Virtual Inequality Beyond The Digital Divide

VIRTUAL
Karen Mossberger
Caroline J. Tolbert
Mary Stansbury
-^ » 10 o-^
H'ESSS
cko& 3>
VIRTUAL INEQUALITY
. RARY
<
<0 .
Cj?orr,
\
American Governance and Public Policy series *
Series Editor: Barry Rabe, University of Michigan
After Disaster: Agenda Setting, Public Policy, and Focusing Events

Thomas Birkland
Budgeting Entitlements: The Politics of Food Stamps

Ronald F. King
Collaborative Public Management: New Strategies for Local Governments

Robert Agranoff and Michael McGuire
Conflict Amid Consensus in American Trade Policy

Martha L. Gibson
Controlling Technocracy: Citizen Rationality and the NIMBY Syndrome

Gregory E. McAvoy
Dam Politics: Restoring America’s Rivers

William R. Lowry
Expert Advice for Policy Choice: Analysis and Discourse

Duncan MacRae, Jr., and Dale Whittington
Federalism and Environmental Policy: Trust and the Politics of Implementation

Denise Scheberle
Federalism in the Forest: National versus State Natural Resource Policy

Tomas M. Koontz
Fenced Off: The Suburbanization of American Politics

Juliet F. Gainsborough
From Revenue Sharing to Deficit Sharing: General Revenue Sharing and Cities
Bruce A. Wallin
The Government Taketh Away: The Politics of Pain in the United States
and Canada
Leslie A. Pal and R. Kent Weaver, Editors
Green Politics and Global Trade: NAFTA and the Future

of Environmental Politics
John J. Audley
Hazardous Waste Siting and Democratic Choice

Don Munton, Editor
How Governments Privatize: The Politics of Divestment
in the United States and Germany
Mark Cassell
Improving Governance: A New Logic for Empirical Research

Laurence E. Lynn, Jr., Carolyn J. Heinrich, and Carolyn J. Hill
Justice & Nature: Kantian Philosophy, Environmental Policy, and the Law
John Martin Gillroy
Lobbying Together: Interest Group Coalitions in Legislative Politics

Kevin W. Hula
Pluralism by the Rules: Conflict and Cooperation in Environmental Regulation

Edward P. Weber
Policy Entrepreneurs and School Choice

Michael Mintrom
The Political Economy of Special-Purpose Government

Kathryn A. Foster
The Politics of Automobile Insurance Reform: Ideas, Institutions,

and Public Policy in North America
Edward L. Lascher, Jr.
The Politics of Ideas and the Spread of Enterprise Zones

Karen Mossberger
The Politics of Unfunded Mandates: Whither Federalism''

Paul L. Posner
Preserving Public Lands for the Future: The Politics of Intergenerational Goods
William R. Lowry
Rethinking Health Care Policy: The New Politics of State Regulation

Robert B. Hackey
Taking Aim: Target Populations and the Wars on AIDS and Drugs
Mark C. Donovan
Taking the Initiative: Leadership Agendas in Congress and the

“Contract with America”
John B. Bader
Virtual Inequality: Beyond the Digital Divide

Karen Mossberger, Caroline J. Tolbert, and Mary Stansbury
Welfare Policymaking in the States: The Devil in Devolution

Pamela Winston
V
VIRTUAL liEQUALITY
Beyond the Digital Divide
KAREN MOSSBERGER
CAROLINE J. TOLBERT
MARY STANSBURY
Georgetown University Press

Washington, D.C.
Georgetown University Press, Washington, D.C.
© 2003 by Georgetown University Press. All rights reserved.
Printed in the United States of America
10 987654321 2003
This book is printed on acid-free recycled paper meeting the requirements

of the American National Standard for Permanence in Paper for Printed
Library Materials.
Library of Congress Cataloging-in-Publication Data
Mossberger, Karen.
Virtual inequality : beyond the digital divide / Karen Mossberger, Caroline J.
Tolbert, Mary Stansbury.
p. cm. — (American governance and public policy series)
Includes bibliographical reference and index.
ISBN 0-87840-999-8 (pbk. : alk. paper)
1. Digital divide. I. Tolbert, Caroline J. II. Stansbury, Mary
Catherine, 1957— III. Title. IV. American governance and public policy.
HN49 .I56M67 2003
303 .48'33—dc21
2003004571
Contents
List of Tables and Figure IX
Preface xi
Acknowledgments xv
Chapter I
Redefining the Digital Divide
Chapter 2
The Access Divide 15
Chapter 3
The Skills Divide 38
Chapter 4
The Economic Opportunity Divide 60
Chapter 5
The Democratic Divide 86
Chapter 6
Beyond the Divides: Toward Opportunity and Equity 116
Appendix I
Multivariate Regression Tables 141
Appendix 2
Survey Questionnaire 155
Glossary 169
References 171
185
Index
List of Tables and Figure
Tables
2.1 Measuring the Access Divide 33
Tables listed below are in Appendix 1.
A2.2 Respondent Has Internet Access 141
A2.3 The Access Divide: Respondent Has Access

to Information Technology 142
Technical Competence and Information Literacy 143

A3.1
Computer Skills 144

A3.2
Attitudes about Library Use 143

A3.3
Instructional Preferences 146

A3.4
A4.1 Attitudes about Computers and Economic

Opportunity 147
A4.2 Support Use of Internet for Employment

and Education 148
A4.3 Support Use of Internet for Employment and

Education in a Public Place 149
A4.4 Respondents’ Internet Use for Employment

150
and Education
Support for Digital Democracy and E-Government 151

A5.1
\
X List ofTables and Figure
A5.2 Support for Digital Democracy vs. Actual Experience f52
A6.1 Digital Experience 153
A6.2 Support for Vouchers 154
Figure
6.1 Distribution of Digital Experience in the United States 126

Preface
T he issue of the “digital divide”—or disparities in information

technology based on demographic factors such as race, ethnic¬
ity, income, education, and gender—captured headlines as the
Internet made otherwise steady progress in permeating American
society. Casting a shadow over the country’s newfound fascination
with the information “superhighway” were persistent reports that
the poor, minorities, and others remained disconnected. The num¬
bers contained in the reports were troubling indicators of a potential
problem, but they said little about the causes or consequences of the
problem and the possible remedies. For example, was the problem
affordability or the ability to learn how to use the technology? Was it
the ability to find and use the information on the Internet? Was it
a lack of awareness of the possible uses and benefits of information
technology? And what difference did it make anyway, even if some
people never use computers and the Internet—would they and soci¬
ety be appreciably worse off? Much of the prior research has focused
less on these questions than on counting the number of people
who have access to technology at home. This is a first step—as Deb¬
orah Stone has said, counting is a political act and raises awareness of
an issue—but it does little to inform debate or to offer guidance to
policymakers.1
To us, there was a larger story to be told, and through telling the
story we could better define the problem and appropriately target
public resources. The real story of the digital divide could be found
by understanding more about the experiences, attitudes, and needs of
the individuals caught in the gap. A survey that focused on low-
income communities and included a number of minority individu¬
als would enable those most affected to tell their own story. By
thinking about why disparities in information technology are a pol¬
icy issue—about the possible consequences for society and for nor¬
mative values such as equality of opportunity we could also begin
xii Preface
to move “beyond the digital divide” as an issue of access and to pro¬

voke a fuller public discussion about what the aims and obligations
of public policy should be.
Our interdisciplinary team of three Kent State University re¬
searchers brought different perspectives to the topic. Mary Stansbury
is a professor of library and information science, and her previous
research on underserved populations sensitized her to this issue early
on, particularly because public libraries became a focal point for
public access to technology. Mary sought out faculty from the uni¬
versity’s graduate program in public policy to carry out collaborative
research on the issue and found two political scientists eager to take
up the challenge—Caroline Tolbert, who had conducted research on
issues such as race and ethnicity and direct democracy, and Karen
Mossberger, who specialized in urban policy. Together we obtained
research support from a national foundation and the Ohio Board of
Regents, in order to conduct a national, random-sample telephone
survey that included a separate sample drawn from low-income
communities.
The results of that survey form the basis of this book, but we hope
that what we present is much more than an aggregation of data. Our
aim is to push the boundaries of the policy debate and make the infor¬
mation accessible to a wide audience.
We hope to influence the debate on this issue by redefining it as
multidimensional—through consideration of an access divide, a skills
divide, an economic opportunity divide, and a democratic divide.
Although Pippa Norris’s excellent cross-national study, in which she
uses the term the “democratic divide,” differs in many ways from
ours, we share her concern that the relationship between technology
disparities and public objectives merit more discussion.2 We argue
that public policy needs to address issues of skill as well as access and
be focused on the way in which access and skill influence economic
opportunity and political participation, now and in the future.
Our research is presented in a way that requires no knowledge of
statistical methods, even though the study has been conducted
according to academic standards and uses more sophisticated analy¬
sis than most of the prior studies on technology access. We hope to
reach policymakers, interested citizens, and all sorts of thoughtful
individuals who are engaged in this issue but who may not have for¬
mal training in statistics and research methods. Throughout the text,
Preface xiii
we use “What Matters” boxes to highlight the results that are statis¬
tically significant and to keep track of what might be an otherwise
confusing array of outcomes on a number of questions. We list “what
matters” without the numbers or with simple probabilities. Behind
these simple comparisons, however, are multivariate regression analy¬
ses that statistically control for the effect of multiple influences, and
we offer the regression tables in the appendix for those who are inter¬
ested in examining our statistical findings in more detail. We discuss
our methods to some extent within the text, because we feel that this
is an important advantage of the research presented here. In the inter¬
est of reaching as broad an audience as possible, though, we carefully
explain our approach in clear language and leave the more detailed
and technical discussions for the endnotes. In short, we hope to
change the landscape of policy debate by moving beyond the limita¬
tions of current research and beyond the definition of the problem as
one primarily based on access.
Notes
1. Stone 2002, 176.

2. Norris 2001.
'
Acknowledgments
A project such as this one—a book based on a national survey—

is inevitably the product of many people, not just the authors.
We are indebted to our sponsors at a national foundation
(which makes it a policy to withhold publicity for all grants). The
challenge grant we received from the Ohio Board of Regents provided
crucial funding, allowing us to proceed without delay. The Com¬
puter-Assisted Telephone Interviewing Laboratory in Kent State’s
Department of Sociology provided trained interviewers, supervision,
and consultation for the survey. Tom Hensley, chair of the political
science department, supported this project in many ways beyond the
call of duty. We also thank Rick Rubin, interim dean for the College
of Communications, Joseph Danks, dean of the College of Arts and
Sciences, and Richard Worthington, dean of the College of Fine and
Professional Arts.
We have noted on several chapters special assistance from two of
our graduate research assistants—Ramona McNeal and Lisa Dotter-
weich, who are doctoral students in the public policy program in Kent
State University’s Department of Political Science. Carrie Hribar, also
a public policy graduate student, and Tinnie Banks, of the College of
Education, provided valuable research assistance as well.
A number of colleagues and other observers offered comments on
initial data, arguments, or early chapters presented at conferences
such as the American Library Association, American Political Science
Association, Midwest Political Science Association, Western Political
Science Association, American Society for Information Science and
Technology, and Association of Internet Researchers. Among those
we thank for their insights are Larry Ledebur, Genie Stowers, Paul
Baker, Rodney Hero, Hal Wolman, Susan Clarke, Donald Norris,
and two anonymous reviewers. We tried to follow your suggestions,
and we hope that it shows. Barry Rabe and Gail Grella, of George¬
town University Press, gave us sage advice throughout this process,
xvi Acknowledgments
and we are grateful to be working with a publishing team of George¬

town’s caliber.
We appreciate the efforts of the more than 1,800 people who
answered our survey questions, in an age when telemarketers and oth¬
ers mercilessly clog the telephone lines. We thank the government
officials and community activists in Berkeley, California, who took
time to welcome us to the city and answer our questions.
Finally, we thank those who are closest to us for their support over
the past two years, especially our spouses and our children (who range
from ages one to twenty-two). One theme that emerges from this
book is the role that young people will play in developing the social
potential of information technology. In that spirit, we dedicate this
book to our children: Heather, Lauren, Jacqueline, Eveline, Edward,
Amanda, Elizabeth, and Thomas.
Chapter I
Redefining the Digital Divide
I nformation technology is now an integral part of the workplace

and the home for many Americans. More than half of American
households now use the Internet, and two-thirds own home com¬
puters. Their presence transforms the way in which we work, play,
socialize, and discover and disseminate information.
While the use of information technology is growing rapidly, some
segments of society remain largely disconnected from this trend. The
term “digital divide” has been used to describe the patterns of unequal
access to information technology based on income, race, ethnicity,
gender, age, and geography that surfaced during the mid-1990s. Data
showing that computer ownership and Internet access are lower
among certain groups of the population have generated both a spate
of public and private initiatives and a chorus of critics who dismiss
the divide as either a myth or of little real consequence. Heated con¬
troversy currently surrounds federal programs addressing the digital
divide, as congressional proponents of these efforts battle attempts by
the Bush administration to eliminate them.
We aim to contribute to these debates and to public policy in this
area by providing a more accurate assessment of the problem and more
information about the needs, attitudes, and experiences of the low-
income and minority communities that have been the focus of current
initiatives. The research presented here is distinctive in two ways.
First, we advocate moving beyond the narrow boundaries of the
way in which the digital divide is currently defined. The preponder¬
ance of programs, debate, and research has been restricted to the
problem of access to technology. But having access to a computer is
insufficient if individuals lack the skills they need to take advantage
of technology. Skills development has taken a backseat to the provi¬
sion of wiring and hardware in most programs, and there is scant
\
2 Virtual Inequality
research evidence on skills. Moreover, little is known about the expe¬

riences or attitudes of disadvantaged groups in relation to the public
objectives that support government intervention—economic oppor¬
tunity and political participation. We propose a broader definition of
the problem as consisting of multiple divides: an access divide, a skills
divide, an economic opportunity divide, and a democratic divide.1
Second, our research has several advantages over existing studies
because of its breadth, its low-income sample, and its methods of
analysis. As suggested by the multiple divides, the survey on which
this book is based includes a wide-ranging set of policy concerns
beyond access. In order to paint an accurate picture of the attitudes
and experiences of disadvantaged groups, we surveyed a random sam¬
ple of respondents from high-poverty census tracts, as well as a more
general sample for comparison. Finally, nearly all of the existing data
have been reported as simple percentages. These percentages are use¬
ful for understanding general trends, but they cannot sort out the rel¬
ative significance of different factors in the same way that statistical
methods such as multivariate regression can. Using these methods,
we can distinguish the causes of technology disparities, or the differ¬
ences that matter. Nevertheless, we present the results of our study in
a way that requires no statistical background, as explained later in this
chapter. Together, our research and analysis allow us to explore new
ground in the multiple divides and to test the validity of previous con¬
clusions about access and other issues that were not based on multi¬
variate statistics.
In this chapter, we trace the evolution of the “digital divide” as a
policy issue, explain our approach and research methods, and offer an
overview of the chapters that follow, particularly the different objec¬
tives that we set out to achieve in each.
Issue Evolution: The Access Divide and Beyond
Once the province of scientists, the Internet’s potential as mass

medium was unleashed by the development of a graphically based
web browser in 1994, which made use of the web possible through a
few simple clicks of a mouse. In 1995, Vice President A1 Gore
announced the development of a National Information Infrastruc¬
ture—the information superhighway—as a priority of the Clinton
Redefining the Digital Divide 3
administration. Also during 1995, the first of several studies under¬

taken by the National Telecommunications and Information Admin¬
istration (NTLA) appeared. The report, Falling through the Net: A
Survey of the “Have Nots” in Rural and Urban America, revealed
inequities in access to personal computers and the Internet.2 Subse¬
quently, the Telecommunications Act of 1996 extended telecommu¬
nications policy beyond traditional concerns with phone service to
include new digital media. This represented the first policy response
to the disparities featured in the NTLA report and also established this
issue as a problem of access, akin to telephone service, which has been
cross-subsidized in rural regions by national taxes on telephone util¬
ities. As part of this act, the E-Rate program, funded by taxes on local
and long distance telephone carriers, provides schools and libraries in
poor communities with discounted rates for Internet access, high¬
speed data connections, phone service, and wiring.3
Later iterations of the NTLA study have tracked changes in com¬
puter and Internet access over the ensuing years. The exact origin of
the term “digital divide” is unclear, but it became popular after NTLA
used the phrase to describe disparities in access in its 1998 report.4
As discussed in the next chapter, home access has continued to be a
central concern in research, although later studies also examined
access in various places.5
Access concerns predominate in programs as well as research
addressing the technologically disadvantaged. The E-Rate is the
largest federal program, with total spending of about $2.25 billion in
fiscal year (FY) 2001 for technology infrastructure. Two other federal
programs, the Technologies Opportunities Program (TOP) and the
Community Technology Centers (CTC) assist communities with
other needs, such as hardware, software, content development, and
training. Federal resources committed to skills development are
decidedly more modest than infrastructure investment. The TOP
supports pilot projects, and the CTC program funds about 400 cen¬
ters nationwide. Together, their spending totaled only $110 million
in FY 2001, or about 5 percent of the amount expended on E-Rate.
Federal efforts are supplemented by private initiatives. That the
issue of the digital divide has captured the imagination of many is evi¬
dent in the programs funded by foundations, corporations, and non¬
profits. The Digital Divide Network lists approximately 20,000
initiatives in its directory, but many of these programs are small-scale,
\
with a few notable exceptions, such as the Gates Foundation’s assis¬

tance for public libraries.6
Federal policy still constitutes a mainstay of funding in this policy
area and is now the focus of controversy over future policy direction.
Existing policy supports public access in schools, libraries, and com¬
munity centers, but proposals to partially subsidize the purchase of
computers or Internet services for low-income farriilies have also cir¬
culated in Washington in the past.7 To date, however, home access
has been supported primarily through small nonprofit programs that
offer loans and used computers to low-income families.
More recently, conflict has erupted on Capitol Hill over the fate of
the CTC and TOP programs, which the Bush administration desig¬
nated for termination. Clashes over federal policy revolve around two
issues: whether disparities will disappear of their own accord, in light
of the rapid diffusion of the Internet, and whether government inter¬
vention is appropriate, regardless of continued inequities. The 2002
installment of the NTLA reports takes a decidedly optimistic turn,
entitling its findings “A Nation Online,” in contrast to the “digital
divide” of a few years ago.8 Contained within the report, however, are
data that show that rapid diffusion has not yet erased previous dis¬
parities. Will this gap eventually close through the natural course of
events, without the commitment of public revenues? If not, is pro¬
moting digital equity even a legitimate policy goal?
In the chapters that follow, we offer evidence in response to these
questions and contend that, in order to adequately answer them, we
must redefine the “digital divide” beyond the confines of the access
issue. Research and policy debate have been hobbled by a simplistic
view of the factors that contribute to technology use and a paucity of
information about the relevance of these disparities to the public
interest. Access is undeniably important, but the real policy question
is how well society will be able to take advantage of the opportunities
offered by technology.
An issue definition based primarily on access contains an implicit
assumption that the policy problem is affordability rather than abil¬
ity to use technology. Yet certain skills are necessary to exploit the
potential of computers, including the ability to use complex software
programs or to locate and evaluate information on the web.
Even if evidence demonstrates the existence of an access and skills
divide, that is not justification enough for public action. For a con-
dition to qualify as a policy issue rather than a personal concern, there

must be something at stake for the larger society. Current Federal
Communication Commission chief Michael Powell has dismissed the
issue of the digital divide by saying that there is a Mercedes divide as
well.9 If computers and the Internet are just the latest luxury item, or
technological toys, little different from microwaves or DVDs, then
there is sparse justification for public intervention. What defines the
access and skills divides as appropriate issues for public policy are the
uses of information technology. Computers and the Internet are,
among other things, tools for participation in the economy and the
political arena. Technology disparities merit policy attention because
of their implications for important normative issues such as equality
of opportunity and democracy. Information technology also has
“positive externalities,” or spillover benefits.
Information technology skills and access are “public goods,”
because, like education and libraries, they are capable of providing
positive externalities associated with economic growth and demo¬
cratic governance. Economists justify government intervention in the
market when there are externalities, or effects that ripple beyond the
individuals who are directly involved in a transaction. Positive exter¬
nalities mean that the market, left to its own devices, will likely under¬
provide such commodities. Because individuals fail to “capture” all of
the benefits of the knowledge and skills they acquire, they will tend
to undervalue them and underinvest from the point of view of soci¬
ety as a whole. Public subsidy or public provision in such cases is more
efficient than the market, because governments are able to act in the
public interest and to realize the additional social benefits. How then,
does society gain from increasing technology use?
Beyond the economic benefits that accrue to individuals when
they acquire new skills, a community or a nation with a well-educated
workforce is more productive and competitive, particularly within
the context of the knowledge-intensive economy that has developed
in the advanced capitalist countries over the past few decades.
“Human capital”—or “the acquired skills, knowledge, and capacities
of human beings”—is a productive resource, as necessary for pro¬
duction as other inputs such as physical capital.10 Just as widespread
education raises the level of human capital in the economy, so do crit¬
ical technology skills that are increasingly important throughout the
economy.
V
Likewise, there are positive externalities to be found in the politi¬

cal arena. An educated and informed citizenry is more interested in
and knowledgeable about political issues and is thus more likely to
vote, voice opinions, organize, volunteer on behalf of the community,
run for office, and govern. Political knowledge and the skills to par¬
ticipate not only empower individuals, but are also necessary for the
functioning of a democracy. The “public good” provided the primary
rationale for Thomas Jefferson’s advocacy of schools as a central fea¬
ture of a ward system of “little republics”11 and Horace Mann’s cam¬
paign for a system of common schools in the nineteenth century.12
Similar arguments have supported the establishment of public
libraries since Benjamin Franklin. Like schools and libraries, the
Internet has become a resource for political and civic information—
from campaign to government websites. In addition, computers and
the Internet may become increasingly important for access to the
political process, as suggested by recent experiments with Internet
voting and electronic town meetings (see chapter 5).
Beyond the potential for social gain through public support for
information technology access and skills, we believe that there is also
a compelling argument for attention to these issues based on equity.
The idea of equal opportunity (if not equal results) resonates power¬
fully in the American polity, from the Horatio Alger myth to the
struggles of the civil rights movement. Most frequently, this ideal is
evoked in connection with the ability to compete economically, but
equal opportunity also applies to the exercise of the rights of demo¬
cratic citizenship. Universal access to public education has been called
“America’s answer to the European welfare state,” emphasizing prepa¬
ration for economic and political participation rather than equitable
outcomes.1 ^ Although the United States provides fewer redistributive
social programs than most other developed nations, it was the first to
guarantee free public education at both the primary and secondary
levels. To the extent that computers and the Internet offer tools for
economic self-sufficiency and political engagement, the price of
unequal technology skills and access may be to perpetuate or even
widen existing social inequities. Robert Putnam argues that dispari¬
ties in information technology represent a “cyberapartheid” that
diminishes the store of bridging social capital necessary for connect¬
ing disparate elements of society.14
Our goal, then, is to explore these multiple dimensions of dispar¬

ities in information technology—to establish whether in fact signifi¬
cant technology divides exist for access, skills, economic opportunity,
and democratic participation. The next section discusses our research
design, the analytical methods we use, and the advantages of these
approaches.
Methodology
In view of our task of exploring multiple divides, we chose to examine

the skills, attitudes, and experiences of the individuals who were most
likely to be affected by a lack of computer access and skills. Three
features of our study improve upon previous research on the digital
divide. First, our study draws upon a unique source of data, a survey
that features a broader set of questions relevant to all four divides rather
than merely technology access or use. Second, our data are derived from
a larger sample of low-income respondents than most information
technology surveys (other than the very large sample studies used by
the NTIA), which increases confidence in our findings. We drew one
sample from high-poverty areas and another from the general popula¬
tion. Third, we use multiple controls in analyzing the data, which allow
us to sort out which factors account for observed differences and which
are statistically significant. Our findings on access contradict some pre¬
vious studies that have used simpler techniques.15 Although we use
more rigorous methods, we report most results using probabilities, so
that no statistical background is required for readers.
Collecting the Data
Our primary source of data is a national telephone survey conducted

in June and July 2001 by Kent State University’s Computer Assisted
Telephone Interviewing (CATI) lab in the Department of Sociology.
Respondents were at least twenty-one years of age. One national ran¬
dom sample of 1,190 respondents was drawn from all high-poverty
census tracts in the forty-eight contiguous states, excluding Alaska and
Hawaii. High-poverty tracts were defined as those with 50 percent or
more of the households living at or below 150 percent of the federal
poverty level. The response rate for individuals in the high-poverty

