency to provide feedback to the service providers through operation.
the system. The pre-pilot is still operational, with mobile In the past year, the European aid agency has worked with
phones remaining in the two clinics, and the full pilot will an additional transnational aid agency to fund the expansion

of the project into additional districts. While they may not for the PMO, and one for the TMA. However, the copy that
have a direct impact on the information processes in the remains in the PMO does not have the voucher number, a
project itself, the funders’ internal actions have direct impact critical piece of information, and with stacks of hundreds of
on the project as a whole. In one example, a delay in claims per month, the information is not in a format actually
payment to the European aid agency resulted in a delay in accessible to the program office until the TMA sends back
payment to the two management agencies. As a result the IT