tracts was 92 percent. Federal data show that telephone service now
reaches 94 percent of the population, so telephone surveys are a rea¬
sonable methodology for obtaining sample data even in low-income
communities.16 A second national random sample of 655 respon¬
dents drawn without regard to the poverty rate of the census tract
served as a control group and had a response rate of 88 percent. There
were 1,837 valid responses overall.
Telephone numbers were dialed daily through the months of June
and July (with 37 days in the field) by trained interviewers. Up to 524
callbacks were attempted to contact potential respondents for the
general population sample, and 371 were attempted for the low-
income sample. Answering machines were treated as “no answer” and
called back on a regular no-answer rotation, a minimum of three
hours later. After securing the cooperation of interview subjects,
interviewers used CATI systems to administer questions and record
responses. The telephone survey included fifty items and averaged 8.5
minutes to complete. The specific wording of each survey question is
given in appendix 2.
The process of identifying all high poverty census tracts in the
forty-eight states and drawing a separate random sample from them
was laborious, which is why few other national surveys pursue such
an approach. We believed, however, that such an effort was necessary
in this case. Because our survey targeted high-poverty areas, our sam¬
ple included a relatively large proportion of racial and ethnic minori¬
ties, compared to standard surveys. Of the 1,837 respondents, 70
percent were white non-Latino, 19 percent were African American, 9
percent were Latino, and 1.5 percent Asian American. Thus, Latinos
and African Americans comprised 28 percent of the sample popula¬
tion, compared to 25 percent of the U.S. population in the 2000 cen¬
sus. Thirty-eight percent of our sample had household incomes below
$30,000. This allowed us to make accurate inferences to minority and
low-income populations as a whole. The survey generated data that
were comparable to large-sample studies, adding to our confidence in
the validity of our findings. As shown in the next chapter, our figures
closely track the numbers in the large-sample U.S. Department of
Commerce study conducted in August of 2001 on parallel questions
of access and use, providing further evidence that our sample is a rep¬
resentative one.
Measuring the Divides
To measure the multiple divides, we use various survey responses as

the dependent variable or the outcome to be explained.
The access divide. For the access divide, we are interested in access
(whether the individual has home access to a computer, home access
to the Internet, an e-mail account), the location of computer and
Internet use (at home, at work, at the home of a friend or relative, at
a library, etc.), and the frequency of use at each location. This gives
us a more complete picture than home access alone and allows us to
compare patterns of usage by frequency and location.
The skills divide. For the skills divide, we develop two different
indices of skill, one for technical competence and one for informa¬
tion literacy. We also explore preferences for assistance (one-on-one
help, group instruction, online help, and printed manuals) and atti¬
tudes regarding public access such as computer use at libraries or com¬
munity centers.
The economic opportunity divide. To examine the opportunity

divide, we document beliefs about computers and economic advance¬
ment and attitudes and experiences for online job search and taking
a course online.
The democratic divide. For the democratic divide, we measure atti¬

tudes and experiences regarding Internet use for voting, registering to
vote, looking up government information, looking up political infor¬
mation, and participating in an electronic town meeting.
Policy recommendations. In our discussion of policy recommenda¬

tions in the final chapter, we also include attitudes about vouchers
and a summary index measuring differences in digital experience.
Analyzing the Data
In our analysis of the survey data, we begin by listing the simple per¬
centages (or frequencies) for the responses we received, in order to
describe overall tendencies. Where we differ from previous studies,
however, is our use of a technique called multivariate regression. This
\
widely used statistical method allows us to list a number of “indepen¬

dent variables,” or possible explanatory factors for each result (for
example, home access), and to identify which are statistically significant
predictors for the result (for example, whether education, income, race,
gender, and age are related to home access). As we demonstrate in chap¬
ter 2, such crucial statistical techniques are lacking in nearly all of the
research that has been disseminated on the “digital divide.” The inde¬
pendent variables we use vary somewhat based on the research ques¬
tion, but we examine a number of socioeconomic and demographic
factors, employment status, and political variables such as partisanship.
Our income measure deserves explanation. We define low-income
individuals as having a household income below $30,000, which cap¬
tures working families of modest means as well as the very poor. There
was a very low response rate to questions about income beyond our
initial screen of above or below $30,000, which makes it difficult to
compare respondents with different levels of poverty. Although ide¬
ally we would have liked more specific information about income, we
feel that our measure is still a reasonable one, for several reasons. The
$30,000 threshold is only about 70 percent of the 2000 median
household income of $42,151 reported by the U.S. Census Bureau.17
In terms of absolute measures of poverty, $30,000 is about 200 per¬
cent of the official poverty threshold for a family of three.18 The fed¬
eral poverty threshold is calculated by taking a minimal food budget
and multiplying that figure by three. The National Academy of Sci¬
ences and a number of social scientists have criticized this method as
inadequate to account for all a household’s needs.1'1 Some estimates
have placed a more realistic poverty level for a family of four at about
$28,000 (in 1999).20 The Department of Labor has also defined a
2001 lower living standard income level” for its programs as a little
over $30,000 for a family of four in metropolitan areas in either the
Northeast or the West.21
Our results indicate that many of our “under $30,000” category fall
considerably below this threshold. For example, around one-third of
survey respondents (32 percent) are currently unemployed, and 87 per¬
cent are between twenty-one and sixty-five years of age, indicating most
respondents are the working poor or unemployed, not retirees. The fact
that nearly two-thirds of our sample was drawn from high-poverty cen¬
sus tracts (with a poverty rate of 50 percent or more) increases the like-
Redefining the Digital Divide I I
lihood that many of the individuals listed as having household incomes

below $30,000 actually fall below the official poverty level.
To facilitate interpretation of the statistical findings, we present
“What Matters” boxes within the text, listing all of the explanatory
variables that are statistically significant, without the numbers. For
those who are familiar with regression analysis, we also present the
multivariate logistic and ordered logistic regression tables in appen¬
dix 1. On key questions, we use an additional statistical technique to
present our findings. We present probabilities (also in the What Mat¬
ters boxes) to compare magnitudes for the influence of different
explanatory variables—whether income, education, race, age, or gen¬
der have a greater impact on access, skills, attitudes, or experience. We
convert the regression coefficients from the multivariate models to
probabilities using a Monte Carlo simulation technique.22 As a result,
we can report that holding other factors constant, there is a 35 per¬
centage point increase in the probability that individuals with a grad¬
uate education will have Internet access, in comparison to individuals
with less than a high school diploma. This technique makes the
results of the regression models as easy to understand as percentages,
requiring no knowledge of statistical methods.
One case study will be briefly reported in the chapter on the demo¬
cratic divide. The Berkeley, California, city government recently
solicited input for revisions to the city plan using an online forum.
The Berkeley experience, like an earlier experiment in Santa Monica,
California, illustrates that there are numerous challenges in using this
form of direct democracy.23 Ensuring equitable opportunities for par¬
ticipation is only one of these, according to our interviews.
Overview of the Book
The following chapters are organized around four divides, with a con¬
cluding chapter that presents some general recommendations for
public policy.
Chapter 2, on the access divide reviews the findings of major stud¬
ies, introduces our own data on the topic and uses multivariate regres¬
sion to establish the causes of disparities in access. Although previous
research has attracted attention to the issue and described trends over
\
the past few years, most studies fall short insofar as they are unable to
ascribe with any certainty the causes of inequality in access. We use
data from a national study that has tracked Internet access since the
mid-1990s and then analyze our own data from 2001, which cover a
broader range of questions about access and have a representative
sample of low-income individuals. Our findings contradict the con¬
clusions of some previous studies, but affirm that significant dispari¬
ties in access endure, and are based on multiple factors. Because we
use statistical controls, our study can distinguish the influences that
matter for the access divide.
Having documented the continued existence of an access divide
(using more rigorous analysis), in chapter 3 we venture into largely
uncharted territory on the skills divide. We make the case that the Inter¬
net requires “information literacy,” or the ability to locate and evaluate
information through this new medium, in addition to technical com¬
petence with hardware and software. There is little research on these
topics. Information literacy, however, requires basic literacy, the ability
to read and interpret information, as a fundamental prerequisite, and
national data show that low levels of literacy are pervasive in the United
States. Using our own survey, we describe our findings on technical
competence and information literacy. Following our data on the skills
divide, we examine the preferences our respondents express for learn¬
ing new skills and for locations for public access. These preferences are
disaggregated by socioeconomic and demographic characteristics,
offering some guidance for public policy intended to assist disadvan¬
taged groups in acquiring information-age skills.
The next two chapters of the book shift from describing the bound¬
aries of the access and skills divide to an investigation of their possible
consequences. In both chapters, we approach this task by first survey¬
ing evidence on the implications of technology disparities and then dis¬
cussing our data on the attitudes and experiences of our low-income
sample. This allows us to study beliefs and needs within different con¬
texts, rather than attitudes about technology in the abstract.
In chapter 4, we examine trends in the new economy” and find
that technology gaps do indeed affect economic opportunity, but
often in complex ways. Rapid change may further increase the impor¬
tance of technology for jobs and economic mobility in the future.
Our data on attitudes show widespread belief in a connection
between computer skills and economic advancement as well as some
unexpected patterns in the groups most likely to perceive technology

as economically advantageous. Similar relationships appear in regard
to attitudes and experiences with online job searches. Economic
advancement is clearly a salient motivation for learning technology
skills and using the Internet, for the population as a whole and for
some disadvantaged groups in particular.
Chapter 5, on the democratic divide, defines the possibilities for
technology in the political sphere, from campaign websites, Internet
voting and registration, electronic town meetings, and “e-govern-
ment,” initiatives that place information and services at the disposal
of citizens on the web. Scholarly debates tend to either overemphasize
or dismiss the role of technology in democratic participation, and the
empirical evidence so far is sketchy on many of these issues.24 Exam¬
ining attitudes and current use allows us to predict demand for future
innovation as well as who is likely to use technology for political par¬
ticipation and who is not. Our assessment is that computers and the
Internet promise to increase participation among some groups, while
reinforcing or even aggravating disparities in participation for others.
The final chapter summarizes the results for each of the previous
chapters and draws some conclusions across the four divides regard¬
ing the impact of factors such as race and education. We develop an
index of “digital experience” to show some commonalities in who cur¬
rently uses information technology for economic and political par¬
ticipation. Following a brief survey of current policy efforts, we
present a discussion of our recommendations for public policy.
Within this context, we include results from the survey on support
for vouchers but also discuss other alternatives. The next chapter
begins with a reexamination of the access divide, as a first step toward
defining the outlines of this multidimensional landscape.
Notes
1. Pippa Norris (2001) provides evidence regarding political participation

in a cross-national context, and we emulate her use of the term democratic
divide.” We approach the topic somewhat differently, however, with a more
extensive examination of disparities in the United States and survey data
based on a large sample of low-income and minority respondents.
2. U.S. Department of Commerce 1995.
3. Carvin, Conte, and Gilbert 2001; Puma, Chaplin, and Pape 20Q0.
4. Hoffman, Novak, and Schlosser (2001, 55) have credited Lloyd Mor-
risett of the Markle Foundation with coining the term, “digital divide,” but
Benjamin Compaine (2001, xiv), in the same volume, cites Morrisett’s
uncertainty about the term’s origins.
5. See for example, the most recent NTIA report, U.S. Department of
Commerce 2002.
6. See Trotter 2001 for a description of these initiatives. Information on

the Digital Divide Network can be found at www.digitaldivide.org.
7. Lacey 2000; Thierer 2000.
9. Shadid 2001.
10. Salamon 1991, 3.
11. Jefferson [1816] 1988.
12. Mann [1839] 1988.
13. Hochschild and Scovronick 2000, 209.

14. Putnam 2000, 175.
15. Nie and Erbring 2000.

17. U.S. Census Bureau 2001.
18. U.S. Census Bureau 2002a.
19. National Academy of Sciences 1996.

20. Uchitelle 1999.
21. See U.S. Department of Labor 2002.
22. See King, Tomz, and Wittenberg 2000. This technique involves cal¬
culating the change in the probability of Internet access, for example, that
is caused by moving from a variables minimum to maximum value while
simultaneously keeping all other variables set to their mean.
23. Dutton 1999, 184-85.
24. See also Norris 2001.
Chapter 1
The Access Divide
with Ramona McNeal
W hen public officials talk about policy with regard to the dig¬
ital divide, the topic usually centers on access. During the
Clinton administration, policies implemented to address
disparities in information technology usage were expressly designed
to increase Internet access. Programs such as the Technology Oppor¬
tunities Program (TOP) under the Department of Commerce and
the Community Technology Center (CTC) initiative and the E-Rate
administered by the Department of Education were put into place to
increase access to disadvantaged groups. When President Bush
released his budget proposal for the fiscal year 2003, it called for the
termination of both the TOP and CTC initiatives.1 The reasoning
for ending these programs, like the justification for creating them, is
based on access. The 2002 Department of Commerce report A
Nation Online: How Americans Are Expanding Their Use of the Inter¬
net argued that these programs had met their goals and that the Amer¬
ican public is coming online at a satisfactory rate.2
Whereas the Bush administration predicts sunny skies in its digi¬
tal divide forecasts, others see darker clouds on the information tech¬
nology horizon. More than 100 groups, including the National
Urban League, the National Congress of American Indians, and the
American Council of the Blind, came together on Capitol Hill in
May 2002 to launch the Digital Empowerment Campaign to oppose
Bushs decision to cut the CTC and TOP programs. Joining these
groups was a number of lawmakers, including Senator Barbara
Mikulski (D-Md.) and Representative Ted Strickland (D-Ohio).
They argue that, although it is true that Americans are getting con¬
nected in increasing numbers, there are still gaps in online use based
on race, ethnicity, and income. For example, whereas 60 percent of
the white households had Internet access in 2001, only 34 percent of
African American and 38 percent of Latino households did. In addi¬
tion, whereas nearly 78 percent of households with income between
$50,000 and $75,000 have Internet access, only 40 percent of those
with household incomes between $20,000 and $25,000 have web
access. Groups protesting proposed cuts in the federal budget fear
that elimination of these programs will make it more difficult to
address inequalities related to technology access.^
This chapter presents the findings on access from our low-income
survey and subjects our results to more rigorous analysis than the
methods used in previous studies. We therefore provide more deci¬
sive evidence that an access divide in fact exists and is not fading over
time. We review the previous research, the conflicting conclusions,
the shortcomings of various studies, and the advantages of our
approach. We include an original analysis of data from the American
National Election Studies (ANES), which allows us to track the access
divide over time, as well as more detailed evidence from our own July
2001 survey.
Evidence from Prior Research
Although both the Bush administration and members of the Digital

Empowerment Campaign assessed computer access in the United
States using the same report, A Nation Online: How Americans Are
Expanding Their Use of the Internet, one saw a glass that is half full
while the other concluded that the access glass is half empty. “Facts”
are often interpreted in starkly different ways in the midst of politi¬
cal battles, of course,^ but in this case the existing empirical literature
on computer and Internet access is indeed murky and incomplete. A
Nation Online, like most studies on Internet access, reports only sim¬
ple frequencies, or percentages, which tell little about the strength of
the association between variables or the relationship between vari¬
ables. It is well-known that race, ethnicity, income, and education are
highly correlated, thus simple frequencies or percentages can overes¬
timate the gaps in information technology based on any one of these
factors. African Americans and Latinos, for example, tend to have

lower incomes and educational attainment than do whites. The ques¬
tion, then, is whether race is an independent (nonspurious) predic¬
tor of access to information technology or whether, for example,
education is really driving differences in access.
The overwhelming majority of current studies rely primarily on
bar charts based on simple percentages. At most, these reports might
examine two variables at a time—for example, African American
respondents by income.5 These approaches are useful for illustrating
trends and suggesting possible relationships. Descriptive data pub¬
lished by the U.S. Commerce Department and Pew Charitable Trust
have served as important indicators of a developing policy problem,
thrusting the issue into the public spotlight; however, data of this type
are inadequate for making claims about the root causes of the prob¬
lem and can be open to different interpretations. The methods that
have been used to analyze the access divide are insufficient to sepa¬
rate the effects of overlapping influences and to establish with any cer¬
tainty what factors matter—race, education, income, or all of the
above. Our task in this chapter is to sort out the causal determinants
in who has access to information technology in the United States.
Before discussing our own analyses, we review the evidence presented
in major studies, from the U.S. Department of Commerce, the Pew
Charitable Trust, academic research, and market surveys.
U.S. Department of Commerce —
Reliable Data, No Statistical Controls
A Nation Online is fifth in a series of reports created by the National

Telecommunications and Information Administration (NTLA) that
examine disparities in usage and access to information technology.
The data reported by the NTLA are drawn from survey questions
from the U.S. Census Bureaus Current Population Survey (CPS),
administered to a large sample of 50,000—60,000 households/ When
the first NTLA report, Falling through the Net: A Survey of the “Have
Nots” in Rural and Urban America, was published in 1995, the Inter¬
net had not yet gained widespread acceptance and was still mainly a
tool for academics and the defense industry/ The Internet went
unnoticed in this first report, which instead focused on the penetra¬
tion of telephones, personal computers, and modems. The second
\
(1998) and third (1999) reports. Falling through the Net: New Data
on the Digital Divide8 and Falling through the Net: Defining the Dig¬
ital Divide? measured Internet access, as well as access to computers,
and were instrumental in bringing the term “digital divide” into'the
public consciousness.
The second report found that Americans were in general expand¬
ing their access to telecommunication technology such as telephones,
computers, and the Internet; however, gaps based on demographic
and geographical factors existed. Although differences in access con¬
tinued based on age, gender, education, rural/urban regions, and sin¬
gle/dual parent households, there were growing disparities based on
race, ethnicity, and income when compared to levels from the first
report. These differences were particularly acute when examining
Internet access and computer ownership. At the time of the first
NTIA report (1995), computer ownership among African Americans
(10.3 percent) was 16.8 percentage points lower than for whites (27.1
percent). This gap increased to 21.5 percent (40.8 percent for whites
and 19.3 percent for African Americans) by the second report in
1998. The difference between white home access (27.1 percent) and
Latino home access (12.3 percent) increased from 14.8 percentage
points for the 1995 report to 21.4 percent in the second NTIA report
(19.4 percent Latino home access).
Like the second report, the third report found that, although
Americans were getting more connected, the gaps based on income,
education and race/ethnicity were still widening further. During the
single year between the second and third reports, the gap between
white and African American households in Internet access increased
37.7 percent. For white and Latino households the disparity in
Internet access rose to 37.6 percent. Over the same period, the dif¬
ference in Internet access between college graduates and those with
only a high school degree escalated to 13.2 percent, while the divide
between the highest and lowest income levels expanded to 24.3 per¬
cent. These findings led the NTIA to conclude that disparities in
information technology were based on race and ethnicity as well as
education and income.10
Although the second and third reports told a story of inequality,
the final two reports began to show signs of a narrowing divide. The
fourth report (2000), Falling through the Net: Toward Digital Inclu¬
sion, found that Internet access had increased across all groups and
that the gender gap had largely disappeared. Nevertheless, the report
also observed that an access divide remained based on income, edu¬
cation, race, ethnicity, age, disabilities, and dual/single parent house¬
holds. The gap in Internet access between whites (46.1 percent) and
African Americans (23.5 percent) was 22.6 percent; similarly, the dif¬
ference between whites and Latinos (23.6 percent) was 22.5 percent.
Although the race and ethnicity gaps had narrowed somewhat, they
were still substantial. The difference in Internet access between those
with a bachelor’s degree (64 percent) and a high school diploma (29.9
percent) was larger still, at 34.1 percent. Most striking was the gap
between individuals with a household income greater than $75,000
(77.7 percent) and those with a household income less than $15,000
(12.7 percent), a difference of 65 percentage points.11
The final report (2002), A Nation Online, again noted increases
across all groups and found that urban and rural differences were dis¬
appearing.12 Nevertheless, it reported persistent gaps in “Internet
use” based on age, income, education, race, ethnicity, dual/single par¬
ent households, and, in particular, mental or physical disabilities.
Unlike the other NTLA studies, A Nation Online emphasized Inter¬
net use instead of Internet access at home. This measure is different
from access in that it includes use of the Internet at work and school
rather than only at home, where the divide is greatest. There is noth¬
ing inherently wrong, of course, with measuring use, and it is possi¬
ble to debate whether this is actually a more significant measure than
home access. When compared to the access measure, however, Inter¬
net use figures can give the impression that technology gaps are nar¬
rower than they actually are, especially if frequency of use in locations
outside the home is not taken into account.
Because of the large sample sizes of the CPS, the NTLA reports can
be used to generalize to the American population. As a government
agency, however, the NTIA collects and publishes data rather than
models cause-and-effect relationships.
Pew Research Center Surveys—

Reliable Data, No Statistical Controls
Like the NTIA, the Pew Research Center began conducting a series
of studies on Internet access in 1995. Also like the NTIA reports,
these studies were based primarily on descriptive statistics (frequen-
cies and simple percentages) presented as bar charts. The same limita¬
tions of the NTIA data analysis apply to the Pew studies, for they can¬
not sort out the causes of inequities in access to information
technology. There is considerable overlap between the findings in the
two series of reports. The Pew Research Center found in its Who’s Not
Online report that by spring 2000 gender had reached parity in Inter¬
net access but gaps still remained based on income, education, age,
race, ethnicity, and geographic location.13 Although the Pew report
found gaps, they were not of the same magnitude as those documented
by the NTIA. The NTIA in 2000 reported a 22.6 percent difference
between whites and African Americans and a 22.5 percent gap
between whites and Latinos in Internet access; the differences for the
PEW study were only 14 percent and 6 percent. The NTIA also
reported a 34.1 percent disparity based on educational attainment
(between those with high school and college degrees), whereas the Pew
study found a larger (41 percent) difference. The 2000 Pew Internet
and American Life Project report, “African-Americans and the Inter¬
net,” concluded that, when survey results were broken down by
income and education, differences in access based on race still
remained.14 The 2000 NTIA report reached the same conclusion for
both race and ethnicity. Some research has suggested that racial and
ethnic disparities in Internet access may be partly due to the fact that
minorities have different information and content needs that the
Internet is not meeting.15 Like the NTIA studies, the Pew surveys
include representative samples that can be used to draw generalizations
about the American public (but not specifically the poor), but the data
gauge trends rather than trying to isolate explanations for the trends.
Academic Studies—Older Data, Some Analysis

with Statistical Controls
Academic studies have differed in their assessment of which factors

are responsible for disparities in access, demonstrating how much
research methods influence findings. Hoffman, Novak, and Schlosser
conclude that race, as well as income and education, matters, but
their study is limited by the absence of statistical controls.16 Neu,
Anderson, and Bikson17 and Wilhelm18 conducted two of the rare
studies using multivariate regression, which controls for multiple
influences. Neu, Anderson, and Bikson used the 1993 and 1997 CPS
and logistic regression to predict access to a home computer and
e-mail use. They found that income, education, race, and ethnicity
were important factors for both. Computer access and e-mail use
were greatest among Asian Americans and whites who were more
affluent and more educated. Age and region also played a part in
determining access, but to a lesser extent, with the young more likely
to have a computer and e-mail access.19 Because of their use of more
sophisticated methods, Neu, Anderson, and Bikson provide a more
complete picture of the digital divide than can be puzzled together
from more recent studies of access. Nevertheless, this study was based
on older data that are limited in their ability to predict current trends
in access to information technology.
Wilhelm, like Neu, Anderson, and Bikson, used multivariate sta¬
tistics (logistic regression) and based his research on the 1994 CPS to
examine home computer and modem ownership. The interpretations
and validity of the research, however, are limited by errors in the data
analysis that have the effect of obscuring differences based on race and
ethnicity.20 Not surprisingly, Wilhelm discovered little racial or eth¬
nic influence. In addition, the study relies on older data.
In one of the best-known academic studies, Nie and Erbring found
that only education and age matter in the digital divide.21 Their study
was based on data collected in December 1999 through a national
survey of 4,113 individuals in a 2,689-household panel. Each mem¬
ber of a panel household filled out separate questionnaires. Because
this study focused on online behavior and the survey was conducted
over the Internet, all respondents (even those without previous access)
were provided with a WebTV set-top box and free Internet access and
e-mail accounts. The results on Internet use presented in the study,
however, were based solely on participants who had previous access.
This survey design, while unique, violates the assumption of inde¬
pendence underlying statistical methods by allowing more than one
member of a household to participate in the study. Using multiple
members of the same household biases the sample so that it is not rep¬
resentative of the population as a whole. If survey respondents are not
randomly selected, using statistics to draw conclusions that apply to
the whole population is questionable in terms of reliability. Again,
only simple frequencies and percentages are reported in the study.
Market Research Surveys—Questionable Data and Analysis
Whereas the NTLA reports and Pew Research have found enduring
gaps in Internet access, other studies report conflicting findings. Mar¬
keting research firms (e.g., Forrester Research Inc., Cheskin Research)
provide their clients with studies that show how best to take advan¬
tage of this new market. Their research does not always coincide with
the findings of government, foundation, or academic studies. In par¬
ticular, some surveys from marketing research firms have suggested
that the divide based on race has been overstated. For example, a For¬
rester study based on more than 80,000 mail surveys found that
Latino households were more likely to be online than white house¬
holds. The finding that Latino households are online more than
whites directly contradicts the results of the NTLA and Pew Research
surveys. The Forrester study also found that income played the great¬
est role in promoting Internet access, although age, education, and
technology optimism are also important factors in determining
access. The study suggested that, while African Americans lagged
behind in Internet access, it was because of disparities in income and
not racial factors.2^ To evaluate the validity of these findings, it is
important to know the response rate of the mail survey and the sta¬
tistical methods on which the conclusions were based. A low response
rate, which is common in mail surveys, can bias the findings. Those
who take the time to send back the survey may only be those who are
already using the Internet and are most interested in the topic. Nei¬
ther the response rate nor methodology is apparent from the report,
making it difficult to assess the validity of the findings.
This pinpoints one of the differences between research that meets
academic standards and those of market research. Scholarly research
requires that statistical methods are made public, explained in detail,
and that the data used are available so that others may attempt to repli¬
cate and evaluate the validity of findings. Otherwise, it is impossible
to know how trustworthy the findings are. The market surveys do not
meet these standards of public scrutiny. Government agencies (such as
the Bureau of the Census, which conducted research for the NTLA)
and studies funded by major foundations generally observe the stan¬
dards of academic research. Government agencies aim to provide
information about important policy issues but often do not conduct
advanced statistical analysis to draw cause-and-effect relationships.
Strange results and methodological flaws in market research might

be of little concern if the only audience for these findings was compa¬
nies that contracted with these firms; however, this less rigorous and
transparent research is often publicized in the press or books. The vol¬
ume edited by Benjamin Compaine and published by MIT Press
includes market surveys on the same footing as government or acade¬
mic studies. The editor, in fact, argues that Forrester, Cheskin, and the
academic study by Nie and Erbring, which concluded that race was
not a factor, are more “sophisticated” than the NTIA and the Hoff¬
man, Novak, and Schlosser studies that found that race did matter.24
Summary of the Previous Research
Our review of the research has found that a number of surveys have
been conducted exploring demographic and geographic factors that
may impact computer and Internet access. All of the eight studies
summarized found enduring gaps in Internet access based on educa¬
tion and age. Seven of the eight reported that income played an
important role in promoting technology access. The findings were
mixed in terms of whether factors such as race and ethnicity were
important, with four of the eight studies reporting access gaps based
on race and ethnicity. None of the studies found gaps in technology
access based on gender, but other factors cited in regard to computer
access include dual/single parent households, mental or physical dis¬
abilities, geographic location, employment status, and optimism
about technology.
The Need for Reliable Data

and Analysis with Statistical Controls
While the fast-changing population of Internet users creates an obsta¬

cle for making predictions, the greatest weakness related to previous
research has been the lack of appropriate statistical methods, specifi¬
cally statistical controls. Most studies have simply described the per¬
centage of a certain group that has computer or Internet access. Even
studies that use cross-tabulation tables to report the relationship
between two variables (for example race and income) can be decep¬
tive. Looking at differences across only one or two variables does not
permit the researcher to control for the effect of other related factors.
Many demographic factors, such as income and education, are inter¬
related. If you examine them one at a time, the differences found may
be in reality attributable to other related factors. The independent
effects of race, ethnicity, income, education, age, gender, and employ¬
ment status can be found only by using an appropriate statistical
method, such as multivariate regression.
In an attempt to address some of the weaknesses of previous
research, this chapter will use two different surveys to explore the influ¬
ence of a variety of factors on information technology access. First, we
examine broad trends in our own 2001 survey data, using simple per¬
centages. Second, we analyze data on Internet access using informa¬
tion drawn from the 1996, 1998, and 2000 ANES survey. The ANES
is a nationwide, large-scale study that conducts in-person and tele¬
phone surveys using randomly selected respondents. The ANES allows
us to examine Internet access over time, while our survey includes more
specific questions about access to varying forms of information tech-
nology and also includes a representative sample of low-income respon¬
dents. In addition to using complementary data sources, we employ a
more accurate methodology to analyze the data. Simple percentages are
supplemented by multivariate regression procedures and a Monte
Carlo simulation technique that estimates probabilities and predicts the
likelihood of information technology access.25
Low-Income Survey: Patterns of Access and Use
Frequencies provide descriptive trends and a first cut at the data in

terms of understanding who does and does not have access to infor¬
mation technology. Sixty-one percent of our respondents reported
having access to a home computer, and 54 percent reported having
home Internet access. This closely tracks the figures on Internet use
for the U.S. Department of Commerce study conducted in Septem¬
ber 2001.26 Compared to the 61 percent of respondents who have
home computer access, 58 percent had an e-mail address through
which they can send or receive e-mail. Only a small percentage of
respondents (14 percent) said they had high-speed Internet access.
We also included more detailed questions on alternative ways to
access information technology. This is especially important for low-
income respondents who do not have a computer or Internet access

at home.
The survey allowed respondents the opportunity to identify mul¬
tiple locations for access to computers and the Internet. Almost equal
percentages of respondents used a computer at home (54 percent)
and at work (49 percent). Of employed respondents, however, 65 per¬
cent used the computer at work. Relatively small percentages of
respondents (15 percent) used the computer at a school or a public
library. Yet more than a quarter of those surveyed used the computer
at a friend or relative’s house (26 percent). This suggests the inter¬
personal potential of information technology use, perhaps for enter¬
tainment, information search, or informal instruction.
A parallel story emerges when examining venues for Internet access.
Compared to the 51 percent of respondents who actually used the
Internet at home, 34 percent used the Internet at work (45 percent of
employed respondents used the Internet at work), while close to the
same number use the Internet at a school (11 percent) or a public
library (10 percent). Twenty percent used the Internet at a friend or
relative’s house, which is double the rate of Internet use at libraries and
schools. For the general population, usage of information technology
at home, however, far outweighed usage at other locations, even work.
This finding demonstrates the importance of measuring home Inter¬
net access rather than lumping together Internet access at home and
other locations. Home access allows more privacy and greater flexibil¬
ity in terms of length of use and round-the-clock availability.
How do these figures change if we examine the answers given by
people who do not have a computer at home? Do more of these indi¬
viduals use computers at other places? Of the 710 respondents with¬
out a home computer, 30 percent use the computer at work, 9 percent
at school, and 13 percent at the library. Twenty-four percent used a
computer at a friend’s house. These frequencies are comparable to
those for the overall population. The data suggest that patterns of
computer use outside the home do not differ significantly among
those with and without a home computer. Individuals without a
home computer are not using public access more than those with
home access. The Internet is used even less outside the home by those
without a home Internet connection. Of the 841 respondents with¬
out Internet access at home, 17.5 percent indicated they use the
Internet at work, 7 percent at school, 9 percent at the library and 16
percent at a friend’s house. These figures are lower than the percent¬
ages for the overall population. The data show that work or the homes
of friends and relatives are the most common venues to access infor¬
mation technology outside of the home, regardless of home access.
Home and work are clearly associated with more frequent access
to information technology than a friend or relative’s house, schools,
or public libraries. We asked respondents about the number of times
they used computers and the Internet at varying locations. When
asked, “Last month, how often did you use a computer at home,” 14
percent reported low usage (1-10 times), 20 percent moderate usage
(11-30 times), and 17 percent high usage (31-100 times), while 3
percent reported very high usage (more than 100 times). Very high
frequency of computer use was more likely to take place at work
rather than at home: 8 percent reported low use at work (1-10 times),
12 percent moderate use (11-30 times), 17 percent high use (31-100
times), and 12 percent very high use of over 100 times. In contrast,
frequency of access was much lower at a friend or relative’s house: 22
percent reported low use (1—10 times), 2 percent moderate use
(11-30 times), and less than 1 percent of the respondents high or very
high use at a friend or relative’s house. Frequency of access at public
libraries was even lower. Only 12 percent reported low usage, 1.5 per¬
cent moderate usage, and less than 1 percent high or very high usage.
Although friends, relatives, and libraries may provide exposure to
information technology, work or home access is associated with con¬
sistent use.
Exploring Patterns of Access

with Statistical Controls
The descriptive statistics presented in the last section give us a picture

of current access and use but tell us little about the causes of dispar¬
ities or the development of trends. Our survey provides data from a
single point in time. To assess the validity of our low-income survey
data and to understand changes over time, we analyze access to infor¬
mation technology drawing on the 1996, 1998, and 2000 ANES.
This large-scale, nationwide study selects respondents randomly for
in-person and telephone surveys. The ANES is conducted every two
years and provides one of the most comprehensive sources of data
regarding popular attitudes toward government. Beginning in 1996,

the ANES also began collecting data on Internet access and use. For
the remainder of this chapter, we use multivariate statistics to ascer¬
tain which factors are statistically significant when we control for
other possible explanations.
Before presenting our findings, a few words are necessary to
describe our model. The dependent variable—the result that we want
to explain—is Internet access.27 The independent variables—the fac¬
tors that potentially influence Internet access—are individual-level
attitudinal and demographic factors suggested by the findings of pre¬
vious studies and the fifth report (2002) of the U.S. Department of
Commerce series, A Nation Online: How Americans Are Expanding
Their Use of the Internet,28 This most recent installment of the
Department of Commerce’s digital divide series indicates that,
although Internet access has been increasing for all groups, there are
still significant gaps based on race, ethnicity, age, and, in particular,
income and education.
In our analysis of the ANES, personal income in 1996 and 1998
is measured on a 24-point ordinal scale, where 1 indicates annual
family income ranging from $0 to $2,999 and 24 indicates that
annual family income is $105,000 or more. In 2000, income is mea¬
sured on a 22-point scale, where 1 indicates annual family income
ranging from $0 to $4,999 and 22 indicates that annual family
income is $200,000 or more. Education is measured on a seven-point
ordinal scale, with a “dummy” variable for gender, where respondents
are coded 1 for female and 0 for male. (Dummy variables are used for
gender, race, ethnicity, partisanship, and income. This means that
they are coded as categories, with female, African American, Latino,
Asian American, Democrat, Republican, and those with an annual
income less than $30,000 coded 1 and 0 otherwise.) To control for
race and ethnicity, variables for African Americans, Asian Americans,
and Latinos were coded 1 and 0 otherwise, with non-Latino whites
as the reference group. Age was measured in years. Because we also
discuss political participation in chapter 5, we introduce partisanship
in the analysis. Chew has found that, when individuals are interested
in the political process, they seek out greater levels of media coverage
of elections to become better informed.2^ Strongly partisan individ¬
uals may also be motivated to gain access to the Internet, so we mea¬
sure the strength of partisan identification as well. A seven-point scale
\
measures partisanship (1 = strong Democrat to 7 = strong Republi¬

can). A series of dummy variables are used to account for political atti¬
tudes, including strong Democrat, strong Republican, and pure
independent.
The Findings: ANES Data
By examining Internet access over time using statistical controls, we

discover some unexpected patterns. One of the most stable findings,
shown in table A2.2, is that the young are significantly more likely to
have access to the Internet, when we control for other factors.
Although the young are more likely to have Internet access, they also
tend to have lower incomes. Previous research has modeled the rela¬
tionship between income and computer and Internet access using a
simple linear relationship—that is, as income increases, Internet
access increases.3® We find the relationship between income and
Internet access to be more complex and more accurately measured by
a nonlinear (quadratic) model, which we explain below.
Because young people were the first to embrace the Internet,
regardless of income, when going online was a relatively new phe¬
nomenon, low-income individuals were actually more likely to have
Internet access in 1996 and 1998. Thus, in 1996 and 1998 the low-
income groups (which often included the young and students) and
upper-income groups tended to have the highest access to the Inter¬
net. By 2000 the number of Internet users had expanded sufficiently
that the expected relationship of increased income is associated with
more Internet access.31 In the analysis of the low-income data in the
next section, we model a linear relationship between income and
Internet access, reflecting this finding.
Across the years, individuals with more education are more likely
to have access to the Internet, holding other factors constant. In
1998, females are less likely to have Internet access than males, but
this relationship does not hold in 1996 and 2000. There is evidence
that strong Republicans are more likely to have Internet access and
strong Democrats significantly less, confirming previous research.32
We will return to this issue in the chapter on the democratic divide.
The other notable finding is that African Americans have significantly
lower access to the Internet in both 1998 and 2000 compared to
whites.33 Our analysis of the ANES survey is fairly consistent with

government reports that are not based on multivariate statistics and
earlier studies based on e-mail access alone.3^ The main difference we
discovered was that relationship between income and access was com¬
plex at the beginning of the Internet’s diffusion, because of the num¬
ber of students and other young people who adopted the technology
early on. A more straightforward relationship between income and
access has developed over time.
Low-Income Survey Data
The ANES survey data, however, do not include a representative sam¬

ple of the poor or of African American and Latino populations. Accu¬
rately measuring the experiences of these important populations is
necessary to sufficiently define the access divide. The ANES data also
do not include measures of access such as e-mail. Do the same demo¬
graphic patterns associated with Internet access using the ANES data
appear in our national representative sample?
Table A2.3 uses logistic regression and multivariate analysis based
on low-income survey data to examine factors associated with com¬
puter use and Internet access. We present three different analyses,
based on the questions, “Do you personally have a home computer?’
“Do you have an e-mail address with which you can send and receive
e-mail?” and “Do you have access to the Internet from home?”35
Consistent with the analysis based on the ANES data, explanatory
variables include gender, race, ethnicity, partisanship, income, and
education. For race and ethnicity whites were the reference, or left-
out, group that was not coded. For partisanship, those without strong
partisan identification—independents—were the reference group
that was not included. Education was measured on a five-point scale
with responses ranging from 1 (less than a high school degree) to 5
(postgraduate work). Age was recorded in years. We present the
results of our regression analysis in the following What Matters box.
The factors listed are those that are statistically significant, when con¬
trolling for other explanations.
As in previous research, the poor, the less educated, and the old
were significantly less likely to have a home computer, an e-mail
address, or Internet access. Unlike the ANES data, which include
\
WHAT MATTERS
Who Is Least Likely to Have Internet Access at Home?
Poor
Less-educated
Old
Democrats
Latinos
African Americans
Who Is Least Likely to Have an E-Mail Address?4

Poor
Less-educated
Old
Females
Latinos
African Americans
Who Is Least Likely to Have a Home Computer?*’

Poor
Less-educated
Old
Latinos
African Americans
Note: The only statistically significant differences are the ones reported above
(see table A2.3).When multivariate regression is used, these are the variables that
matter, holding other factors constant.
Republicans are statistically more likely to have an e-mail address than indepen¬
dents, but Democrats are not statistically less likely to have an e-mail address.
Republicans are statistically more likely than independents to have a home com¬
puter, but Democrats are not.
small samples of racial and ethnic minorities, our data indicate that
both African Americans and Latinos were significantly less likely to
have home computers, e-mail addresses, or Internet access than
whites, after controlling for socioeconomic conditions. Findings
from the low-income sample demonstrate that race and ethnicity
clearly matter in the access divide, even after accounting for variations
in income and education. Asian Americans and whites were found to
have comparable access. These findings also provide evidence that

partisanship is an important factor in measuring the access divide.
Republicans were more likely to have an e-mail address and home
computers than the reference group (independents), while Democ¬
rats were less likely to have Internet access than those without strong
partisan ties.
To facilitate interpretation of the statistical findings, our results for
home Internet access—arguably the most important of the informa¬
tion technology access variables—were converted to expected proba¬
bilities using a Monte Carlo simulation technique.36 This allows us
to compare the magnitude of differences, for example, whether edu¬
cation or income is more important, even if they are both statistically
significant. The results are easy to understand, because they resemble
simple percentages. It is important to remember, however, that these
are probabilities based on regression models rather than percentages.
In other words, a 17 percent difference based on race should be read
as a 17 percentage point difference in the probability that African
Americans will have Internet access, compared to the probability for
whites, holding all other factors constant.
The estimates provide an interesting comparison of access based
on income, education, gender, race, age, and partisanship. We calcu¬
late the change in the probability of access caused by moving from a
variable’s minimum to maximum value while simultaneously keep¬
ing all other variables set to their mean (or the change in the proba¬
bility of home Internet access when moving from the 1 to 0 category
for dichotomous, or dummy, variables).
The simulations show that income plays an important role in
determining home Internet access, controlling for other factors,
including education, age, race, ethnicity, gender, and partisanship. All
else equal, individuals in the lower income category had a 39 percent
probability of having Internet access, compared to those in the high¬
est income group, who had a 63 percent probability of enjoying home
Internet access, a difference of 24 percentage points.
Education was important as well. Holding other factors constant,
college graduates were 21 percent more likely to have home Internet
access than those with only a high school diploma; those with some
graduate education were 35 percent more likely to have Internet
access than those without a high school degree. Of respondents with
some graduate education, 71 percent had access to the Internet at
WHAT MATTERS
Who Is Least Likely to Have Internet Access at Home?
Poor (39% for low-income vs. 63% for high-income)—24-point difference

Old (43% for 61 -year-olds vs. 67% for 28-year-olds)—24-point difference
Less-educated (44% for high school graduates vs. 65% for college gradu¬
ates)—21 -point difference
African Americans (37% vs. 54% for whites)—17-point difference
Latinos (41% vs. 54% for whites)—13-point difference
Democrats (54% vs. 64% for Republicans)—10-point difference
Note: Estimates are based on a hypothetical respondent who is female, white,

and independent, with values for education, age, and income set at their mean.
The only statistically significant differences are the ones reported above (see
table A2.3). We have calculated the probability of access, holding other factors
constant.
home, compared to 65 percent of college graduates, 54 percent of

individuals with some college, 44 percent of high school graduates
and 36 percent of respondents without a high school degree. Educa¬
tion and income emerged as important factors in the access divide,
consistent with other studies, even in multivariate statistical analyses.
As the earlier analysis of the ANES data showed, age continues to
be an important factor. The simulations show a 24 percent increased
probability of having access for twenty-eight-year-olds (one standard
deviation below the mean) in comparison to individuals who are
sixty-one years of age (one standard deviation above the mean). For
the young, the likelihood of having Internet access from home was
67 percent, all else equal, compared to older respondents, for whom
there was a 43 percent likelihood of having home Internet access.
Holding other demographic and attitudinal factors constant, differ¬
ences based on gender were not statistically significant.
Contradicting other reports that find only education and age mat¬
ter in the digital divide, we find that race and ethnicity clearly do
matter, though somewhat less than education and income.37 Asian
Americans had the highest predicted probability of Internet access
(72 percent), with whites significantly behind (54 percent). Latinos
trail whites (41 percent), and the probability of having home Inter-
net access is lowest for African Americans (37 percent). The differ¬
ence in the probability of home Internet access between Asian Amer¬
icans and African Americans is 33 percent. Even after holding
constant socioeconomic status, some racial and ethnic minorities
(African Americans and Latinos) are significantly less likely to have
access to the Internet at home than whites and Asian Americans.
The findings for partisanship were mixed. Democrats were less
likely to have Internet access (54 percent) than Republicans (64 per¬
cent) and independents (58 percent). Because individuals with
Republican partisanship are 10 percentage points more likely to have
access to the Internet at home than Democrats, this may have some
influence on policy. Republicans may be less concerned about issues
such as the access divide, for example, or may be more interested in
issues such as e-government, given their more wired constituency.
Conclusion
The aims of this chapter have been both descriptive and analytical.
Our low-income sample offers a detailed picture of computer and
Internet access for disadvantaged groups. Several trends are visible.
Most Americans use computers and the Internet at home or work,
and a quarter use computers at a friend or relative’s house. A much
smaller percentage (about 15 percent) use public access services at
libraries. The percentage of people who use computers in places out¬
side the home is similar for both those with and without home access,
and in fact those lacking a home connection are less likely to use the
Internet in other places such as libraries. This suggests a lack of inter¬
est or a dearth of knowledge and skill regarding the Internet. Frequent
use of computers and the Internet occurs at work or at home. Infor¬
mation technology use in other places tends to be sporadic. These
simple percentages demonstrate the significance of home access but
can tell us little about what influences access.
With the proliferation of sophisticated statistical analyses in the
social sciences, it is surprising to find the research on such a well-pub¬
licized policy issue dominated by descriptive data. Although an exten¬
sive number of surveys on information technology access and use
have been conducted by government agencies, nonprofit organiza¬
tions, and market research firms, few studies report findings based on
multivariate regression that can isolate the effect of specific factors on

who has access to computers and the Internet. Contradictory find¬
ings provide ammunition for both sides as debates rage on in Wash¬
ington, D.C., the policy community, and academic circles about
whether an access divide really exists.38 Using two sources of data, a
representative sample of low-income respondents, and multivariate
methods, we find unequal access to the Internet over a period of years
and continuing inequities for the Internet, e-mail, and computer
ownership. Our analysis of the ANES data over time demonstrates
that gender disparities for computer and Internet access have faded.
Statistically significant differences still exist in e-mail access, and per¬
haps this can be explained by differences in jobs. The complex rela¬
tionship between income and age that existed when young people
were among the earliest adopters of the Internet has given way as
income has increased in importance. The ANES data show that
African Americans are statistically less likely to have access, but the
more representative sample in our survey shows that this is true for
Latinos as well.
Table 2.1 compares the reported gaps in access to the Internet
based on race, ethnicity, education, and income from the NTIA 2000
survey,39 Pew Research 2000 survey,40 and analysis of our low-
income survey (Tolbert, Stansbury, and Mossberger 2001). While
Pew and the NTIA report only simple percentages, our analysis
reports expected probabilities based on a multivariate regression
analysis that controls for correlation, or overlap, between factors such
as race, education, and income.
Although the NTIA survey overestimates racial disparities in Inter¬
net access between African Americans and whites, Pew underesti¬
mates the racial divide. Our data and analysis suggest that, after
controlling for other factors, whites are 17 percentage points more
likely to have Internet access than African Americans. In measuring
the ethnicity gap, we again estimate a middle position, lower than the
NTIA survey and higher than the Pew Research survey. All else equal,
we find that whites are 13 percentage points more likely to have Inter¬
net access than Latinos.
Because simple percentages tend to exaggerate, or overestimate,
the true relationships, our analysis reveals that the education gap is
smaller than that reported by Pew and NTIA, but is still substantial
and larger than disparities in access based on ethnicity or race. Indi-
Table 2.1
Measuring the Access Divide
Department of
Commerce/NTIA Pew Research Low-Income
Internet Access Survey a (%) Surveyb (%) Surveyc (%)
Race gap (African 22.6 14 17

American vs. white)
Ethnicity gap (Latino 22.5 6 13
vs. white)
Education gap (high 34.1 41 21
school diploma vs.
college degree)
Income Gap (below 39 34 24
$30,000 vs. above
$30,000)
Sources: a U.S. Department of Commerce 2000b; b Pew Research Center 2000; c Tolbert,
Stansbury, and Mossberger 2001.
Notes: For the low-income survey, data are predicted probabilities based on multivariate
regression. Data reported on income gaps for NTIA and Pew are estimates because income
is not measured in exactly the same intervals in the three surveys.
viduals with a college degree are 21 percentage points more likely to

have Internet access than those with only a high school diploma.
Finally, all three studies report enduring gaps between the affluent
and poor. Flolding other demographic factors constant, we estimate
that individuals with incomes above $30,000 are 24 percentage
points more likely to have Internet access than those with incomes
below $30,000. Our findings are consistent with those reported by
the other studies.
The striking result is that all three studies based on different sur¬
vey data and statistical methods report persistent gaps in access to the
Internet based on race, ethnicity, education, and income. This is so
even for the most recent data, which have been heralded by the Bush
administration as evidence that the digital divide is vanishing and
insignificant. The data reveal that a “digital divide” in terms of infor¬
mation technology access is an undeniable reality. Even as more
Americans purchase computers and flock online, most of the dispar¬
ities that emerged during the latter half of the 1990s remain.
Notes
1. Benner 2002, 1.
3. Wright 2002, 1-2.
4. Stone 2002, chap. 13; see also Mossberger 2000, 156.
5. Hoffman, Novak, and Schlosser 2000.
6. U.S. Department of Labor 2002 and U.S. Department of Commerce
2002.
10. U.S. Department of Commerce 1999, 8.
11. U.S. Department of Commerce 2000b. For the racial gap see p. 98;
for the educational gap see p. 99; and for the income gap see p. 8.
13. Pew Research Center 2000, 5.
14. Pew Internet and American Life Project 2000, 5.
15. Novak, Hoffman, and Venkatesh 1997, 3; Hoffman, Novak, and
Schlosser 2000.
16. Hoffman, Novak, and Schlosser 2000.
17. Neu, Anderson, and Bikson 1999.
18. Wilhelm 2000.
19. Neu, Anderson, and Bikson 1999, 146-48. The authors used logis¬
tic regression because they constructed dummy variables for age, gender,
race, ethnicity, income, and region (urban, rural, and suburban).
20. Unlike most studies, Wilhelm (2000) included two variables for
white (non-Latino) in his analysis: (1) a dummy variable for ethnicity coded
1 for non-Latino and 0 otherwise and (2) a series of dummy variables for
race that included white. This dummy variable coding of race/ethnicity cre¬
ates a situation of near perfect multicollinearity, which makes it impossible
to determine the impact of race or ethnicity on access to a home computer
or modem. In addition to the dummy variable for white, Wilhelm also
included dummy variables for African American, Asian Ajmerican, and
Native American respondents, leaving “other race” as the reference group for
race. Given that very few individuals would have been in the reference group,
the race variables also suffer from near perfect multicollinearity. Multi¬
collinearity is a violation of the assumptions of linear regression that can lead
to invalid findings.
21. Nie and Erbring 2000.

22. Nie and Erbring 2000, 5.
23. Walsh, Gazala, and Ham 2001, 279-84.
24. Compaine 2001, 267.
25. King, Tomz, and Wittenberg 2000.
26. Our figures fall near the data for September 2001 cited by the NTIA.
The NTIA reported 66 percent and 54 percent for computer and Internet
use by individuals, and 57 and 51 percent for household ownership of com¬
puters and home Internet access, respectively. Our data for individual home
computer and Internet access are within the standard margin of error of ±4
points in comparison to the NTIA’s reported household data (U.S. Depart¬
ment of Commerce 2002).
27. Because Internet access is measured by a dummy variable, with Inter¬
net access coded 1 and 0 otherwise, logistic regression coefficients are re¬
ported in table A2.2.
29. Chew 1994.
30. Neu, Anderson, and Bikson 1999; Department of Commerce 2002.
31. We measure income by including an additional squared term (income
multiplied by income) to model a quadratic equation.
32. For similar findings, see Bucy 2000. This does not, however, match
the prediction that all strong partisans are likely to have access, as Chew’s
(1994) findings on other media suggest.
33. Recent research based on the CPS suggests that Asian Americans have
higher Internet access than whites (U.S. Department of Commerce 2002).
34. Neu, Anderson, and Bikson 1999.
35. We create three models with the binary dependent variables for home
computer access, e-mail, and home Internet access. The dummy variables
are coded 1 for access to technology, and 0 for no access. Since the depen¬
dent variables are binary, our estimates are based on logistic regression.
36. King, Tomz, and Wittenberg 2000.
37. Nie and Ebring 2000.
38. See Compaine 2001.
39. U.S. Department of Commerce 2000b.
40. Pew Research Center 2000.
Chapter 1
The Skills Divide
A lthough there have been many reports and studies of the state
of access to technology and the Internet, few studies, and none
with the scope of the one presented here, have addressed the
state of computer-related skills. Two distinct concepts describe the
knowledge and skills needed to use information technology effec¬
tively. Technical competencies are the skills needed to operate hardware
and software, such as typing, using a mouse, and giving instructions
to the computer to sort records a certain way. Information literacy is
the ability to recognize when information can solve a problem or fill
a need and to effectively employ information resources. Information
literacy is needed to navigate the Internet for work, school, political
information, medical information, news, entertainment, and other
purposes. Both technical competency and information literacy are
needed to fully exploit the potential of information technology,
although there are a number of computer applications that only
require technical competence. Computers and the Internet are more
than data storage or communication devices. They are gateways to a
seemingly limitless but complex array of information sources of var¬
ied uses and quality.
The data from our survey indicate that the technology skills
divide is quite pronounced for many groups. The lack of funda¬
mental technology-related skills—such as using a mouse and typing,
using e-mail, locating information on the web, and using word pro¬
cessing and spreadsheet programs—is a clear indicator of the need
for policy attention to this issue. The acquisition of skills such as
technical competence and information literacy may be particularly
problematic for some individuals, however, in particular those who
lack basic literacy.
This chapter also discusses the preferences for acquiring these skills
expressed by survey respondents. Public policy to some extent has
supported public libraries and community technology centers as
instructional providers for technology as well as locations for public
access; however, public libraries and other providers need guidance in
the selection of instructional methods. Do those who need to learn
how to use a keyboard or a mouse prefer one-on-one instruction? Do
those who need to learn how to search for information on the web
prefer group instruction? Our survey findings provide some answers
to these questions and assist instructional providers in using their
resources advantageously.
Because public access can potentially play an important role in
addressing the skills divide, we also explore attitudes about public
access locations. In the previous chapter, we found that only a small
percentage of the population has used libraries for public access and
that such use has been almost equally modest even among those who
have no computer or Internet access at home. Possible explanations
for this finding are that libraries are are not frequented by disadvan¬
taged groups who lack home access, because they are inconvenient or
are viewed as unwelcoming or irrelevant to such groups. In this chap¬
ter, we explore attitudes about other possible public access sites as well
as attitudes about libraries. Along with the data on instructional pref¬
erences, our findings on public access offer concrete guidance for pol¬
icy addressing the skills divide.
This chapter serves as further argument against those who say that
the gap is shrinking. If some individuals cannot use computer tech¬
nology, then all the access in the world will do no good. Further, if
people cannot find the assistance they need to use the technology,
then access alone does little to alleviate the problem.
Skills for the Information Age
The National Academy of Sciences has declared that for individuals to

“participate intelligently and thoughtfully in the world around them,
they need the ability to evaluate and select technological solutions to
problems.1 Their definition of “technological literacy” is a far-ranging
one, well beyond the more circumscribed set of digital skills that we
are concerned with in this chapter.2 Yet the academy’s appeal under¬
scores technology use as an intrinsic need in modern society.
Technical competence and information literacy, though linked,
represent distinct skill sets within this overarching idea of technolog¬
ical literacy. As skill sets, they should transcend operating system plat¬
forms, software programs, computer makes and models, and database
interfaces.
Technical Competence
Technical competence is the ability to operate a computerized or elec¬

tronic device. For example, using word processing software requires
technical competencies such as initiating a file; inputting, formatting,
and outputting data; and managing the file itself.
Little research addresses the extent of technical competence among
Americans, particularly among groups that are disadvantaged in
terms of home access. Both the Pew Internet and American Life pro¬
ject and the Commerce Department’s NTIA reports have docu¬
mented some of the ways in which the Internet is used, from which
we can extrapolate technical competencies. In a 2001 survey, Pew
found that 45 percent of Americans used e-mail, 36 percent used the
Internet for product and service information, 39 percent made online
purchases, and 35 percent searched for health information on the
Internet.3 The U.S. Department of Commerce has collected data on
the prevalence of different tasks in the workplace by occupational cat-
egory that suggest demand for certain competencies.4 For example,
managers and professionals were found to use the Internet and e-mail
at a 66.8 percent rate, word processing and desktop publishing at a
63.2 percent rate, and spreadsheets and databases at a 56.6 percent
rate.5 The extent of competency was not addressed in either study—
whether respondents who used e-mail could apply filters, organize
folders of mail, and open e-mail attachments, for example, or whether
they sometimes needed help with word processing or spreadsheets.
Nevertheless, simply being able to perform these tasks implies a cer¬
tain level of technical competence.
We examine both technical competence and a set of skills called
information literacy. With the advent of the Internet, the ability to
use technology requires not only technical acumen, but also the abil¬
ity to search for and use information.
Information Literacy
Information literacy is the ability to recognize when information is

needed and to locate, evaluate, and use effectively the needed in¬
formation while adhering to principles of social responsibility.6
Professional organizations associated with the American Library
Association have defined information literacy standards for K-12 and
for higher education. Although these standards were originally cre¬
ated with respect to students, they can easily be applied outside an
academic setting and have special relevance for an individual’s ability
to use information sources on the web. An information-literate per¬
son would have the capacity to do the following:
• Determine the nature and the extent of the information needed

• Access needed information effectively and efficiently
• Evaluate information and its sources critically and incorporate
selected information into his or her knowledge base and value
system
• Use information effectively to accomplish a specific purpose
• Understand many of the economic, legal, and social issues sur¬
rounding the use of information, accessing and using informa¬
tion ethically and legally.7
Computer technology may be used for locating, evaluating, and

using information, but a printed dictionary may also suit informa¬
tion needs. In fact, recognizing whether a printed or electronic source
is more appropriate is one hallmark of an information-literate person.
How is information literacy needed to use the web? Someone look¬
ing for information about a political issue might go to a common web
search engine such as www.yahoo.com. Choosing key words for the
search may require sufficient command of a topic to formulate the
question using the correct terminology. Next, the user may be
directed to consult an extensive list of directories. Choosing the
appropriate directory is one more step to take. Reliability or bias of
information is a common concern when consulting web resources,
although it may be an issue in using print resources as well. With the
fluidity of the web—the ease and low cost of widely disseminating
information—these concerns are magnified. Sources may have been
created by a scholarly research organization, a partisan group, or, as
mentioned in Next: The Future Just Happened, by a very bright four¬

teen-year-old.8
There are very few works that address information literacy levels
in the United States or in any country, for that matter.9 One reason
for this lack of research is the absence of measurement tools to assess
information literacy, although some efforts are being made to develop
the means to research this issue.10 Of the few studies that exist, most
of them are case studies that offer limited ability to assess the extent
of information literacy in the population as a whole. One study con¬
ducted in a workplace setting in Australia, however, demonstrated
that workers needed information literacy skills consisting of basic lit¬
eracy (reading skills) and the ability to access information using tech¬
nology, among other skills.11
Basic Literacy
Basic literacy is clearly a prerequisite for information literacy in the

context of using resources on the web. Not surprisingly, one study of
a technology center in a low-income community documented the
problems that individuals with poor reading skills had in under¬
standing content on the Internet.12 Comprehensive and systematic
data are available on literacy in the United States, and we can use this
research to draw some inferences about information literacy needs.
Funded by the U.S. Department of Education’s National Center for
Education Statistics, the National Adult Literacy Survey (NALS) was
conducted in 1992.13 A random sample of 26,000 American adults
was interviewed. The survey defines three aspects of literacy: prose,
document, and quantitative literacy. The first two categories pertain
to the reading and comprehension skills traditionally associated with
literacy, and the third is sometimes referred to as numeracy, or the
ability to use and understand numbers. For the purposes of under¬
standing information on the Internet, prose and document literacy
are most relevant. Prose literacy calls for locating, integrating, and
demonstrating comprehension of information in a variety of text
forms, such as newspaper editorials, poems, fiction passages, or
weather reports. Document literacy tasks include using forms or
graphically displayed information found in everyday life, including
job applications, payroll forms, transportation schedules, maps,
tables and graphs.’ 14 Literacy levels are scored separately for each type
of literacy, and scores range from level 1 to level 5, with level 1 rep¬
resenting the lowest level of attainment.1^
Results from the 1992 NALS are stunning. Approximately 21-23
percent of the adults in this country have skills at the lowest level of
prose, document, and quantitative literacy.16 Level 1 document liter¬
acy skills, for example, include the ability to complete a brief job
application with just a few items or to sign on the correct line on a
social security card. Prose literacy skills at this level consist of the abil¬
ity to locate a single item in a short passage of text. An additional
25—28 percent were able to demonstrate skills at level 2, which
required slightly more advanced, but still limited, reading and prob¬
lem-solving abilities.17 The lowest levels of attainment are of partic¬
ular concern, because more than a fifth of the population was unable
to perform simple daily tasks adequately. Limited fluency in English
accounted for some, but certainly not all, of the results for the low¬
est categories of literacy.
What are the consequences for individuals with limited literacy,
and for society? Although the results suggest startlingly low levels of
literacy in the United States, participants were also asked whether
they felt incapable of day-to-day functioning because of their literacy
skills. For the most part, even those scoring at the lowest levels said
they felt fairly comfortable coping with daily demands and expressed
little need for assistance. The poor and near-poor have levels of liter¬
acy well below average, however, demonstrating a connection
between literacy and earning ability.18
As the authors of the NALS study note, “If large percentages of
adults had to do little more than be able to sign their name on a form
or locate a single fact in a newspaper or table, then the levels of liter¬
acy seen in this survey might not warrant concern. We live in a nation,
however, where both the volume and variety of written information
are growing and where increasing numbers of citizens are expected to
be able to read, understand, and use these materials.”19
There are obvious parallels between the literacy definitions used by
the NALS and the concept of information literacy. For example, the
NALS was concerned with the ability of readers to process a number
of categories or features, to decipher central information despite irrel¬
evant distractors, and to cope with text that is lengthy and dense.
These are prerequisites for understanding information on many
Internet sites, which feature distractors such as advertisements or
formats that are less than user-friendly. Beyond, this, however, Inter¬
net users need skills in searching for and evaluating information. Both
of these require knowledge and experience, as a large body of research
on information seeking shows.20
Although basic literacy appears to be most crucial to Internet use,
its relationship to technology skills goes further. Anyone who has
used technology of any sort knows that troubleshooting and problem
solving are necessary at some point. Basic literacy is required to read
manuals or online instructions. Being literate makes it possible to
solve a problem or to learn the use of a more complicated device. Hav¬
ing noted the significance of basic literacy for technology use, we
focus, however, on technical competence and information literacy.
Survey Results: Skills
In this section, we present data on skills needed to use computer tech¬

nology from our national telephone survey. In order to assess the
prevalence of both technical competence and information literacy, we
asked a number of questions in our survey about the assistance indi¬
viduals would need. The simple percentages, presented below, show
ample need for assistance. More than half of the respondents reported
some need for assistance with the most complex applications.
A substantial portion of our survey respondents reported needing
assistance to carry out basic computer tasks. Nearly 22 percent said
that they needed assistance using the mouse and keyboard—the most
simple and yet fundamental skills involved in operating a computer.
As tasks required more sophistication, the percentage of respondents
needing help rose, with well over one-third needing assistance to find
information on the web, find books, or do homework. One interest¬
ing finding was that the responses for our measures of information
literacy were almost exactly the same, even though the questions were
not asked consecutively.21 For technical competence, there is a clear
hierarchy of tasks. More than half the survey respondents require
assistance to use common software applications such as word pro¬
cessing and spreadsheets. Looking at the simple percentages, we get
a picture of a substantial minority that needs help taking the very first
steps toward technical competence and larger segments of the popu-
WHAT MATTERS
Technical Competence
Do you need assistance doing the following computer tasks?

Yes
Using the mouse and keyboard 22%
Using e-mail 31%
Using word processing/spreadsheet programs 52%
Information Literacy
Do you need assistance doing the following computer tasks?

Yes
Finding books in a library 37%
Doing homework 37%
Finding information on the web 37%
lation needing help with more sophisticated applications, including

those involving information literacy.
Using multivariate regression, we are able to find out which groups
in the population are most likely to need help. We created an index
for “technical competence,” using the first three questions listed in
the box above (assistance for using the mouse and keyboard, assis¬
tance for using e-mail, and assistance for using word processing and
spreadsheet programs). We created a similar index for information lit¬
eracy, using the responses for books, homework, and the Internet.
The same explanatory variables used in the analysis of the access
divide are included in our examination of data in this chapter, with
one exception. We add the variable “library patronage,” to find out
whether individuals who use the library frequently are those who are
most likely to need assistance. Other explanatory variables are gen¬
der, race, ethnicity, income, education, and age. The education vari¬
able is measured on a scale ranging from less than a high school degree
to postgraduate study.22
The regression analyses reveal a skills divide that largely mirrors the
access divide.2^ Individuals who are low-income, less-educated, older,
and African American or Latino are more likely to need more com¬
puter assistance, controlling for other factors. This is true whether
“skills” are defined as technical competence or information literacy.

The results suggest that developing these skills is a complex process,
involving access to computers, frequency of use, and also education
or cognitive skills such as reading. Frequent library patrons, for exam¬
ple, are less likely to report needing assistance. Although there is no
longer a gender divide in access, women are slightly more likely to
report needing assistance for technical skills. In light of a prior gen¬
der gap in access, some women may have limited experience and less
confidence about their technical skills. There are no gender differ¬
ences for information literacy. All of these measures involve self-
reported assessments of skills, of course, and therefore they are better
viewed as perceived needs rather than objective assessments of the
skills differences between groups.
We used Monte Carlo simulations to calculate the magnitude of
skills differences—that is, the probability of needing computer assis¬
tance, holding other factors constant. Our hypothetical respondent
is a white female who is a library patron with average education and
income and of average age. We ran both the regression analyses and
the simulations with the indices of technical competence and infor¬
mation literacy combined into a single index of skills, and also as sep¬
arate indices. With the exception of gender, results were similar
whether the indices were run separately or together. For the sake of
simplicity, we present the results as a single index, except in the case
of gender, where the results are for technical competence only.
While many factors influence skills, differences based on age and
education are sharper than those based on other criteria. This is most
noticeable for age, where there is an expected 32 percentage point dif¬
ference between younger and older respondents. The predicted prob¬
ability that a young individual, defined as age twenty-eight, will
report needing assistance is only 20 percent, compared to an older
individual, age sixty-one, who has a 52 percent probability of need¬
ing assistance (controlling for other factors). The predicted impact of
education is somewhat less, 18 percentage points rather than 32. The
expected probability that an individual with only a high school
diploma will report needing assistance is 43 percent. For individuals
with a college degree, the probability of needing assistance is 25 per¬
cent. Predicted differences for other factors were also significant, but
more modest—a 12 percentage point difference between low- and
high-income respondents, an 11 percentage point difference between
WHAT MATTERS
Who Is Most Likely to Need Assistance (technical competence
and information literacy)?
Old (52% for 61-year-olds vs. 20% for 28-year-olds)—32-point

difference
Less-educated (43% for high school diploma vs. 25% for bachelor’s
degree)—-18-point difference
Poor (42% vs. 30% for affluent)—12-point difference
African Americans (45% vs. 34% for whites)—11 -point difference
Latinos (42% vs. 34% for non-Latinos)—8-point difference
Who Is More Likely to Need Assistance for Technical Competence?
Females (34% vs. 33% for males)—I-point difference

No gender difference for information literacy
Note: Estimates are based on a hypothetical respondent who is female, white, and
a library patron, with education, income, and age set at their means.The only sta¬
tistically significant differences are the ones reported above (see tables A3.1 and
A3.2).We have calculated the probability that respondents agree with the above
statements, controlling for other factors.
African Americans and whites, and an 8 percentage point difference

between Latinos and whites. For technical competence, the difference
between men and women was statistically significant, but decidedly
minor—only a little over 1 percentage point. Overall, then, skills dif¬
ferences largely duplicate disparities in access. This indicates that
public policy enhancing both access and skills development would be
effective if targeted to low-income communities in particular, where
many residents also have lower levels of education.
Public Policy Addressing Skills: Libraries

and Community Technology Centers
Despite the absence of data on technical competence and informa¬

tion literacy skills in the United States, some programs are intended
to promote skills as well as access. Both government and nonprofit

efforts have focused on two venues for skills development among
adults—public libraries and community technology centers (CTCs).
CTCs may be located in a multiservice agency, a stand-alone com¬
puting center, housing development, church, library, or other places
in low-income urban and rural communities. There are approxi¬
mately 400 federally funded CTCs across the country, and some cen¬
ters are supported by nonprofit organizations as well as grants from
private foundations.24
Libraries provide the most extensive network for public access and
are an important source of computer assistance. Almost half of the
public libraries in the United States now offer some type of computer
training, either in technical skills or in the skills needed to evaluate
information on the web.25 We know little, however, about the extent
of the training offered, its quality, its availability in poor communi¬
ties, or whether the supply of training satisfies the demand. Both the
public and the library profession strongly support the technological
function of libraries.
The role of the public library as “safety net” for information have-
nots is an important one in the eyes of many library leaders and library
users. A Benton Foundation report, based on a public opinion poll and
interviews with library leaders, provided a sense of the place the public
library has now and in the future in American culture. Some of the roles
of the public library include providing a physical place for a commu¬
nity to gather, “the library as a provider and protector of equal access
and equal opportunity; [and] the library as community builder, civic
integrator, and community activist in a digital world.”26
In the foundation’s national public opinion poll of 1,015 adults,
respondents indicated a strong public support of the library as a
provider of access. For example, 85 percent of the respondents
believed that it is either very important (60 percent) or moderately
important (25 percent) for a public library to spend its money “pro¬
viding computers and online services to children and adults who
don’t have their own computers.”27 Similarly, 86 percent of the
respondents thought the library should spend its money “providing
a place where librarians help people find information through com¬
puters and online services.”28
A few studies have examined the activities and impact of CTCs.
One survey of 123 CTC administrators found that access to technol-
ogy and the Internet were the most important functions of CTCs but
that literacy classes and tutoring were also provided at some centers.29
The CTCNet Research and Evaluation Team conducted a survey of
817 people at forty-four CTC affiliates in 1998. Findings from this
study indicate that activities in CTCs include the development of Eng¬
lish-language skills, homework, GED (general education develop¬
ment, for high school equivalency) courses, use of e-mail, Internet
searches, and the creation of webpages. Reasons for taking classes at a
CTC were dominated by job-related needs. For example, “65 percent
of respondents took classes at a technology center to improve their job
skills [and] 30 percent used the Internet at their center to look for a
job.”30 More than a third of the respondents felt they had improved
their job skills or had success in finding a job as a result of programs
at the CTC. Participants also felt more positive about their ability to
learn new skills or pursue further education. More than half of the
respondents rated “finding out about local events, local government,
or state/federal government as important reasons for coming to their
center” (ctcnet.org/impact98.htm).31 Some of these CTC users—
about 23 percent—used computers in their own homes but apparently
went to the CTC for training. The top reason for attending a CTC
was its “comfortable, supportive atmosphere,” and 94 percent of the
CTC users who participated in the study were overwhelmingly sup¬
portive of the CTC and its programs.
This federally sponsored evaluation is interesting for the picture it
paints of activities within CTCs, but it is important to note that many
areas lack such facilities. Community-based programs are often small
and frequently rely upon unstable funding sources and part-time or
volunteer staff. A Los Angeles-area study conducted by the Tomas
Rivera Policy Institute at Claremont College found that, in many
poor neighborhoods, the public library offered the only available
public access.32
The sparse evidence on CTC and library technology programs
affords little guidance for policy. Does it matter where access and
skills programs are available—would more people attend them at
libraries or at other places in the community? Do disadvantaged
groups view libraries, where the majority of programs are located now,
in a negative light? Help with computer skills can take several forms,
from personal instruction to manuals. We also wondered whether any
of these methods were preferred over others and whether the choices
made by disadvantaged groups differed from the population as a

whole.
Survey Results: Preferences

for Public Access and Learning
As part of our survey, we asked questions about willingness to use dif¬

ferent locations for public access, and as reported in the next section,
about preferences for learning new skills.
Public Access Sites
The majority of respondents were willing to go to a number of places

to use computers and the Internet: recreation centers (64 percent),
senior centers (60 percent), churches (66 percent), and government
offices (59 percent). The most popular sites, however, were schools
(74 percent) and public libraries (93 percent). Schools and libraries
are public institutions that bear responsibility for education and that
could be expected to offer assistance and expertise. They have also
been a focus of government programs to increase the number of com¬
puters in low-income communities, although not all school computer
facilities are open after hours or available for the use of nonstudents.
Do any disadvantaged groups differ from the general population in
their willingness to use public access sites? Six questions from our sur¬
vey measure willingness to use computers and the Internet in differ¬
ent public places—recreation centers, senior centers, local churches,
government offices, libraries, and public schools after hours. We com¬
bined these questions to create an index ranging from 0 to 6 of
reported willingness to use public access. The index serves as the
dependent variable in our statistical model. The explanatory factors
included in the model are the social and demographic variables used
in the multivariate regression model for the access divide.
We found that the individuals who expressed the most willingness
to use public access sites were more affluent, better educated, and
African American.33 Controlling for income and education, African
Americans are more likely than whites to express interest in using a
range of public access sites. Asians were less likely than other racial
groups to express willingness to use public access. Unfortunately,

however, the low-income and less-educated individuals for whom
public access is intended are least likely to be willing to use public
access sites. And, we know from our question about the places where
people actually use computers, that only about 13 percent of those
who have no home computer use library computers, in contrast to 15
percent of those who do have home computers.
Libraries are already cornerstones of public access policy and were
the most preferred setting for public access. Are there ways in which
use of the library could be expanded or improved, particularly for dis¬
advantaged groups? Perhaps. According to our survey, they are among
the groups in the population who are most likely to think of libraries
as “community gathering places.” Only slightly more than half (51
percent) of respondents viewed libraries as community gathering
places, yet low-income individuals, women, African Americans, and
Latinos were most likely to see libraries in this manner, controlling
for library patronage and other factors.34 This image may be one that
libraries should capitalize on in poor and minority communities, par¬
ticularly in efforts to expand technology access. The CTC evaluations
indicate that sociability and a “comfortable, supportive” atmosphere
WHAT MATTERS
Who Is Most Willing to Use Public Access?
Affluent
Educated
African Americans
(Asian Americans were less willing than whites to use public access.)
Who Sees Libraries as “Community Gathering Places”?
Poor
Women
Latinos
African Americans ___
(table A3.3). When multivariate regression is used, these are the variables that
_
were critical reasons for using the facilities, and this is likely impor¬
tant in providing assistance in other venues, such as libraries.
Instructional Preferences
We also asked our respondents how they felt about various methods
of providing assistance—one-on-one help, group ihstruction, online
help, and printed manuals. We wondered whether any particular
method of instruction was preferred in general, or by specific groups,
and whether any particular method posed obstacles for learning for
specific groups. Such information could be used to improve assistance
provided in libraries, CTCs, and other settings. Although many peo¬
ple were willing to use any form of assistance, negative reactions rose
as instruction became less personal and hands-on. A majority of
respondents was willing to use any method of assistance, but a size¬
able minority expressed discomfort with less personal methods, par¬
ticularly those that require more reading and self-help. Support for
online instruction and printed manuals was weakest (about two-
thirds) and support for one-on-one instruction was highest (89 per¬
cent). The simple percentages for instructional preferences are
reported in the box below.
To explore the differences between groups, we used multivariate
regression to analyze responses to each of the four types of instruc-
WHAT MATTERS
Would you be willing to learn a new skill through
One-on-one instruction 89% agree; 6% disagree

Group instruction (such as a class) 78% agree; 16% disagree
Online help or tutorials 64% agree; 22% disagree
Printed manuals _ 67% agree; 23% disagree
Note: We have combined the “strongly agree” and “agree” responses and the
“strongly disagree” and “disagree” into the categories agree and disagree.The per¬
centages total less than 100% because we have omitted missing responses as well
as those recorded as “neutral” or “don’t know.”
tion. The explanatory variables used for each question were the demo¬
graphic variables used for the access divide, and library patronage was
used as a control.35
For instructional preferences, we found that certain groups were
more likely to be reticent about some methods of assistance. Women
were more likely than men to prefer one-on-one instruction, whereas
men liked do-it-yourself methods such as online instruction and
tutorials. Income mattered for only one category, with poor people
more hesitant than affluent respondents about using computer-based
help such as online instruction and tutorials. Older individuals and
WHAT MATTERS
Who Is Most Likely to Prefer Learning New Skills in This Way?
One-on-One Instruction
Educated
Young
Women
Group Instruction
Educated
Young
Latinos
African Americans (Asian Americans are less likely to prefer group
instruction than whites.)
Online Instruction orTutorials

Affluent
Educated
Young
Men
African Americans
Printed Manuals
Young
African Americans _
(see table A3.4).When multivariate regression is used, these are the variables that
less-educated individuals are more inclined toward rejecting mosqor

all methods of instruction. This may indicate a general anxiety about
computer instruction or a lack of interest in learning computer skills.
Some interesting patterns emerged with regard to race and eth¬
nicity. Controlling for other factors, African Americans have even
more positive attitudes about learning computer skills than the gen¬
eral population. African Americans were similar to whites in terms of
personal instruction and were less likely than whites to oppose other
methods. Both Latinos and African Americans were less likely than
whites to oppose group instruction or computer classes. Asian Amer¬
icans, however, were less likely than whites to be willing to participate
in group instruction. Overall, this suggests that a range of approaches
is needed, particularly for men and women. Education and age clearly
affect general willingness to receive instruction. Racial and ethnic
groups reporting lower skills—African Americans and Latinos—are
not particularly opposed to any method of instruction, compared
with whites. African Americans, in fact, are less likely than whites to
express opposition to several types of instruction.
Conclusion
This chapter serves as further argument that an information tech-

nology gap indeed exists and is characterized by a skills divide as well
as an access divide. The two are not entirely separable in practice, of
course, but the implications of a skills divide lie at the heart of the
policy problem—do individuals have the skills they need to partici¬
pate fully in society, particularly in the economic and political arenas?
Without appropriate skills, access is meaningless. Information tech¬
nology requires a set of computer-specific skills that we call technical
competencies, but many uses of the Internet also demand more gen¬
eral information literacy, rooted in basic literacy. The proliferation of
websites offering information on health, jobs, parenting, political
issues and campaigns, homework topics, home remodeling, cooking,
money management, and an unfathomable number of subjects
requires the ability to locate, read, comprehend, and evaluate infor¬
mation (both its content and its source).
Our findings on skills show a sizeable minority bereft of the most
basic technical abilities. About one-fifth of the population reported
needing assistance using a mouse or a keyboard. Likewise, more than

one-third of respondents felt they needed help in negotiating their
way through information sources on the web and in databases. Pub¬
lic provision of access accompanied by on-site assistance offers oppor¬
tunities to teach basic skills, to provide more sophisticated technical
support, to instruct individuals in using search engines and evaluat¬
ing sources, and to help with specific information searches. Some
libraries and CTCs are already providing library Internet service,
though we know little about the extent or quality of the help that is
offered in most of these settings.
Large majorities of individuals expressed willingness to use com¬
puters and the Internet at public access sites, and the majority of
respondents showed flexibility in expressing willingness to use multi¬
ple methods for learning new skills or coping with computer prob¬
lems. Respondents were most amenable to personal instruction and
classes and slightly less favorable toward self-help in the form of online
instruction, tutorials, or manuals. More sobering, though, is the real¬
ity revealed in the prior chapter, which showed that only about 10-15
percent of the population uses technology resources at libraries, which
are the most common site for public access. People without home
access are slightly less likely to use such services. Apparently, the will¬
ingness reported in our survey is seldom translated into deed.
By disaggregating this data, we get a better picture of who lacks
skills and the preferences that these same groups have regarding pub¬
lic access and assistance. For the most part, the skills divide replicates
the access divide—those who lack skills are older, less-educated, poor,
African American, and Latino. The same pattern of disparity charac¬
terizes both technical competence and information literacy. Although
there was a small gender difference in technical competence, con¬
trolling for other factors, it was too small to raise much concern about
a skills gap affecting women.
Confirmation of a skills divide that parallels the access divide pre¬
sents a policy challenge. Even if computers are becoming increasingly
affordable, some individuals may face significant hurdles in using
them, particularly if those left behind are the least educated with the
fewest resources for learning through trial-and-error and through the
manuals and tutorials provided by manufacturers. The web offers new
opportunities for enriching people’s lives in many ways, but it also
poses more skill demands than simple computer tasks. The policy
challenge is how to offer assistance to those who lack skills, at least*to

those who are interested. The poor and the least educated are less will¬
ing to use public access sites, where assistance could be provided.
Older and low-income individuals are also less willing to learn new
skills, in whatever form. These findings, however, need to be placed
in the context of the favorable attitudes that most people had about
public access and instruction. Those who are poor, older, and less-
educated are somewhat less positive.
Race and ethnicity influence preferences for addressing the skills
divide, but the news here is hopeful rather than disheartening.
African Americans were more willing to use public access and more
positive about various methods of assistance than white respondents.
Latinos expressed slightly more positive attitudes toward libraries as
community gathering places than non-Latinos, but were equally will¬
ing to use public access. Latinos preferred group instruction to a
greater extent than whites did but differed little in other respects.
Although Asian Americans were more likely to reject public access
and group classes, they enjoy higher rates of home access than did
whites and are not disadvantaged in terms of skills.
On a final note, gender made a difference in learning preferences.
Men were more likely to want to find their own way with online help
and tutorials, while women tended to choose personal instruction.
This suggests that personal assistance and public access may be more
important for women or used as occasional technical support by men.
Having established the outlines of the access and skills divides,
we turn next toward their implications for daily life—for earning
a living and exercising the rights and responsibilities of democratic
citizenship.
Notes
1. Pearson and Young 2002, 3.

2. See Pearson and Young 2002, 3. The National Academy of Sciences
provides a three-part framework for technological literacy that includes
knowledge, ways of thinking and acting, and capabilities that apply to any
type of technology, not just information technology (1996, 4). The Interna¬
tional Technology Education Association (ITEA) has defined standards for
technological literacy in K-12 education (International Technology Educa-
tion Association 2002). There are twenty different standards, including

knowledge of historical and cultural attributes of technology, engineering
design, and transportation, manufacturing, and construction technologies.
3. Horrigan 2002. The survey was conducted by Pew researchers in
2001.
5. Horrigan and Rainie 2002.
6. American Library Association 1989.
7. In 2000, the Association of College and Research Libraries, a divi¬
sion of the American Library Association, created Information Literacy
Competency Standards for Higher Education, upon which the above defini¬
tion is based. Similar competencies have been defined for the K-12, as well
as adult, populations. The K-12 population is addressed by the American
Association of School Librarians (1998) in Information Power: Building
Partnerships for Student Learning with its information literacy standards for
student learning.
8. Lewis 2001.
9. The absence of data applies for information literacy in the K-12 set¬
ting, higher education, or adults in the workplace. The information literacy
movement was born in school library media centers, and there is an exten¬
sive body of publication on the topic. These publications, however, tend to
be either tool kits and worksheets for teaching information literacy skills or
professional editorials and opinion pieces about the necessity of teaching
information literacy skills. There are some case studies and anecdotal reports
of information literacy programs, but there is no body of data.
10. A promising project undertaken by Kent State University Libraries
and Media Services faculty is the development of a measurement tool for
higher education. This tool, tentatively titled Standard Assessment of Infor¬
mation Literacy Skills (SAILS), is being considered for broad use by mem¬
ber institutions of the Association of College and Research Libraries. If this
tool is adopted and widely applied, data can be collected and analyzed for
establishing norms of information literacy skills. Until that adoption occurs,
though, there is no source of data for describing the level of information lit¬
eracy skills.
11. McMahon and Bruce (2002) published a study of information liter¬
acy needs. Their phenomenographic approach was intended to derive a bet¬
ter understanding of the perception of information literacy needs among staff
participating in community development projects in Australia. Using inter¬
views with five development workers, MeVUhon and Bruce constructed a
multilayered model of necessary information literacy skills for one particu¬

lar work setting. Their model, which is based on the perceptions of the work¬
ers, began with basic literacy. From basic literacy outward, the model
incorporated workplace system skills, information technology skills, cultural
experiences, ability to translate between cultures, information presentation,
and relevance to the community being served. A broadening of the initial
model accommodates five layers of information literacy: (1) basic literacy
skills, (2) understanding workplace systems, (3) communication skills, (4)
accessing information sources, and (5) understanding the dominant society.
12. Penuel and Kim 2000.
13. Another survey is scheduled for 2002.
14. Kaestle et al. 2001.
15. Ibid.
16. The range of percentages reported reflects estimates for the whole
population based on the sample of 26,000 participants.
17. Level 3 was achieved by about 61 million adults, which constitutes
nearly one-third of participants. Levels 4 and 5 were attained by 18-21 per¬
cent of the participants, with only 3 percent of adults performing at the
highest level.
18. National Center for Education Statistics [no date],
19. Kirsch et al. 2002, xxi. For a complete description of the definitions,
questions and tasks used in the survey, and the results of the 1992 study, con¬
sult Kaestle et al. 2001.
20. The vast body of information-seeking research is primarily out of the
disciplines of library science, information science, communication science,
cognitive psychology, cognitive science, human-computer interaction, and
education.
21. Percentages are rounded. There was some minor variation in the fre¬
quency of responses for the three questions.
22. Library usage was measured on a five-point Likert scale based on the
number of times the respondent reported going to the library in the last
month. 0 times, 1—5 times, 6—10 times, 11—15 times, and more than 15
times. Dummy variables were created for gender, race, ethnicity, and
income. Women, African Americans, Latinos, Asian Americans, and those
with annual incomes less than $30,000 were coded 1; otherwise, respon¬
dents were coded 0. For race/ethnicity, non-Latino whites are the reference,
or left-out, group. Education was measured on a five-point Likert scale, with
responses coded 1 for less than a high school degree and ranging to 5 for
postgraduate work. Age was recorded in years.
23. See tables A2.1 and A2.2 for the results of the ordered logistic regres¬
sion analysis.
24. For a description of the CTC program, see U.S. Department of Edu¬
cation 2002.
23. Trotter 2001.
26. Benton Foundation 1996, 11. The report was funded by the W. K.
Kellogg Foundation.
27. Ibid., 44.
28. Ibid., 44.
29. Servon and Nelson 2001, 285-86.
30. Chow et al. 1998.
31.Ibid.
32. Trotter 2001.
33. For the results of the ordered logistic regression analysis, see table
A3.3.
34. These results are based on logistic regression reported in table A3.3,
using a binary dependent variable (agree or disagree that the library is a com¬
munity gathering place). The independent variables used in the multivari¬
ate model were income, education, age, gender, Latino, African American,
and Asian American, as measured in the access chapter, with the addition of
library patronage as a control variable. Library patronage was measured on
an ordinal scale as the number of times the respondent had visited the library
in the past month.
35. The results of the multivariate regression analysis are reported in table
A3.4 We used negative responses for coding the dependent variable,
although, for clarity, we present the results in the positive. (For example,
where poor individuals were less likely to favor a certain method of instruc¬
tion, we show that affluent individuals prefer that method.) Since the major¬
ity of respondents supported any single type of instruction, we were most
interested in whether certain groups of respondents were likely to shun any
particular method.
flap-tir 4
The Economic
Opportunity Divide
A consensus has emerged among observers in academia, govern¬

ment, and business that there is indeed something new about
the economy and that technological change is one of the fac¬
tors underlying this transformation. On the left, Manuel Castells has
characterized the present period of capitalist development as “infor-
mationalism. On the right, William Niskanen, conservative econo¬
mist and director of the Heritage Foundation, calls the current era the
“third industrial revolution,” ushered in by digital technology and
biotechnology. The U.S. Department of Commerce has dropped the
term “emerging” from its reports on the “digital economy.” And the
Conference Board, a business organization whose economic reports
are closely watched by policymakers and industry, has compared
information technology to “steam power at the turn of the nineteenth
century and electricity at the beginning of the twentieth.”^
What skills do workers need to survive, and perhaps even prosper,
in this new economy? Are information-technology skills linked to
economic opportunity; if so, how are they related? Does an “eco¬
nomic opportunity divide,” based on lack of computer skills, exist?
The first section of this chapter traces the trajectory of economic
change over the past few decades. A broad economic restructuring has
widened economic disparities, automated some jobs out of existence,
created new types of jobs, modified organizational practices, and
altered traditional career ladders. In the new economy,” workers are
more likely to hold a number of jobs over a lifetime. Less-educated
workers have watched their standard of living erode, and skills
demands are increasing even in jobs requiring only a high school
degree or less. We contend that computer skills must be viewed as one

element of changing skills requirements in the new economy and that
they are particularly important for mobility beyond the lowest-pay¬
ing jobs.
Against this backdrop of economic change, we discuss our data on
the attitudes and experiences of our respondents in regard to com¬
puters and economic opportunity. Technology has the potential to
enhance economic opportunity in three distinct ways: for use on the
job, for finding a job, and for learning. First, technology skills can
facilitate employment or advancement in jobs or business. Second,
the Internet offers tools for searching for a job and for gathering infor¬
mation about occupations and their requirements. Third, computer
software and the Internet provide users with choices for improving
other basic skills or earning educational or occupational credentials.
Lifelong learning promises to be an important consideration for eco¬
nomic advancement now and in the foreseeable future.
We ask respondents about their attitudes regarding computer skills
and economic opportunity in all three of these areas, as well as their
experiences in using technology for job searches and online courses.
Our low-income sample allows us to explore the perceptions and pos¬
sible motivations of those who are most likely to lack computer skills.
Do disadvantaged groups see information technology as less impor¬
tant for economic opportunity than do other Americans? Are they
indifferent to its potential for economic advancement? What are the
characteristics of those who have used new technologies for job search
and learning, and how interested are disadvantaged groups and other
Americans in using the Internet for these purposes? Some major stud¬
ies have tracked the way in which Americans are using the Internet,
but none of these previous studies examine the attitudes of both users
and nonusers.2 Understanding the priorities and aspirations of
nonusers or those with limited skills suggests future directions for
public policy.
The beliefs expressed by survey respondents demonstrate that the
problem lies not with limited awareness of technology’s benefits, but
with issues of access and skill. An economic opportunity divide does
in fact exist, but some disadvantaged groups see technology in a par¬
ticularly positive light, and those attitudes may help to eventually
bridge the divide.
Economic Opportunity in the “New Economy”

*
Rapid growth in the productivity of the U.S. economy in the late

1990s resulted primarily from a groundswell in information technol¬
ogy use across a wide range of industries rather than from growth in
firms that produce hardware or software.3 Despite the demise of the
dot.com mania, technology has enabled broader innovations in pro¬
duction processes and products that will continue to affect the econ¬
omy. According to the international Organization for Economic
Cooperation and Development, innovations made possible by infor¬
mation technology “are transforming the ways in which economies,
and the people within them, are working.”4 Among these dramatic
shifts are rising skills requirements in the labor force and changes in
the opportunity structure for American workers.
Education and Economic Opportunity
Educational attainment is increasingly linked to both economic

opportunity and widening disparities in the labor market. Control¬
ling for race, experience, and other factors, male college graduates
averaged 42 percent more in earnings than male high school gradu¬
ates in 1999. This college “premium” has more than doubled since
1979, when male college graduates earned an average of 20 percent
more than did male high school graduates.5 The expanding wage dif¬
ferential is partly due to gains made by college-educated workers, who
experienced real growth in wages over this period. It also reflects dete¬
riorating earnings for every category of educational attainment below
a bachelor’s degree, particularly those with only a high school educa¬
tion or less.6
Income inequality has been growing in the United States for more
than two decades now, and it was only during the rapid growth and
tight labor market of the late 1990s that this trend abated somewhat.7
Two divergent tendencies characterized the 1980s and early 1990s:
rising incomes at the top of the income distribution and falling real
wages for many at the middle and bottom. The number of full-time,
year-round workers whose earnings fall below the poverty line has
been climbing since 1973, even when the figures are adjusted for
changes in the size of the workforce over this period.8 During the lat-
ter half of the 1990s, the flush economy began to improve the situa¬
tion of the poorest Americans, narrowing the gap between the bot¬
tom and middle of the wage distribution. Nevertheless, these changes
were insufficient to reverse the trends of the past few decades—
inequality today is still worse than in previous periods of economic
expansion, such as the 1960s. Because improvements for the lowest-
paid workers were due primarily to low unemployment, the econ¬
omy’s subsequent slide has jeopardized the gains realized at the close
of the 1990s.9 Recent reports indicate that the both the proportion
of Americans living in poverty and income inequality sharply
increased during 2001.10
Women and racial and ethnic minorities continue to lag far behind
white men in earnings, despite gaining some ground over the past few
decades. For racial and ethnic minorities, some, but not all of the dif¬
ferences are attributable to median age and education. Fewer African
Americans and Latinos of working age have bachelor’s degrees than
whites—only one-fifth of African Americans and one-seventh of Lati¬
nos, compared to one-third of whites.11
Education affects the ability of workers to chart a course through
an increasingly volatile labor market. Steady employment and up¬
ward mobility are more difficult for less-educated workers. Blue col¬
lar workers are most likely to be displaced through layoffs, although
during the 1990s white collar workers suffered from downsizing more
frequently than in previous decades. Women, nonwhites, and work¬
ers lacking college degrees are the slowest to find new employment
after a job loss.12 Approximately 11 percent of the workforce is per¬
sistently “stuck” in low-wage jobs for five years or more. Most fre¬
quently, these workers lack more than a high school education and
work in jobs that offer the least in the way of training and advance¬
ment. Even some postsecondary education makes income mobility
more likely.13 One study of career mobility and lifetime earnings
found that the penalty for lack of education experienced by workers
with a high school diploma or less actually grew with years of work
experience. 14
The growing returns to education are driven in part by techno¬
logical change and the demand for higher levels of skill in the work¬
force. Education is clearly an important dimension of opportunity in
the new economy. Flow do computer skills affect the fortunes of
American workers, especially those who are in nonprofessional jobs?
Looking at the Numbers:

Computers and the Workplace
Information technology has wrought a number of changes in the

workplace and continues to grow rapidly by several measures. The
U.S. Bureau of Labor Statistics, for example, projects that the fastest-
growing occupations through the year 2010 will include software
engineers, computer support specialists, network and systems admin¬
istrators, database administrators, systems analysts, and desktop pub¬
lishers. In fact, of the ten fastest-growing occupations, eight are
computer-related.
Technology makes its real impact felt, however, through its integra¬
tion in a variety of workplace tasks. Office work has been revolution¬
ized by word processing, spreadsheets, and databases. Advanced
manufacturing technology uses computers to control production
processes and to diversify and customize products. The trucking indus¬
try now uses software for planning routes, controlling costs, schedul¬
ing, and real-time tracking.15 Computers have also made possible new
practices such as “just-in-time” inventory, and this potential has been
expanded by the development of the Internet and e-commerce.
The diffusion of information technology throughout U.S. industry
is at an early stage, according to some observers. Digital technology
promises to be adopted in a broader range of applications in the
future.16 Data on the use of the Internet support this contention. In
the span of just thirteen months, between August 2000 and Septem¬
ber 2001, the percentage of employed adults twenty-five years of age
and older who use the Internet and/or e-mail at work rose from 26 per¬
cent to nearly 42 percent. This suggests that newer applications, espe¬
cially those using the Internet, are proliferating in the workplace.17
Computers and Less-Educated Workers
What these general data do not tell us, however, is what the acceler¬
ating diffusion of computer applications means for job applicants and
current workers who are stuck on the wrong side of the access and
skills divides. Jobs that require a high school diploma or less are com¬
monly referred to as low-skill jobs.111 How, then, do computer skills
affect the chances that workers have for obtaining these low-skill jobs
and for possibly advancing to moderately-skilled jobs that require
some postsecondary education, such as an associate’s degree, an
apprenticeship, or vocational training?
Nearly half of the workers in low-skill jobs currently use comput¬
ers in some way at work, according to our survey and other data col¬
lected over the past few years.19 Forty-eight percent of our
respondents who were employed and had a high school education or
less used the computer at work at least once during the month prior
to our survey. This was substantially lower than the 65 percent of all
employed persons who reported using computers at work, but it still
represents a near majority of noncollege workers. A smaller share of
noncollege workers reported intensive use of information technology
on the job—17 percent used the computer frequently (between 31
and 100 times) at work during the past month, and 10 percent used
it very frequently (more than 100 times). Our data, based on the
reports of individual workers, are consistent with employer surveys,
which show that 50 to 60 percent of jobs requiring a high school edu¬
cation or less involve some computer use. Our data strengthen the
case that computer use is significant in these types of jobs, because
they are based on reports of actual use rather than reported employer
requirements.
One of the problems with tracking only “computer use” is that
individuals and employers may be reporting simple applications, such
as optical scanning, that do not involve any of the skills discussed in
the last chapter. A more challenging threshold for measuring com¬
puter use, however, is Internet use on the job. The results of our own
survey reveal that a sizeable minority of employed workers with a high
school education or less—25 percent—used the Internet at work dur¬
ing the past month. This is significantly lower than the 45 percent of
all employed workers who reported using the Internet at work, but
clearly even less-educated workers are using relatively demanding
forms of digital technology.
Our survey data and other reports establish that computer use is
prevalent in the workplace. Computer use rises with educational
level, but even among those with only a high school diploma or less,
nearly half of those who work reported using a computer at work, and
one-quarter used the Internet. The recent surge of Internet use in the
workplace demonstrates that job requirements for technology use are

important and may become more widespread in the near future.
Are workers who lack any type of computer skill essentially shut out
of the job market? The answer is clearly no, based on our survey find¬
ings as well as employer surveys.20 Despite the increased use of tech¬
nology in the workplace, jobs requiring little computer use or very
simple computer use are among those projected to have the largest
growth (in terms of the actual number of jobs rather than the rate of
growth). These include fast food workers, retail salespersons, cashiers,
security guards, and waiters and waitresses.21 The problem, then, is not
so much a matter of access to any employment. A number of low-skill,
low-wage jobs with limited upward mobility will continue to be avail¬
able to workers lacking even the most rudimentary computer skills.
Workers with no computer skills, however, may be less able to
move into jobs that offer better pay and benefits. Whereas service
workers, factory workers, and laborers have the lowest rate of com¬
puter use in the workforce, the demand for computer use is higher in
occupations requiring more skill or experience, such as skilled trades
or technical, sales, and administrative support.22 To examine the rela¬
tionship between computers and economic opportunity, however, we
need to look more closely at the issue of skills.
Rising Skill Requirements
Computer use is proliferating, but computer skills alone are not

enough to open the door to economic opportunity. There is a con¬
sensus among researchers that skills requirements are rising, even in
jobs requiring a high school degree or less.23 Computer skills are often
cited as the most common change in job requirements.2"1 But other
basic skills, such as reading or math, and soft skills,” or interpersonal
skills such as dealing with customers, are often at least as important
to employers.25 As the chapter on the skills divide demonstrates, tech¬
nical competence is necessary but not sufficient to use the Internet.
Likewise, technical competence should be viewed as just one part of
a package of skills that are increasingly in demand.
Technological change has helped raise skills requirements. Adapt¬
ing to new technologies requires basic literacy and numeracy_to
read and write instructions, study new training materials, and per¬
form simple math. Computers have facilitated the adoption of par¬
ticipatory “high-performance” workplace practices, such as total
quality management, which requires statistical control of production
processes. These organizational practices tend to upgrade necessary
skill levels for front-line workers.26 Some jobs have also experienced
an “upskilling” effect, where workers assume new, more demanding
responsibilities with the assistance of information technology. In the
insurance industry, for example, clerical workers using desktop PCs
now perform underwriting as well as basic data entry.27 The requisite
skills may include learning new software programs, but demands for
problem-solving, literacy, and numeracy are at least as critical.
A widely cited government report released in the early 1990s made
the case that improving a broad range of skills throughout the work¬
force was critical for competition in a global economy. The Secretary’s
Commission on Achieving Necessary Skills (SCANS) defined a num¬
ber of core workforce competencies for the new economy, on the basis
of research and consultation with employers. The SCANS require¬
ment for information competence is similar to technical competency
and information literacy as defined in the previous chapter. Workers
meeting these standards “can acquire and evaluate data, organize and
maintain files, interpret and communicate, and use computers to
process information.”28 Such competencies, however, rely upon
foundation skills such as reading, writing, and arithmetic.
Computer and cognitive skills may be more valued in jobs that
offer better pay and benefits for less-educated workers. One study
found that jobs requiring only soft skills were often part time, with
no benefits and median pay just above the minimum wage.29 Another
study found that companies in rural areas that required computer,
reading, writing, and math skills were often larger firms that paid
their employees more.30
Computer skills are best viewed as part of a package of basic skills,
along with literacy and numeracy, which can enhance an individual s
employability for a broader range of occupations. Moreover, techno¬
logical competence and information literacy can be valuable when
searching for better jobs or for learning. Computers can help work¬
ers hone basic skills, take advantage of job-specific training opportu¬
nities, and further their education.
Attitudes about Computer Skills *

and Economic Opportunity
What do we know based on the survey, about beliefs regarding infor¬

mation technology and opportunity? Are those who are less likely to
have access and skills also less interested in technology? In particular,
are they less likely to be motivated by the prospect of economic
returns for computer skills? The attitudes expressed by respondents
reveal the relative value that different groups, especially those from
low-income communities, place on the connection between technol¬
ogy and opportunity.
A solid majority of all respondents—more than two-thirds—artic¬
ulated the belief that computer skills are linked to economic opportu¬
nity in a variety of ways. Of the 1,837 valid responses to the survey, 69
percent agreed with the general statement that it is “necessary for peo¬
ple to use the Internet to keep up with the times.” Similar majorities
also agreed that learning new computer skills was important for career
advancement. When asked, “Do you believe you need to learn new
computer skills to get a job?” 70 percent said yes. Similarly, 71 percent
thought learning new computer skills was important for securing a
higher-paying job, and three-quarters of respondents believed learning
new computer skills was necessary for starting a small business. A lower
percentage of respondents—61 percent—felt that learning new com¬
puter skills was necessary to obtain a promotion.
What do these findings tell us? The similarity between responses
for the general statement about “keeping up with the times” and the
responses to the various measures of economic advancement—get¬
ting a job, getting a promotion, getting a higher-paying job, and start¬
ing a small business—indicates that these individuals are expressing
general beliefs about technology and opportunity as well as com¬
menting on their own specific skills or needs. For example, although
the vast majority of respondents—70 percent—agreed with the state¬
ment “You need to learn new computer skills to get a job,” only 6.5
percent reported having been denied a job because they needed more
computer skills.
What factors account for variation in beliefs about the connection
between computer skills and economic opportunity? Using multi¬
variate regression, we compared the results of two models. In the first
one, the dependent variable, or the result to be explained, is the state-
ment that it is necessary for people to use the Internet to keep up with
the times. Positive responses were coded 1, and negative responses
were coded 0.31 For the second model, the dependent variable is an
index of responses to the following four questions measuring beliefs
about the significance of computer skills and economic opportunity:
(1) Do you believe you need to learn new computer skills to get a job?
(2) Do you believe you need to learn new computer skills to get a
higher-paying job? (3) Do you believe you need to learn new com¬
puter skills to get a promotion? (4) Do you believe you need to learn
new computer skills to start a small business?^2 For both models,
higher scores are associated with more favorable attitudes toward the
use of computer skills for economic opportunity.
The same explanatory variables used in the analysis of the access
divide are included in our examination of data in this chapter, with
one exception. We add the variable employment status so that we can
take into account whether respondents are employed. Other explana¬
tory variables are gender, race, ethnicity, income, education, and age.
The education variable is measured on a scale ranging from less than
a high school degree to postgraduate study.
WHAT MATTERS
Who Is Most Likely to Think You Need the Internet to Keep Up?
Latinos (80% vs. 65% for whites)—15-point difference

Who Thinks You Need More Computer Skills to Get Ahead?

Young (73% for 28-year-olds vs. 55% for 61-year-olds)—18-point
difference
Unemployed (74% vs. 67% for employed)—7-point difference
Females (67% vs. 63% for males)—4-point difference__
employed, with values for education, age, and income set at their mean for the
sample.The only statistically significant differences are the ones reported above
(see table A4.l).We have calculated the probability that respondents agree with
the above statements, controlling for other factors.
Our analysis indicates that race and ethnicity are particularly

important for predicting attitudes regarding information technology
and economic advancement. When we asked a general question
about technology use that was unrelated to the respondent’s beliefs
about his or her own skills, attitudes did not differ by age, education,
or income. Racial differences were the only statistically significant
predictors of beliefs about the Internet. After ,we controlled for
socioeconomic conditions, gender, age, and employment status,
African Americans and Latinos were statistically more likely than
whites to believe that it is necessary to use the Internet to keep up
with the times. On the basis of the multivariate analysis, we estimate
the probability of agreeing that the Internet is necessary to keep up
with the times as 80 percent for Latinos, 78 percent for African Amer¬
icans, and 65 percent for whites. The paradox is that Latinos and
African Americans have significantly lower access to the Internet and
report the need for more assistance in using it.
Similarly, African Americans (but not Latinos) were significantly
more likely than whites to view the acquisition of new computer skills
as a means of career advancement—for starting a business or for
obtaining a job, a higher-paying job, or a promotion. There was a 76
percent probability of African American respondents agreeing that
they needed to learn new skills for economic advancement versus a
66 percent probability of a similar response from whites.
Women, the young, and the unemployed were statistically more
likely to cite the necessity of learning new computer skills, although
they were not more likely to believe that using the Internet was nec¬
essary to keep up with the times. Women were slightly more likely to
state that new skills were necessary for economic advancement—with
a 67 percent probability of agreement with the statement, compared
with a 63 percent probability for men. The unemployed were also
more likely to agree with the statement “You need to learn new com¬
puter skills” for economic advancement. For the unemployed, the
probability of agreeing with this statement is 74 percent, versus 67
percent for employed respondents. In light of the fact that African
Americans are among the likely groups to report needing assistance
with computers, these responses may reflect some feelings about their
own personal skills as well as general beliefs about the significance of
these skills for economic opportunity.
The greatest differences in attitudes are based on age. The proba¬

bility that young respondents agreed with the statements in the index
was 72 percent versus 66 percent for middle-aged respondents and
55 percent for older respondents. For the young (age twenty-eight),
who are more likely to have both access and skills, agreement with the
statement reflects generally positive attitudes about the significance
of technology. Older individuals have possibly reached a point in
their careers at which they are less concerned with advancement and
learning new skills, even if they have not yet retired. Older respon¬
dents may also be more apathetic in general about technology use.
Surprisingly, attitudes about computer skills—using the Internet
to keep up with the times and the linkage between computers and
economic advancement—did not vary by income or education. Low-
income and less-educated individuals, therefore, are just as likely as
other Americans to value the potential connection between technol¬
ogy skills and economic opportunity. This reflects a broad consensus
on these issues and suggests that, for most groups (other than older
individuals), the access and skills divides are not usually the result of
indifference to information technology and its potential.
Economic Opportunity
and Information Networks
Labor force projections depict a future in which workers of all skill

levels will change jobs several times over the course of their careers.33
Computers and the Internet offer information for job search and net¬
working that can be advantageous to less-educated workers. Our sur¬
vey findings explore attitudes toward this new use of technology, as
well as experience with online job search.
Ongoing career development is more important in the new econ¬
omy for several reasons. First, as the earlier discussion of the diffusion
of technology indicated, continued change is projected in the types
of jobs that are in demand. Second, evidence from the past few
decades reveals uan increasingly turbulent labor market with few
guarantees of job security.34 Despite economic expansion in the
1990s, the job dislocation rate increased.35 Corporate reengineering,
relocation, and automation have exacerbated the churning of firms
that some observers see as a hallmark and long-term attribute of the

new economy.36 Third, changing jobs represents a strategy for career
mobility, as career ladders within firms are less prevalent than in the
past.37 Management positions often require a college degree, and
some mid-level positions have disappeared, with responsibilities
pushed downward in the organizational hierarchy.38 Research on
career mobility demonstrates that workers with high school diplomas
who escape the low-wage threshold do so primarily by finding new
jobs outside low-paying industries rather than through promotion
within the low-wage sector.39 Likewise, some studies indicate that
former welfare recipients who voluntarily change jobs tend to have
higher wages than those who persist in their original placement.40
Digital technology offers a new portal for career information and
job search. For-profit websites such as Monster.com and Career-
builder.com list hundreds of thousands of job openings and provide job
seekers and employers with a variety of services. The “biggest and
busiest job market in cyberspace” is Americas Job Bank, sponsored by
the U.S. Department of Labor.41 The job bank lists more than a mil¬
lion openings and is linked to several companion websites collectively
called Americas Career Kit.” Through these partner sites and other
links, job seekers can do all of the following online: research occupa¬
tional requirements and salaries; find information on more than
350,000 training and educational programs; apply for admission and
financial aid for a college program or a single web-based course; find
advice on job search, interviewing, and resume writing; submit
resumes; store applications in a customized database; receive automatic
updates on jobs fitting their profile; and create customized searches by
job title, geographic region, salary, and educational requirements.
Americas Job Bank includes jobs at all educational levels. A recent audit
of the site turned up 550,000 jobs requiring a high school diploma or
less. Local workforce development efforts also employ information
technology. San Diego’s public jobs program has created a “virtual one-
stop career center,” where all transactions can be completed online.43
Other public programs feature computerized resource centers that offer
assistance for online job search and resume writing.44
Some proponents of government policy addressing the access
divide have pointed to the job-networking potential of the Internet
as an important reason for government intervention.45 Casual
acquaintances, or “weak ties” within personal networks, are valuable
as informal referral systems for jobs.46 Poor people living in areas of

concentrated poverty often lack networks with sufficient information
about jobs, particularly well-paying jobs. Because of residential seg¬
regation, low-income minority workers may be particularly disad¬
vantaged and may be unable to find out about available jobs elsewhere
in the metropolitan area.47 In lieu of effective personal networks,
workers are forced to rely on more formal mechanisms for job search,
such as job-training programs, schools, classified ads, and employ¬
ment services. This suggests that online job search may be particu¬
larly valuable for low-income and minority individuals.
Survey Results for Online Job Search
According to our survey, 30 percent of respondents have searched for

or applied for a job online. A clear majority, 64 percent, said that they
would feel comfortable using a computer for these purposes, so online
job seekers may become more numerous in the future. Because lim¬
ited access to computers and the Internet may have biased responses
to the questions, we asked, “Would you use a computer located in a
public place to search for or apply for a job online?” A public place was
defined as a library or a community technology center, where com¬
puter access would be provided. The same proportion—64 percent
was willing to search for a job online, with or without public access.
Support for online job search was most likely among educated,
young, male, and African American respondents, controlling for other
factors.48 Holding other demographic factors constant, the predicted
probability of support for online job search was 14 percentage points
higher for college graduates than for high school graduates. Similarly,
the probability of support for online job search was 27 percentage
points higher among younger respondents (defined as twenty-eight
years old, or one standard deviation below the mean) compared to
older respondents (defined as sixty-one years old, or one standard devi¬
ation above the mean). Racial and gender differences were smaller, but
still statistically significant. The probability of favorable attitudes
toward online job search was 77 percent for men compared to 70 per¬
cent for women and 76 percent for African Americans compared to
70 percent for whites, holding all other factors constant. Those who
express interest in online job search are the technologically skilled,
WHAT MATTERS
Who Is More Willing to Search for a Job Online?
Young (81 % for 28-year-olds vs. 54% for 61 -year-olds)—27-point

difference
Educated (76% for college graduate vs. 62% for high school diploma)—
14-point difference
Males (77% vs. 70% for females)—7-point difference
Employed (70% vs. 63% for unemployed)—7-point difference
Who Is More Willing to Use Public Computer Access for Job Search?
African Americans
Educated
Males
Young
Who Is More Likely to Have Searched for a Job Online?

African Americans
Educated
Males
Young
Employed
sample.The only statistically significant differences are the ones reported above
(see tables A4.2.A4.3, and A4.4). We have calculated the probability that respon¬
dents agree with these statements, controlling for other factors.
with the exception of African Americans. Income makes no differ¬

ence in attitudes about online job search. These patterns remained
the same when we asked respondents whether they would be willing
to search for jobs using a computer located in a public place.49 Pub¬
lic access did not make those lacking access more willing to engage in
online job search.
How do patterns regarding attitudes about Internet job search com¬
pare to actual experience? The same factors are statistically significant
predictors for use of online job search, with the exception that the
unemployed are statistically less likely to have searched for a job online
than are employed respondents. Better-educated, younger, male,
employed, and African American individuals are the most likely to have
used the Internet for a job search, after controlling for other factors.50
Our findings are consistent with other surveys that have found that
a higher percentage of African Americans have used the Internet for
job search.51 In our own survey, 37 percent of African Americans had
searched for a job online, compared to only 28 percent of whites. The
added value of our analysis is that we go beyond these simple per¬
centages and find that the difference between African Americans and
whites is statistically significant, controlling for other variables such
as income, education, and age.
What could explain somewhat higher support for Internet job
search, and markedly greater online job search activity, among
African Americans? One explanation is that African Americans value
additional network opportunities in particular, perhaps because they
lack the informal networks that often serve as referral systems for job
seekers. Interest in online job search and application is also consistent
with African American perceptions about the connection between
economic advancement and computer skills. This indicates a general
willingness to use technology in pursuit of economic opportunity.
Lifelong Learning
Turbulence in the labor market and changing skills requirements also

signal an increased need for continued training and education, or life¬
long learning. Adult learners have many alternatives: vocational pro¬
grams, degree-granting programs in community colleges and
universities; adult basic education, including courses leading to a gen¬
eral education development (GED) credential, and on-the-job train¬
ing, among others.
The trend over the past few decades has been toward a growing
cohort of “nontraditional” adult students in colleges and universities.
In fact, only “27 percent of today’s undergraduates are ‘traditional’
students who have a high-school diploma, enroll full time right after
high school, and depend on parents for financial support.
Computers assist adult learners in two ways. First, they provide a
useful tool for carrying out assignments in the traditional classroom.
Second, they have extended the walls of the classroom into the home
and the community through web-based distance learning and educa¬
tional software. In addition to the many distance-learning alternatives
available through postsecondary institutions, web courses have made
on-the-job training programs more convenient and cost effective for
employers.53
The Organization for Economic Cooperation and Development
(OECD) has urged member states to expand the use of distance learn¬
ing for adult basic education programs focused on literacy and sec¬
ondary education. Adult literacy programs enroll only about 10
percent of low-literacy Americans, and in all countries such programs
have a high rate of turnover, with individuals frequently dropping in
and out of classes.54 Distance learning, according to the OECD, may
ease the burden of managing education and adult responsibilities.
There are other technological alternatives as well. Interactive software
with self-paced drills can be used to develop basic skills such as liter¬
acy and numeracy. The Internet itself may serve as a teaching tool,
encouraging individuals with limited literacy to increase their read¬
ing by following areas of interest. Limited home access and deficient
computer skills present some hurdles for increasing distance learning
and home study for adult basic education. Schools, libraries, and
other institutions might use computers and the Internet to offer flex¬
ible options for independent study on their premises along with on¬
site technical and educational support.
Survey Results for Online Courses
In our survey, we asked respondents about their attitudes and expe¬

riences with online courses. As with online job search, more than two-
thirds of respondents answered that they would be willing to take a
class online. The percentage of respondents willing to take courses
online was nearly identical to interest in online job search, but actual
experience in distance learning was much more limited—11 percent
had taken a course online versus 30 percent who had looked at job
information on the Internet. Online courses obviously involve a
higher commitment of time and money than online job search. Web-
based courses can require information literacy and skills for use of
bulletin boards or chat rooms, e-mail, file attachments, word pro-
cessing, and streaming video. Technical glitches often require prob¬

lem-solving abilities and patience. In fact, 53 percent of our respon¬
dents replied that they would need assistance in taking a course
online. This was especially true for less-educated individuals. Of
respondents without a high school diploma, 77 percent reported that
they would need assistance to take a class online, and 57 percent of
those with a high school diploma said they would need help. A sub¬
stantial number of more-educated workers expressed a need for help,
too—47 percent of those with some college, and 39 percent of those
with a bachelor’s or graduate degree.
Interestingly enough, substantially fewer individuals said that they
would take an online course in a public place, where assistance could
be offered. Only 53 percent of respondents were willing to do so, ver¬
sus 61 percent of respondents who answered that they would take a
course online. This 8-percentage-point drop contrasted with online
job search, where public access made no difference. The more-sus¬
tained commitment of taking a course apparently made public access
unattractive for some respondents. Even so, more than half of our
respondents expressed a willingness to take a course through a pub¬
lic access site such as a library, community center, or school.
Interest in online education was statistically more likely among
the educated, the young, the affluent, and the employed, controlling
for other factors.55 The pattern for online education diverged some¬
what from attitudes about careers and job search, because interest
was more consistent with the probability of access and skill. Income
was a significant factor for interest in online courses, but it was not
related to attitudes about careers or online job search. Age mattered,
but not as much as for online job search. Interest in online educa¬
tion was only 10 percentage points higher for a hypothetical twenty-
eight-year-old than for a sixty-one-year-old, compared to a 26-point
difference for online job search.56 Older respondents were more
likely to be willing to take a course online than to search for a job
online, reflecting higher interest in education than career develop¬
ment at this stage of life.
Although older, unemployed, less-educated, and low-income indi¬
viduals are statistically less likely to express interest in web-based
courses, there is still a high level of willingness to consider such an
alternative, even among these less-interested groups. Translating this
willingness into the commitment to actually take such classes is
WHAT MATTERS
Who Is More Willing to Take Online Courses?

difference
Educated (76% for college graduate vs. 66% for high school diploma)—
10-point difference
Affluent (75% for high income vs. 65% for low income)—10-point
difference
Employed (71% vs. 64% unemployed)—7-point difference
(Asian Americans and Latinos were less likely than whites to favor
online courses.)
Who Is More Willing to Take an Online Course in a Public Place?

Young
Male
African American
Who Is More Likely to Have Taken Online Courses?

Young
Educated
Employed
African American
Note. Estimates are based on a hypothetical respondent who is female, white, and
sample. The only statistically significant differences are the ones reported above
(see tables A4.2,A4.3, and A4.4).We have calculated the probability that respon¬
dents agree with the above statements, controlling for other factors.
another matter, however, in light of the need for assistance with com¬
puters expressed by many of these same groups, particularly the least
educated.
Providing public access (and assistance) apparently did not make
many disadvantaged groups more likely to cite interest in online
classes—with one exception.57 African Americans were more inter¬
ested than whites in taking a class using a computer in a public place.
Men are more willing than women to use public access sites for
courses. The educated and the affluent drop out of the ranks of the
“more interested” when online courses are located in a public place

rather than being accessible from home.
Those who have actually taken classes online are statistically more
likely to be young, better-educated, and employed, with the excep¬
tion of African Americans, who were more likely than white respon¬
dents to have participated in online education.88 This is consistent
with the findings on online job searches.
To date, a small percentage of the population has actually partici¬
pated in online courses. The options for online study are, however,
expanding. The interest expressed by our respondents indicates
potential for future growth as well.
Conclusion
Surveying the landscape of the new economy demonstrates clearly

that information technology plays a role in economic opportunity.
The word opportunity is key. Jobs will certainly be available, in the
near and perhaps distant future, for individuals devoid of computer
skills, but economic opportunity suggests the ability to subsist above
the poverty level, to enjoy some choice and mobility in the labor mar¬
ket, and to realize higher returns for additional experience, skills, and
training. An increasing number of jobs, including those for less-edu¬
cated workers, require some type of computer use. Recent evidence,
including our survey, suggests that even jobs requiring only a high
school education sometimes involve rather sophisticated computer
use beyond routine data entry, including navigation of the Internet.
This may be particularly true for better-paying jobs and, as census
data show, for mobility into supervisory and skilled jobs. Moreover,
the boundaries of the landscape have rapidly changed over the past
decade, and technology use in the workforce promises to continue to
gain momentum. Computer use, however, cannot be considered in
isolation from other dimensions of skill. The topography of the new
economy indicates rising skills requirements overall, with increasing
returns to education, and falling wages for the less-educated over
much of the past few decades.
Our survey results demonstrate that more than two-thirds of
Americans are convinced that a connection exists between computer
skills and various types of economic opportunity. This is particularly
true among disadvantaged groups such as African Americans, the

unemployed, and women, as well as the young. African Americans
and Latinos are even more likely than other racial and ethnic groups
to believe that using the Internet is necessary to keep up with the
times. This suggests that the policy problem is the inequitable distri¬
bution of opportunities for access and skills development rather than
awareness of technology and its potential benefits,
Instability in the changing landscape of the new economy means
that workers may need to change jobs more often, to respond to the
threat of involuntary layoffs, or to pursue new prospects in the
absence of well-defined career ladders. For many individuals, eco¬
nomic opportunity will entail job search and continued training and
education. Technology offers new tools for obtaining information
about jobs, for connecting with employers, and for enhancing skills
and educational credentials. More than two-thirds of our respondents
expressed a willingness to search for a job or take a course online,
although only 30 percent have used the Internet to look for a job, and
11 percent have used it to take a class. Among those most willing to
use the Internet for these purposes are those who are the most digi¬
tally savvy—respondents who are better-educated and young. Age
differences are less stark for online education, although still signifi¬
cant. Income matters for online education, with the poor being less
willing to take an online course, but it does not matter for attitudes
about online job search or careers. Poor individuals apparently have
many of the same attitudes and aspirations as other Americans regard¬
ing economic opportunity. Limited home access for the poor may
present higher hurdles for online study than for online job search. A
number of individuals also cited the need for assistance in taking an
online course, especially the least educated.
Overall, income was not an important factor for the opportunity
divide, at least not for attitudes about careers and online job search.
The belief in technology and economic opportunity is widely shared,
and groups more likely to suffer from technology skills deficits are at
least as likely to believe that computer skills are necessary for eco¬
nomic opportunity.
What emerges from the survey is a depiction of African Americans
as particularly attuned to the use of technology for economic oppor¬
tunity. This is contrary to what would be expected, if a “digital divide”
along the lines of race were a matter of attitudes rather than access and
skills. This likely reflects a more general priority among African Amer¬
icans regarding the need for economic opportunity, in view of the con¬
tinued stark disparities in income. Despite economic progress over the
past few decades, African Americans are still disproportionately repre¬
sented among the poor. The median African American household
income is only two-thirds that of white households.59 Closing the
income gap is not merely a matter of technological skills. Differences
in educational attainment matter, too. So do other factors, such as seg¬
regation, the quality of education in central city neighborhoods, a
“spatial mismatch” between the knowledge-intensive jobs available in
central cities and the skills of inner-city residents, and discrimination
by employers.60 Still, the relatively widespread interest African Amer¬
icans express in technology and economic advancement offers possi¬
bilities for closing the access and skills divides.
This favorable assessment of technological opportunity on the part
of African Americans parallels other findings on racial attitudes about
economic opportunity. A review of several decades of survey research
concluded that, although African Americans perceived discrimina¬
tion to be a serious problem, they were even more optimistic than
white respondents about the future and their own chances for success
in achieving the “American dream.”61 Poor African Americans were
more likely than their middle-class counterparts to express their
hopes for the future in terms of economic gains.62
Our findings build on these observations. In our heavily low-
income sample, we discover a deep current of optimism about tech¬
nology and economic opportunity among all respondents, but
particularly among African Americans. Controlling for other factors,
African Americans are more likely to believe that computer skills
facilitate economic advancement, are more willing to use computers
for job search (and actually use them more frequently than whites),
and are also more willing to take courses online, when public access
is provided (and to have taken online courses). These findings dove¬
tail with some of the results in the prior chapter on the skills divide.
African Americans were also more likely to agree to most methods of
instruction or assistance, demonstrating a receptiveness beyond the
generally favorable attitudes expressed by the majority of the sample.
Latinos also expressed more positive attitudes than whites on some
measures in this chapter and the previous one. For Latinos, however,
the results were less consistent and less pronounced.
The focus in this chapter has been on individual aspirations Tor

economic opportunity. Individual opportunities matter for public
policy when systematic patterns of inequality rob some individuals of
a fair chance for bettering their lot. Low levels of skill among the poor
also have implications for welfare-to-work programs, if the goal of
such initiatives is sustained employment sufficient to provide a stan¬
dard of living above the poverty level; however, individual skills mat¬
ter in a collective sense as well.
In the aggregate, skills and knowledge represent “human capital”
for the development of the economy and society. Lester Thurow has
characterized knowledge as “the new basis for wealth” in the infor¬
mation age.63 The debate over skills in the early 1990s framed the
problem as the need to maintain a nationally competitive economy
amid the pressures of globalization and technological change. In
recent years, state governments and business organizations have
defined educational issues as economic development concerns.6^
Impoverished rural communities and cities suffering from deindus¬
trialization have likewise turned toward human capital strategies,
hoping to attract new industries, including high-tech firms, on the
basis of a skilled, educated workforce.65 By addressing the divide in
technological skills together with the need for other training and edu¬
cation, public policy can take steps to ensure opportunities for indi¬
viduals and for their communities.
Notes
L See Castells 2000, 77; Niskanen 2000, 93; U.S. Department of Com¬
merce 2000a, 1; McGuckin and Van Ark 2001, 10.
2. U.S. Department of Commerce 2002; Pew Internet and American Life
Project 2000.
3. Mishel, Bernstein, and Schmitt 2001, 19-20; Barrington 2000.
4. Organization for Economic Cooperation and Development 2000, 3.
5. Mishel, Bernstein, and Schmitt 2001, 154.
6; Atkinson and Court 1998, 25; Bernhardt et al. 2001, 48; Ellwood
2000.
7. Mishel, Bernstein, and Schmitt 2001, 1; U.S. Department of Labor
1999, 5; Ellwood 2000, 31.
8. Barrington 2000, 8.
9. Mishel, Bernstein, and Schmitt 2001, 1.

10. Census bureau figures for 2001, reported in Pear 2002.
11. Carnevale and Rose 2001, 52.
12. Kruse and Blasi 2000.
13. Carnevale and Rose 2001, 64; Strawn and Martinson 2001, 111.
14. Bernhardt et al. 2001, 133.
15. Herzenberg, Alic, and Wial 1998, 63.
16. McGuckin and Van Ark 2001; U.S. Department of Labor 1999,
chap. 2.
17. U.S. Department of Commerce 2002, 57.
18. See, for example, Moss and Tilly 2001, and Holzer 1996.
19. According to the 1997 National Employer Survey, nonsupervisory
workers were engaged in computer use at 52 percent of firms—up from 42
percent in 1994 (Kruse and Blasi 2000, 72). A telephone survey of employ¬
ers in several major cities revealed that computers were used in nearly 59 per¬
cent of jobs not requiring a bachelor’s degree. They were reportedly used on
a daily basis in 51 percent of these non-college jobs (Holzer 1996, 49).
20. Holzer 1996, 49; Kruse and Blasi 2000, 72; Moss and Tilly 2001,83.
21. The rest of the ten occupations with the largest expected growth
include computer-related professions, such as computer support and soft¬
ware engineers, and jobs likely to involve some computer use, such as office
clerks, customer service representatives, and registered nurses.
22. U.S. Department of Commerce 2002, 58, 60. For the category of
operators, fabricators, and laborers, only 20.7 percent reported using com¬
puters at work, and, of them, 9.2 percent used the Internet or e-mail at work.
This was the lowest category for computer use, with service workers report¬
ing only a slighter higher percentage of computer use, at 25.4 percent. For
more-skilled workers with specialized training, the rates of usage are higher.
For the category of precision production, craft, and repair, computer use rose
to 31.7 percent.
23. Technology can result in either deskilling or upskilling of jobs. The
introduction of optical scanners, for example, has lowered the skills require¬
ments for cashier jobs, because it reduced the need for mathematical skills.
Most analysts argue that skills requirements have risen in the new economy
(Holzer 1996, 2-3; Teixeira and McGranahan 1998, 120-21; Kruse and
Blasi 2000, 55, 128; Moss and Tilly 2001, 49, 56; Osterman 2001, 75) on
the basis of evidence from the National Employer Survey and other research.
The 1994 National Employer Survey revealed that 56 percent of employers
experienced an increase in the required skills for production or support jobs
over the prior three years (Kruse and Blasi 2000, 55, 128). Studies of bpth
urban and rural areas have found that between 30 and 40 percent of jobs in
these economically distressed areas have experienced some upskilling in
noncollege occupations (Holzer 1996, 49; Teixeira and McGranahan 1998,
121; Moss and Tilly 2001, 56, 83).
24. Moss and Tilly 2001, 65.
25. Holzer 1996, 49; Moss and Tilly 2001, 56, 83; Teixeira and
McGranahan 1998, 121-23.
26. Osterman 1999, 108; U.S. Department of Labor 1999.
27. Herzenberg, Alic, and Wial 1998, 63; Appelbaum and Albin 1998,
155-56.
28. Secretary’s Commission on Achieving Necessary Skills 1992, 10—11.
29. Osterman 2001, 74.
30. Teixeira and McGranahan 1998, 123.
31. Because the dependent variable is binary for the model “it is neces¬
sary for people to use the Internet to keep up with the times,” logistic regres¬
sion coefficients are reported in table A4.1.
32. Since the dependent variable for this economic opportunity index is
measured on an ordinal scale, coefficients are based on an ordered logistic
regression model. See table A4.1.
33. Hall and Mervis 1995, 323; Herzenberg, Alic, and Wial 1998, 5.
34. Osterman 1999, 53.
35. Osterman 1999, 48; Kruse and Blasi 2000, 44.
36. Atkinson and Court 1998, 5.
37. Bernhardt et al. 2001, 192; Herzenberg, Alic, and Wial 1998, 13.
38. Herzenberg, Alic, and Wial 1998.
39. Bernhardt et al. 2001, 167-68.
40. Strawn and Martinson (2001) cite Cancian and Meyer (2000), Glad¬
den and Taber (2000), and Rangarajan, Schochet, and Chu (1998). The lat¬
ter is available online at www.mathinc.com.
41. See www.ajb.org.
42. Stiglitz, Orszag, and Orszag 2000, 85.
43. See Heldrich Center 2002.
44. See Heldrich Center 2002; Saulny 2001.
45. Civille 1995, 198-99.
46. Civille 1995, 198-99; Osterman 2001, 78.
47. Civille 1995, 198-99; Moss and Tilly 2001, 254; Holzer 1996,
127-28; Kasarda 1990.
48. Because the dependent variable is binary, logistic regression coeffi-
cients are reported in table A4.2. We calculate predicted probabilities for all
variables holding other factors constant (King, Tomz, and Wittenberg
2000). Our hypothetical individual has average (mean) education, age, and
income and is an employed white female.
49. For results, see table A4.3.
50. Because the dependent variable is binary, logistic regression coeffi¬
cients are reported in table A4.4.
51. The Pew Internet and American Life Project study (2000) showed
that 51 percent of African Americans who have ever used the Internet have
used it to get information about jobs, compared to 37 percent of whites who
had ever used the Internet.
52. Evelyn 2002.
53. Greengard 1998.
54. Ginsburg, Sabatini, and Wagner 2000, 78-79.
56. All predicted probabilities are calculated holding other factors con¬
stant (King, Tomz, and Wittenberg 2000). Again, our hypothetical individ¬
ual has average (mean) education, age, and income and is an employed white
female.
59. The 2001 median household income was $29,470 for African Amer¬
icans and $46,305 for whites. See U.S. Census 2002b.
60. Holzer 1996, 127-28; Moss and Tilly 2001, 254; Kasarda 1990.
61. Hochschild 1995, 69.
62. Ibid., 73.
63. Thurow 1999, xv.
64. Mickelson 1996, 246; Sipple et al. 1997.
65. Swaim, Gibbs, and Teixeira 1998, xi; Clarke and Gaile 1998, chap.
7; Johnson 2002.
Chapter 5
The Democratic Divide
with Ramona McNeal
I n an era when the outcome of a presidential election is decided by

a few hundred votes, and the balance of power in the U.S. Senate
can be determined by a similarly close margin, the question of
what effect the Internet as a medium for political information and
involvement has on the voting public is a pertinent one. The Inter¬
net may enhance citizen information about elections and in turn
stimulate increased participation. Yet, because of unequal access to
technology, the Internet may expand turnout rates only among those
who are already predisposed to vote, broadening the gulf between
those who do and those who do not participate. In light of declining
civic engagement and participation in American politics in the last
three decades, the question of what effects new information technol¬
ogy may have on democracy is ever more important.1
Leading theories of political participation have shown that socioe¬
conomic characteristics of voters—education and income—are the
most important factors in explaining whether one votes in the United
States. Voter turnout is also affected by race, age, gender, and attitu-
dinal factors such as strength of partisanship, political efficacy, and
political interest.2 Although a long tradition of research documents
the demographic and psychological determinants of political partic¬
ipation, there is also evidence to suggest that changes in communi¬
cation technology may play an important role in influencing electoral
behavior. Research has found that those who read about politics in
newspapers learn more than those who watch television.3 In the past
decade, technology has changed the way many people gather news
and participate in politics. The most important of these new tech¬

nologies is the Internet, which has become the mass medium for the
twenty-first century. The Internet combines the audiovisual compo¬
nents of traditional forms of media such as newspaper and television
with the interactivity and speed of telephone and mail. It facilitates
flexibility, allowing individuals to choose what information to access
and when to access it. Technology also permits users to exchange large
amounts of information quickly, regardless of distance.
A limited but developing body of research has explored the rela¬
tionship between Internet use and varying forms of civic participa¬
tion, including voting.^ Few researchers, however, have explored
citizen attitudes toward the use of information technology for polit¬
ical participation and communication with government. We are most
interested in attitudes because of their potential to affect participa¬
tion in the future, as nascent trends toward online government and
politics build momentum. Is there a “democratic divide” that emerges
from the access and skill divides? Are more-educated, affluent citizens
more supportive of using the Internet for political participation than
those with lower socioeconomic status? A lack of access and computer
skills may negatively influence attitudes about using computers and
the Internet to communicate with government and participate in pol¬
itics. In an era when “e-government” is rapidly spreading and the use
of Internet voting may well be on the horizon, the answers to these
questions have clear implications for public policy as well as future
political participation.
Some scholars suggest the Internet may function as a new deliber¬
ative public forum, drawing the less engaged into civic life, strength¬
ening democracy, increasing political participation, and leveling the
playing field. Others claim that the “digital divide” and the growing
corporate ownership of the Internet will merely replicate the patterns
of inequality experienced today. Davis contends powerful groups will
continue to dominate the production of political news and informa¬
tion, the expression of opinion, and the mobilization of political par¬
ticipation, online as well as off.5 The literature on traditional political
participation already shows a substantial gap based on income and
education.6 This research addresses whether practices such as online
voting and e-government will exacerbate or ameliorate existing dis¬
parities in political participation based on demographic factors.
Alongside some hopeful signs, our data do reveal a democratic divide,
as individuals with higher education and income are more suppprt-

ive of digital democracy.
Theories of Digital Democracy
Many scholars and political pundits argue that Americans are becom¬
ing more and more disenchanted with traditional institutions of rep¬
resentative government and disillusioned with older forms of civic
engagement and participation. Although a “crisis of democracy” may
be overstated, indicators suggest an increasing number of “critical
citizens” are characterized by high expectations of democracy as an
ideal yet low evaluations of the actual performance of representative
institutions.7
Participation has become one of the dominant themes of modern
governance.8 Normative theorists in particular have long argued that
direct forms of democracy can motivate participation by energizing
citizens with a sense of civic duty and political efficacy.9 For advocates
of direct democracy, such as Benjamin Barber, opportunities and
mechanisms are needed to increase citizen deliberation and direct
involvement in decision making, for example, through initiatives and
referendums. Calling for more “discursive democracy,” “strong
democracy,” “teledemocracy,” and “deliberation,” scholars have
offered a variety of participatory models of decision making.10 From
radical models of a pure direct democracy to more transparent repre¬
sentative systems, citizen participation is deemed as critical in gov¬
erning accountability and public dialogue.11
These participatory models imply that the system of representative
democracy is far from perfect in transmitting the wishes of the pub¬
lic into policy and that citizen participation can improve politics and
policy, even in a complex modern society.12 The general prescription
for making government function better is to foster greater individual
and collective participation and structure institutions to include mass
citizen participation. In its simplest form, participatory government
is plebescitarian, with the public being asked to decide public issues
by a direct vote.13
Some scholars see information technology as the most important
ingredient for fueling a participatory revolution.1^ Proponents argue
that the democratizing effect of the Internet will encourage citizen
participation at all levels of government. The interactivity, low-cost,

flexibility, and information capacity available on the Internet have the
potential to allow the public to become more knowledgeable about
politics as a first step toward greater participation. New information
technologies generate multiple opportunities for sharing political
information and communication. Chat rooms, listservs, e-mail, and
bulletin board systems represent new modes of information exchange
and opinion mobilization. By allowing individuals to be both
receivers and active providers of information, the Internet may foster
increased political communication. As a new channel of two-way
communication, the Internet may strengthen and enrich connections
between citizens, on the one hand, and political parties, interest
groups, and elected officials, on the other.1 6 Proponents of e-demo-
cracy argue that the Internet offers hope to reconnect citizens to the
political process and revive civic engagement in American politics.16
To date, the Internet has mostly provided a conduit for informa¬
tion and communication. In the future, the Internet may offer new
opportunities for participation through online registration and
voting, virtual town meetings, and petition drives that utilize elec¬
tronic signatures. Some suggest that those states that have been lead¬
ers in using direct democracy will be the first to allow Internet voting
and voter registration. Two unsuccessful citizen initiatives circulated
for the 2000 California ballot, for example, would have required the
secretary of state to implement Internet voting and voter registra¬
tion.17 Arizona, a state with frequent use of ballot initiatives and ref¬
erenda, was the first to hold a binding election using Internet voting
in 2000.18
Others, however, contend that information technology will pro¬
mote further inequality in democratic participation, widening the
gap between those who participate and those who do not.19 Individ¬
uals with higher income and education are already statistically more
likely to vote in the United States.20 Disparities in access to the Inter¬
net based on income, education, race, and ethnicity mean that tech¬
nology resources are far from equally distributed and that online
politics may therefore amplify the voice of the affluent and well edu¬
cated, further marginalizing the underprivileged. In this scenario,
opportunities for online political participation will primarily benefit
those elites with the resources and motivation to take advantage of
them, leaving the poor and uneducated farther behind.21
Research on the Internet *
and Political Participation
How does empirical research inform this largely normative debate?

Early studies on the effects of the Internet on civic engagement have
been mixed. Using a national representative sample from the 1998
American National Election Surveys, Bimber found that access to the
Internet had no impact on voter participation.22 With the exception
of giving campaign donations, the political behavior of those with
access to the Internet and online political information did not differ
from those who did not use the Internet to seek political information.
Access to the Internet and online political information did statisti¬
cally increase the probability that a respondent would contribute
money to political campaigns, suggesting a mobilizing potential.
Bimber’s research, however, is limited to one midterm election.
Recent research, using more sophisticated statistical methods
and longitudinal datasets, concluded that the use of the Internet for
political information had a positive effect on participation during
recent presidential elections. Tolbert and McNeal found the Internet
may enhance information about candidates and elections, and in
turn stimulate increased participation.23 Using National Election
Study (NES) data from 1986-2000, they observed that respondents
with access to the Internet and online political news were significantly
more likely to have voted in the 1996 and 2000 presidential elections.
This raises the question, however, of whether the Internet influenced
political participation or whether political activists happened to be
more likely to be online. The authors used a two-stage model to iso¬
late cause and effect (to control for simultaneity problems) and used
multivariate regression to hold factors other than Internet use con¬
stant. Participation increased even after controlling for education,
income, race/ethnicity, gender, age, partisanship, attitudes, traditional
media use, and state environmental factors. The exception to this pat¬
tern was the 1998 midterm election, the year studied by Bimber. Sim¬
ulations suggest Internet access increased the probability of voting by
an average of 12 percent, and use of online election information
increased the probability of voting by 7.5 percent in the 2000 election,
all else equal. The mobilizing potential of the Internet in 2000 was also
associated with other forms of involvement in election campaigns.
Individuals viewing online political information were significantly

more likely to talk to others about candidates or parties, display but¬
tons or signs, work for a party or candidate, attend rallies, and give
money to candidates, parties, and interest groups.
Other studies have addressed the information and communication
potential of the Internet for influencing political participation,
including activities such as contacting political officials, attending ral¬
lies, or signing petitions. Weber and Bergman found that those in¬
dividuals who engaged in Internet activities such as using e-mail
and chat-rooms were more likely to be engaged in a variety of polit¬
ical activities.24 Weber and Bergman, however, used Survey 2000,
an online survey jointly conducted by academic researchers and
National Geographic Interactive. The survey was self-selected and
nonrandom and therefore subject to selection bias, unlike the stud¬
ies reported above. One nationally representative survey (1999 DDB
Life Style Study) contrasted Internet use for information exchange to
use for social recreation, product consumption, or financial manage¬
ment. Across age cohorts (generation X and baby boomers) individ¬
uals who used the Internet for information exchange reported higher
levels of interpersonal trust and civic engagement, after controlling
for demographic, contextual, and traditional media use variables.25
Another area of participation that has been singled out by
researchers for study is citizen-initiated contact of public officials.
Earlier research found that age, gender, education, political connect¬
edness, and proximity to government institutions are important fac¬
tors in determining whether a citizen will initiate communication.
Older, educated, white citizens have been found to be more likely to
contact government officials, whereas women were less likely to insti¬
gate contact.26 Utilizing a self-selected, nonrandom online survey
conducted in 1996 and 1997 and two phone surveys, Bimber exam¬
ined whether the Internet altered the pattern of citizen communica¬
tion.27 He found that, when comparing traditional means of
communication to the Internet, many of the same relations still
existed. The Internet, however, magnified the gender gap in commu¬
nication but narrowed the difference based on political connected¬
ness. Despite the limitations of Bimber’s nonrandom sample, his
study was one of the few to explore the dernographic impact of new
modes of communication.
Research on the Emerging Issue

of Internet Voting
One of the controversies over the possible introduction of Internet

voting is its differential impact, in light of disparities in access. An
analysis of the 2000 online Arizona Democratic primary offers a win¬
dow into how changing election procedures to accommodate digital
technology may change election outcomes.28 The 2000 primary
allowed registered Democratic voters to cast ballots in four ways:
41.16 percent used the Internet; 37.68 percent used traditional
absentee mail ballots; 4.8 percent used electronic voting machines at
polling booths; and 16.36 percent used paper ballots. Based on apre-
and postelection survey funded by the National Science Foundation,
Solop found that better-educated and younger voters took advantage
of the Internet voting option in the Arizona primary.29 Education has
already been found to be an important factor in determining whether
individuals choose to participate in politics. Inclusion of an Internet
alternative may further bias voting patterns toward the higher edu¬
cated. Younger voters, however, have historically been among those
groups least likely to participate. Online voting may be instrumental
in increasing turnout among the young.
Alvarez and Nagler argue that one way to assess Internet voting is
to compare the group of citizens currently voting to those that would
vote if online balloting were implemented.30 Sharp differences in the
demographics of these two groups would be evidence of a change in
political representation caused by Internet voting. Alvarez and Nagler
use aggregate census and election return data from Arizona’s fifteen
counties and ecological-inference methods to estimate white and
nonwhite Democratic turnout rates.31 They compare turnout in the
1998 statewide Democratic primary with the 2000 Democratic pres¬
idential primary, where Internet voting was introduced. Although
overall statewide turnout was significantly lower in the 2000 primary
(10.59 percent) compared to average primary turnout of 23.94 per¬
cent in the past three elections, the authors find that the average rate
of decrease for nonwhite voters was six times greater than the average
rate of decrease for white voters. White turnout actually increased
from 1998 levels in two counties, but nonwhite turnout declined
from 1998 to 2000 in every Arizona county.32
The research on Internet voting suggests its potential to mobilize

new sectors of the population, particularly the young, but also to
expand existing disparities in participation rates based on race and
ethnicity. A number of factors, however, make it difficult to general¬
ize from the Arizona case study to other state and national elections.
There is an inherent difficulty in comparing turnout in an off-year
(1998) and a presidential (2000) election. There are some problems
with using statistical methods in this particular study, where the
number of “things” being studied—counties—is not very large.33
The unique circumstances of the Arizona election also cast some
doubt on its broader implications. The national Democratic contest
had already been decided by the time of the Arizona primary, result¬
ing in extremely depressed turnout. It may have been this, rather than
Internet voting, that caused turnout rates to plummet among some
groups more than others.
Initial findings about Internet voting are suggestive, but Internet
voting may not be a widespread reality in the near future. Contro¬
versies over e-voting include concerns about election fraud and online
privacy, and the construction of secure voting systems would entail
considerable expense.3^ Some states already allow online voter regis¬
tration,35 and Georgia allows absentee ballots to be cast online. Elec¬
tronic government, however, represents a technology application that
is currently burgeoning and that, according to its advocates, has the
potential to transform the relationship between citizens and govern¬
ment at all levels—local, state, and federal.
Theories of E-Government
E-government “refers to the delivery of information and services

online via the Internet or other digital means,” and may also include
opportunities for online political participation.36 The diffusion of e-
government has been rapid and widespread. The federal government
has emphasized the creation of federal websites as part of its effort to
“reinvent” government, and there is now a central portal for all fed¬
eral services.37 All fifty states have adopted some form of e-govern¬
ment; a recent survey indicates that 80 percent of local governments
maintain websites.38
E-government is characterized by multiple constituencies .and

multiple goals. Streamlining government-to-business transactions,
such as procurement and permits, is one aim of e-government, and the
traditional orientation of state and municipal websites has been topro-
mote business and economic development.39 Our primary concern in
discussing the relationship between e-government and a democratic
divide is “government-to-citizen” transactions.40 E-government is
most relevant to ordinary citizens for its potential to improve service
delivery and to enhance transparency and responsiveness of govern¬
ment agencies.41 Proponents argue that e-government could enable
citizens to interact with and receive services from government twenty-
four hours a day, seven days a week. They describe e-government as
“the continuous optimization of service delivery, constituency partic¬
ipation, and governance by transforming internal and external rela¬
tionships through technology, the Internet and new media.”42
E-government initiatives are an outgrowth of the reinventing govern¬
ment paradigm, particularly at the federal level, that were promoted
by the Clinton-Gore administration.43 Goals touted by the adminis¬
tration included increased efficiency, responsiveness to “customers,”
and citizen empowerment.44 Others suggest e-government has grown
larger than government reform and carries with it expectations and
possibilities of transforming, not just reforming, government, consis¬
tent with the literature on digital democracy.45
What, in practice, does e-government look like?46 At its most basic
level, e-government consists of the posting of information about ser¬
vices, contact persons, and a variety of government documents,
including forms, policies, and legislation. At a more sophisticated
level, it allows completion of government transactions online. The
federal government now permits electronic filing of income tax
forms, and about 25 percent of federal and state websites offer online
transactions for services such as vehicle registration, driver’s licenses,
hunting and fishing licenses, tax filing, and more.47 Service delivery
improvements may include portals or “one-stop” websites that cen¬
tralize services and information through links and perhaps searchable
databases. The federal job bank described in the previous chapter is
an example of e-government services delivered entirely online. Such
transactions are less frequently available on local government web¬
sites.4® To realize their potential for making government more acces¬
sible, websites may need to do more than centralize information
online. Some research points to a need to clarify legal and technical

terminology on government websites and to organize and present
information in ways that are user friendly.49 This may be particularly
important for individuals with limited education or experience with
contacting government agencies or following legislation.
Many observers view e-government as a means for enhancing
democratic participation, as government information online pro¬
motes transparency.90 The Internet also facilitates communication
with agencies and elected officials, especially through e-mail. Eighty-
four percent of federal and state websites include e-mail addresses,
and government receipt of e-mail from constituents is increasing.51
Online public hearings and forums have the potential to allow inter¬
action on a broader scale and to encourage deliberative participation.
These broader forms of technology-enabled participation are fairly
rare. Only 15 percent of federal and state websites provide message
boards for public comment, and less than 1 percent offer real-time
chat rooms.52 The end of this chapter discusses Berkeley, California’s,
experiment with an online forum during the revisions of the city’s
master plan. Other forays into technologically enabled participation
have been tried at the community level in the past. Santa Monica,
California, used its Public Electronic Network and public access sites
for both community information and citizen participation prior to
the advent of the Internet. In general, however, government-spon¬
sored or -sanctioned initiatives for electronic participation have been
rare. As the Berkeley case study indicates, such use confronts a num¬
ber of legal and logistical problems as well as disparities in citizen
access to the Internet.
The ambiguous nature of e-government, like digital democracy,
has resulted in hype and confusion with little systematic considera¬
tion of the expectations and limitations of taking government
online.53 Discussions of e-government are wrapped in the language
of increasing citizen participation, but the reality is that the posting
of information and service delivery are more prevalent than efforts to
promote participation. Surveys of state and local officials show that
most of them view e-government in terms of its potential to increase
efficiency and cut costs.54 Other studies explore the causes promot¬
ing the spread of e-government and find that citizen demand, mea¬
sured by Internet access in the state, is not a significant explanation
for innovation in e-government. Controlling for other factors, states
with Republican-controlled legislatures, more professional networks,

and higher levels of legislative professionalization are likely to engage
in more extensive use of e-government. This research also implies effi¬
ciency needs drive reliance on e-government rather than concerns
about expanding political participation.55 This is consistent with
other recent government reforms, such as “reinventing” government.
At the federal level, the reinvention effort emphasized cost reduction
and efficiency over other stated goals such as citizen empowerment
and responsiveness.56
Service delivery and efficiency concerns are likely to dominate fur¬
ther development of e-government.57 Yet, even if all the starry-eyed
predictions for its potential do not come to pass, e-government has
so far demonstrated important benefits for citizens. First, it can pro¬
vide valuable access to information about government services. Web¬
sites can eliminate the need to travel to government offices to obtain
government forms or other documents. Searchable databases, lists of
frequently asked questions, and links to related sites can make infor¬
mation easily accessible and convenient when it is provided and pre¬
sented in an understandable fashion. Low-income citizens often
depend heavily upon various government services and could benefit
from better access to information. Second, e-government has con¬
tributed to communication and accountability, although it fails short
of the prescriptions of e-democracy proponents. Wider availability of
information and the increased use of e-mail can facilitate civic aware¬
ness and interaction with officials. There are other means, of course,
to obtain information and to contact officials, but they can involve
frustrating trips from office to office and long delays of telephone tag.
Access to e-government for all is a desirable public objective.
To date, there are a few initial studies on the public’s use of e-gov-
ernment sites. The Pew Internet and American Life project surveys
describe use of government information as one of the fastest-growing
online activities in recent years. Of those who use the Internet, 58
percent have visited at least one government website, making this one
of the most popular Internet uses.58 Two surveys show that e-gov-
ernment users are likely to be younger, better-educated, and more
affluent, although neither of these studies controls for other factors
using multivariate regression.59 Both surveys also indicate that
African Americans are somewhat more likely than whites to visit local
government websites, which are otherwise the least used.60
Of those accessing online government information, 77 percent

seek information on tourism or recreation, 70 percent conduct
research for work or school, 63 percent download government forms,
63 percent look for information about services an agency provides, and
62 percent gather information on policies or issues. In terms of trans¬
actions, 16 percent of those who seek online government information
are filing taxes, 12 percent are renewing a driver’s license or car regis¬
tration, and 7 percent are renewing a professional license. A smaller
proportion (4 percent) is seeking a fishing, hunting, or recreation
license, and 2 percent are paying a fine. The Pew survey data reveal that
e-government is well underway, in contrast to online voting.
Some research is beginning to accumulate in the area of e-govern¬
ment more generally, but so far only a few studies address citizen use
and attitudes. Like much of the research on the access divide, these
studies often lack statistical controls, which would impart more con¬
fidence in their findings, and representative samples of the poor and
minorities. The existing research on the Internet and political partic¬
ipation is more developed, but still limited. Studies are often based
on single elections or nonrandom samples, limiting our ability to
draw broad conclusions. Our research on attitudes may provide a bet¬
ter predictor of the way in which the Internet will affect political
awareness and engagement in the near future. Are citizens supportive
of e-government and possible future reforms to implement Internet
voting and online voter registration? How do citizens feel about using
the Internet for town meetings? By examining attitudes, we can find
out whether disadvantaged groups see access to e-government as an
important need and whether there is popular support for participa¬
tory reforms such as Internet voting, online voter registration, and
town meetings in the future.
Findings from the Survey
Will use of the Internet for political participation expand or amelio¬

rate the existing disparities in traditional participation in American
politics? Analysis of our survey data suggest both the potential of
online politics to expand civic participation among some who are cur¬
rently disengaged from politics as well as the potential to widen exist¬
ing disparities in participation based on income and education.
Simple percentages from our survey demonstrate that many who

have Internet access do not use it for political purposes and that some
innovations, such as online voting, are controversial. In comparison
to the 54 percent who had home Internet access and 58 percent-who
had e-mail addresses, 31 percent of respondents had searched for
political information online, but only 17 percent had seen an online
political ad. There was somewhat more interest in e-government than
in obtaining political information—40 percent of all respondents had
looked up information on government services online.
Respondents expressed resounding support for putting govern¬
ment information online but were more reticent about using the
Internet for voting and online town meetings. More than three-quar¬
ters of respondents (78 percent) answered positively to the question,
“How do you feel about looking up government information online?”
This exceeded the two-thirds majorities who said they were willing to
search for a job or take a class online. These attitudes confirm the pop¬
ularity of e-government suggested by the Pew study of current use.
Support for Internet voting was almost evenly split. When asked,
How do you feel about voting in a government election online?” 48
percent agreed and 52 percent were opposed. This could indicate
public qualms about this particular reform (security or privacy), or
more general disinterest in voting. The survey revealed more support
for online voter registration. When asked, “How do you feel about
registering to vote online?” support rose to 58 percent. Support for
participating in an online political forum was modest as well: only 47
percent of individuals responded positively to the question, “How do
you feel about participating in an online town meeting?” It is possi¬
ble that our findings were skewed by the abundant presence of either
voters or nonvoters, but, when we controlled for reported voting in
traditional elections, support for the varying forms of digital govern¬
ment remained virtually the same.
Because limited access to computers and the Internet may have
biased responses to the online participation question, we repeated the
questions asking whether the respondent supported use of information
technology for voting, registration, and looking up government infor¬
mation using a computer in a public place. In this case, access would
be provided and election fraud could be more easily controlled. When
asked, “Would you use a computer located in a public place to vote in
an election? support rose by more than 10 percentage points, with 59
percent agreeing. Sixty-seven percent of respondents supported using a

computer in a public place to register to vote, and 74 percent supported
using a computer to search for information on government services. As
with job search and taking a course online, respondents were slightly
less willing to seek government information using public access. They
were, however, considerably more willing to use new technology for
voting and registration at a public place rather than at home. Even with
a representative sample of low-income individuals, the majority of
respondents were supportive of digital democracy and e-government,
at least when public access (and security) is provided.
Support for Digital Democracy

and E-Government
Using multivariate regression, we compared the results of four mod¬

els. In light of the differing levels of support for e-government versus
voting online, we developed a separate model to explain support for
each of the following: (1) voting in a government election online; (2)
registering to vote online; (3) looking up government information
online; and (4) participating in an online town meeting. The
responses for each question were coded 1 for agree and 0 for dis¬
agree.61 We also created an index of support for online politics and
government overall that combined items 1 through 4.
The same explanatory factors used in the access divide analysis are
included in the appendix tables for this chapter, with one exception.
We added a measure of traditional political participation, where 1
indicates that the individual was both registered for and voted in the
2000 presidential election and 0 otherwise. This measure was created
by combining two survey questions and was used instead of voting to
help control for the problem of overreporting in survey data. The
problem with using self-reported voting alone is that the percentage
of people who ostensibly vote usually far outstrips actual turnout. The
results are explained in the What Matters box that follows, and the
regression tables are provided in appendix 1.
Although overall support for the differing forms of political par¬
ticipation varied significantly—from a low of 48 percent for online
voting to a high of 78 percent for searching for government infor¬
mation online—factors associated with support for digital democracy
WHAT MATTERS
1. Who Is More Likely to Support Online Voting?
Educated, young, Democrats, voted in 2000 elections
2. Who Is More Likely to Support Online Voter Registration?

Educated, young, Democrats, males, voted in 2000 elections
3. Who Is More Likely to Support E-Government (Looking Up

Government Information Online)?
Educated, affluent, young, Democrats, non-Latino, voted in 2000
elections
4. Who Is More Likely to Support Participating in an Online Town

Meeting?
Educated, affluent, young, males, voted in 2000 elections
5. Who Is More Likely to Support Digital Democracy and

E-Government Overall (questions 1-4 combined)?
Educated, affluent, young, Democrats, males, voted in 2000 elections
Note:The only statistically significant differences are reported above (see tables
A5.1 and A5.2).When multivariate regression is used, these are the variables that
and e-government are surprisingly similar (see table A5.1). What

emerges from the data is clear evidence telling a single story—a demo¬
cratic divide exists in support for online politics and government.
After controlling for other factors, the respondents most likely to
favor online politics and government are younger, better educated
and more affluent, and are more likely to take part in traditional
forms of political participation (voted in the 2000 election).
Although there were no significant differences in attitudes toward
online voting or registering based on income, the poor have more
negative attitudes about e-government—participating in an elec¬
tronic town meeting or searching for government information
online than did those with higher incomes. This represents a sig¬
nificant downside to the groundswell of interest in e-government.
Surveys that trace only these general trends are not able to isolate
significant differences in the groups responding to innovations like
e-government. The benefits of e-government—easy access to infor-
mation about government policies, community activities, and ser¬

vices—may be largely untapped by the poor, even though they are
more likely than higher-income individuals to depend upon public
services. Partisan differences also surface: Democrats were more sup¬
portive of online voting, registering to vote, and accessing online gov¬
ernment information than were independents or Republicans. This
is noteworthy, because Democrats are less likely to have access to the
Internet than Republicans (see chapter 2).
Gender appeared as an important factor: men were statistically more
willing than women to register to vote online and take part in online
town meetings. This gender gap finding is consistent with that of
Bimber, who found that women were less likely to use the Internet to
initiate contact with public officials.62 Although men are more likely
than women to have e-mail addresses (as reported in chapter 2), atti¬
tudes about politics rather than technology may be driving gender dif¬
ferences. Factors other than access are also important when assessing
the potential impact of information technology on public policy.
We find that race and ethnicity are not significant factors in pre¬
dicting attitudes toward online political participation. African Amer¬
icans, Latinos, and Asian Americans do not differ significantly from
similarly situated whites in support for digital democracy.63 Latinos
were statistically less likely to be willing to search for government
information online, but otherwise there are no differences. Although
the access divide is clearly characterized by racial and ethnic dispari¬
ties, the democratic divide, for the most part, is not.
We further explored the results for online voting and registration
using predicted probabilities to compare the magnitude of income,
education, and age in shaping support for online politics.64 Because
voting is the most basic component of participation in a democracy
and previous surveys have not explored attitudes toward online
voting and registration, the responses to these questions have special
significance.
The two factors that have the greatest substantive impact on sup¬
port for online voting and registration are education and age. Hold¬
ing other demographic factors constant, support for online voting
and online registration were 19 and 22 percentage points higher
among individuals with a college degree than for those with only a
high school diploma. This mirrors existing disparities in civic partic¬
ipation, which are largely associated with educational differences. Age
WHAT MATTERS
Who Is More Likely to Support Online Voting? *
Educated (59% college degree vs. 40% high school diploma)—19-point

difference
difference
Who Is More Likely to Support Online Voter Registration?
Educated (67% college degree vs. 47% high school diploma)—20-point

difference
difference
Males (65% vs. 59% for females)—6-point difference
independent, with values for education, age, and income set at their mean for the
sample.The only statistically significant differences are reported above (see table
A5. l).We have calculated the probability that respondents agree with the above
statements, controlling for other factors.
was equally important. The simulations show a 19 percent decreased

probability in supporting online voting and a 26 percent decreased
probability in supporting online registration when moving from the
young (twenty-eight years old, one standard deviation below the
mean) to the old (sixty-one years old, one standard deviation above
the mean).
Compared to age and education, gender had a smaller impact on
attitudes toward online participation and was statistically significant
only for attitudes toward online voter registration. After holding
other factors constant, females were 6 percent less likely to favor
online voter registration than were males. Income, race, and ethnic¬
ity do not drive attitudes about online voting and registration, hold¬
ing other demographic factors constant. Partisanship, however,
resulted in significant and interesting differences. Although Demo¬
crats were least likely to have Internet access (54 percent), they were
most likely to favor online voting (50 percent) and registration (59
percent). Republicans were most likely to have access (64 percent)
and less favorable toward digital politics. Independents were least
likely to favor online voting (39 percent) and online registration (50
percent).
Finally, we examined support for online participation in a public
location and current political activity online.65 We created an index
of support for participation in a public place (where public access and
security could be provided) in regard to voting, registering to vote,
and looking up government information online. These questions
allow us to compensate for reluctance to participate online that is due
to the need for computer access, assistance, or security concerns.
As discussed previously, respondents were generally more support¬
ive of online participation in a public place. The findings reveal that
young, educated, higher-income, and male respondents, as well as
those who participate in traditional politics, are more willing to par¬
ticipate in online political activities in a public place. Public access does
not change attitudes about participation for groups that are disadvan¬
taged in terms of access or skills, but partisanship emerges as an impor¬
tant factor. Paradoxically, while Democrats were more supportive of
online voting and registration in general, Republicans were more likely
to support use of computers and the Internet for voting in a public
location. This suggests Republicans may be more concerned with
security issues than Democrats. Overall the data reveal significantly
lower support for online voting, registration, and e-government by
WHAT MATTERS
Who Is More Likely to Support Online Voting and Registration in a
Public Place?
Educated, affluent, young, male, Republican, voted in 2000 elections
Who Is More Likely to Participate Online Now?
Educated, affluent, young, male, voted in 2000 elections
Note:The only statistically significant differences are reported above (see table
A5.2).When multivariate regression is used, these are the variables that matter,
holding other factors constant.
those with lower incomes, lower education, and lower levels of civic
engagement—even controlling for public access.
In contrast to attitudes, who has actually used the Internet to find
information about politics and government? The dependent variable
for this model consists of an index ranging from 0 to 3, created from
three questions: Have you searched for political information online?
Have you looked up information on government services online?
Have you seen an online political ad?66 Again we find that the young,
better-educated, and affluent, males, and voters are more likely to be
currently engaged in online political activities. There was no differ¬
ence between Democrats, Republicans, and independents in present
use, suggesting that none of these groups would benefit from online
voting and registration in the near term. Our multivariate analysis
confirms (and extends) the Pew e-government findings that were
based on descriptive statistics. In sum, individuals with lower
incomes and education and those currently not civically engaged are
the least likely to use e-government or participate in politics online,
paralleling inequalities in traditional participation.
Analysis of our survey responses on voting in the 2000 elections
allows us to compare current participation, as reported by our sam¬
ple, with interest in digital democracy.67 Our analysis of voting in the
2000 election agrees with other research on voting. It indicates the
poor are significantly less likely to vote, while the educated and elderly
are more likely to participate. Those with a political orientation
(Republicans and Democrats) are more likely than independents to
vote, and females more likely than males. Race and ethnicity also mat¬
ter: African Americans are more likely than whites to participate, and
Asian Americans less likely than whites to do so.
What, then, are the likely consequences of moving political par¬
ticipation onto the Internet? In short, our data on willingness to use
information technology for political purposes reveal an online demo¬
cratic divide individuals with higher education and income are
more supportive of digital democracy, and are more likely to partici¬
pate in politics online, than the poor and those with lower education.
In contrast to Alvarez and Nagler’s study of turnout in the Arizona
primary, we do not find that race is significant for attitudes about
most online participation, controlling for other factors.68 The excep¬
tion was Latino attitudes about e-government, which were less favor¬
able than those of whites. African Americans are currently more likely
to vote than whites, controlling for education, age, and income. This
is not true of support for Internet voting and other forms of online
participation. The statistical analysis shows an absence of racial fac¬
tors affecting attitudes about online participation.
Attitudes toward digital democracy are influenced by gender, par¬
tisanship, and age. Women are more hesitant about many political
uses of the Internet, but the differences are relatively small. The par¬
tisan impact is unclear. Democrats are more supportive of online reg¬
istration and voting and e-government, but Republicans are more
supportive than Democrats if these activities occur in a public setting.
More important, there are no significant partisan differences in use
of the Internet for political purposes. The young, however, are clearly
more interested in online participation and may become more
involved in politics if online voting and registration are implemented.
On the down side, the data provide compelling evidence for those
who argue that online politics will mirror, or exacerbate, existing dis¬
parities in the composition of the electorate based on socioeconomic
status. On a positive note, the fact that younger respondents are more
supportive of digital democracy suggests the potential for expanding
the electorate to include a group that has been traditionally under¬
represented. For the young, digital democracy and government may
increase their civic engagement because of its convenience and their
comfort with new technology. The importance of age in our findings
is consistent with previous research on the Arizona Internet voting
primary. 69
Dilemmas for political participation online largely mimic the
problem of traditional political participation—those who are better
educated are more interested and more able to participate. This indi¬
cates that addressing the democratic divide requires more than a
technical solution, but attention to educational disparities as well.
Surveys provide one method of predicting the future path of
online democracy. Our case study of a local experiment with an
online town meeting allows us to probe other issues regarding both
digital democracy and e-government and to connect them to our
survey findings. We are interested in finding out who participated in
the online town meeting and whether digital democracy holds
promise for expanding participation. Talking to public officials
about various uses of the Internet also allows us to put some aspects
of e-government into perspective.
Berkeley, California’s, Online Experiment
Berkeley, California, is a community at the cutting edge of experi¬

mentation with “e-democracy.” Berkeley is a natural incubator for
ideas joining e-government with citizen participation. In the shadow
of Silicon Valley, the city is able to draw upon a regional culture of
digital innovation and a “wired” population of University of Califor¬
nia students and academics. The city has a long tradition of political
participation harking back to the early 1960s and the Free Speech
movement on the Berkeley campus. Although many of the issues have
changed, that tradition survives at the local level in public hearings
and other city meetings that often attract forty or fifty people from a
city of just over 100,000 residents. City officials cite problems with
public hearings that straggle on past midnight because so many citi¬
zens are waiting their turn to be heard.70 Berkeley is the type of city
where digital-divide issues are likely to emerge, for it shelters low-
income, as well as upscale, neighborhoods on its quiet, tree-lined
streets. The city also boasts a kaleidoscope of races and cultures: about
half of its residents are white, more than 16 percent are Asian Amer¬
ican, almost 14 percent are African American, nearly 10 percent are
Latino, and approximately 10 percent are from other races or are
multiracial.71 Innovation, participation, and diversity make Berkeley
a good test site for learning about e-democracy and the impact of the
digital divide for online political participation.
Berkeley residents had the opportunity to register online their
opinions about revisions to the city’s general plan during the year
2000. The city cooperated with a nonprofit group called Moveon.org,
which developed a software program called ActionForum, and used
the Berkeley general plan as its first trial.72 Because of the legal issues
entangled in sponsoring an official online forum, the city did not host
the forum itself.73 Berkeley is technically an example of digital demo¬
cracy rather than e-government, because the city did not sponsor the
online town meeting; however, the city gave its blessing to the group,
announcing the effort in a press release and brochure.
The ActionForum website displayed a copy of the general plan and
allowed citizens to make comments that listed their real names, city
of residence, and occupation. The software included a feature that
allows other site visitors to agree or disagree with comments and rank
them in terms of their importance. Highly ranked comments rise to
the top of the list, and lower-ranked comments drop to the bottom.
Individuals who read or rate comments remain anonymous, in con¬
trast to those who post their thoughts.74
In the assessment of both the city and Moveon.org, the response to
the website was limited but nevertheless useful for highlighting some
aspects that could be improved in the future, as well as some thorny
issues that face such an enterprise. The archived files show thirty-three
individuals who participated, but this number included the software
developers and their friends. Most of those who participated, accord¬
ing to Moveon.org, were Berkeley activists, so the website did not suc¬
ceed in enlarging the circle of participation. It did, however, attract
some attention beyond those who posted comments. According to
Berkeley’s communications manager, city employees found it useful to
review the comments, and seventy-eight people who were not city
employees looked at the site. The forum went online in February
2000, and Moveon.org submitted the final results to the city in
August. Citizen input was not effective in shaping the final policy out¬
comes in this case, because the planning commission decided to scrap
the staff’s draft of the general plan and develop their own. The Action-
Forum was based on the staff version of the proposed revisions.
Limited participation resulted at least partly from the experimen¬
tal nature of the endeavor. The city did not actively promote the ini¬
tiative, because it was an initial pilot and there were many questions
about how to implement it. The general plan was also a complex doc¬
ument, about 170 pages long, covering more than 600 different poli¬
cies. As the communications manager suggested, those who did not
traditionally participate would have found this a “daunting” first step.
It is difficult to know whether issues of technology access and skill
made a difference in the Berkeley project, but access was provided at
twelve public libraries in the city. City officials were concerned about
other aspects of disenfranchisement as well, including the problems
of working parents and others who were unable to attend all-night
meetings.
The quality of civic discussion is a concern for online forums, as
well as the quantity of participation. The prevalence of “junk” on e-
democracy websites presented a problem that the creators of the
ActionForum attempted to solve with their system for rating com¬
ments. One possible difficulty, however, is that judgments may simply
reflect the popularity of the opinion rather than its thoughtfulness. The
software developers were pleased that none of the comments on the

general plan qualified as junk, in their assessment, but more exten¬
sive use of the software is needed to conclude that the rating sysfem
encourages more civil and considered discussion.
Despite the limited participation in the online forum, the Inter¬
net has influenced communication between citizens and government
officials in Berkeley other ways. The general plan manager com¬
mented on the burgeoning use of e-mail to communicate with offi¬
cials. The speed, ease, and informality of e-mail encourage people to
weigh in with their concerns. Getting this input from citizens has
made the plan manager’s job “more fun, and less bureaucratic.” The
communications manager noted that posting documents and other
information online required “a transition from bureaucratic speak to
a more conversational tone on the Internet.” The effort to move gov¬
ernment processes onto the web and to make them more transparent
has forced a reexamination of how government operates. “The soft¬
ware or being online is not really the crux of the issue,” said the city’s
technology manager. The crux of the issue is looking at how we do
things internally, [and whether] our procedures help or hinder civic
engagement.”
Although digital democracy, even in Berkeley, has had limited
impact, e-government is incrementally changing some of the rela¬
tionships between citizens and government. The Berkeley initiative
demonstrated that holding such a forum is technically feasible and
that nongovernmental organizations may have an important role in
facilitating such discussions. But the reality is far from the ideal
espoused by advocates of participatory democracy. The Berkeley
experiment was limited for some reasons that may not apply to other
efforts. Yet legal issues, the tenor of public discussion, and the lack of
interest expressed by less-educated and low-income individuals in our
survey indicate more general potential barriers to widespread use and
effective participation.
Participation in the online forum was dominated by a select group
of political activists, mirroring traditional participation in city coun¬
cil meetings. These findings are consistent with those of Davis, who
argues, “The Internet will not lead to the social and political revolu¬
tion so widely predicted. . . . Internet users will continue to be the
affluent, the already politically interested and active.”75 Although we
take a more optimistic position and applaud governments for innov-
ative attempts to increase dialogue with citizens using information

technology, the case study suggests some limitations in the potential
for expanding participation in government online.
Conclusion
Technology promises to have an increased impact on the way in

which individuals interact with government and participate in poli¬
tics. Online voting and voter registration may be a reality in a num¬
ber of states by the 2004 presidential elections.76 In the first binding
test of online voting, Arizona Democrats decided in 2000 to elect
national convention delegates through Internet voting. In theory, the
Internet may provide a means of updating the election system for an
information-based society. Access to online political news may also
enhance information about candidates and elections, stimulating
increased citizen participation.77
Our findings on attitudes about digital democracy reveal a con¬
tradiction between theory and practice. Many Americans are hesitant
about the use of the Internet for purposes such as voting, and others
are clearly less interested in online political participation than in uses
such as job search and taking courses online. Although e-government
delivery of information and services is popular, Berkeley’s experiment
with an online political forum indicates that there are many hurdles
for participatory uses of the Internet, in contrast to the largely infor¬
mational uses of e-government.
Consistent with cross-national accounts of the digital divide, our
survey data reveal an online democratic divide—individuals with
higher education and income are more supportive of digital demo¬
cracy and e-government, and are more likely to participate in politics
online, than are the poor and those with lower education.78 The rea¬
sons are not entirely clear. Individuals with limited educational back¬
grounds may not have the necessary skills or confidence to go online,
or they may simply have negative or apathetic attitudes toward poli¬
tics and government.79 According to our survey results, the willing¬
ness of individuals to use technology for political participation in its
various forms is low (with the exception of e-government) in com¬
parison with use of the Internet for economic advancement. The
analysis provides evidence that online politics will mirror, or even
I 10 Virtual Inequality
exacerbate, existing patterns of unequal political participation based

on income and education. Representation of racial and ethnic
minorities in online politics is unclear. At present, African Americans
are more likely to vote than similarly situated whites, but they are no
more likely than whites to express interest in online participation,
controlling for factors such as income and education.
On the other hand, the young emerge as a group not only more
likely to have access to the Internet and computers, but that is also
significantly more supportive of digital democracy and e-govern-
ment. Information technology may increase civic engagement among
the young, altering and perhaps expanding the electorate over time.
Our prediction of the impact of digital democracy on the represen¬
tation of the American electorate is therefore mixed. The Internet will
neither serve to replicate politics as usual nor transform governance
and restore levels of mass political participation. Although the Inter¬
net promises to have some positive effects, it will not erase, and may
even underscore, the bias of limited participation and representation
in American politics that E. E. Schattschneider decried decades ago:
“The flaw in the pluralist heaven is that the heavenly chorus sings
with a strong upper class accent.”80
Education emerged as the most important factor in the democratic
divide. Support for online voting and online registration were 19 and
22 percentage points higher among individuals with a college degree
compared to those with only a high school diploma. This suggests
that, in order to close the democratic divide in cyberspace, as well as
traditional politics, education will be crucial as well as access to tech-
nology. More than 200 years ago, Thomas Jefferson argued that pub¬
lic education was necessary for an educated citizenry and for the
health of the republic. In the future, attention to information liter¬
acy and access may be mechanisms for achieving equal opportunity
in the political sphere, but participation will also be rooted in factors
that have traditionally been associated with civic engagement.
Notes
1. Putnam 2000.
2. Abramson 1983; Campbell et al. I960; Conway 1991; Wolfinger and
The Democratic Divide II I
Rosenstone 1980; Rosenstone and Hansen 1993; Piven and Cloward 1988;
Verba and Nie 1972; Verba, Schlozman, and Brady 1995.
3. Smith 1989.
4. Bimber 2001; Norris 2001; Alvarez and Nagler 2002; Shah, Kwak,
and Holbert 2001; Scheufele and Shah 2000; Solop 2000; Tolbert and
McNeal 2003.
5. Davis 1999.
6. Campbell et al. I960; Wolfinger and Rosenstone 1980.
7. Norris 1999; Rosenthal 1997; Dionne 1996; Baldassare 2000,
chap. 2.
8. Peters 1996, chap. 3.
9. Pateman 1970.
10. On “discursive democracy,” see Dryzek 1990; on “strong democracy,”
see Barber 1984; on “teledemocracy,” see Toffler 1995; on “deliberative
democracy,” see Fishkin 1993; on general participatory models, see Peters
1996.
11. Budge 1996.
12. Dryzek 1990; Barber 1984.
13. Butler and Ranney 1994; Bowler, Donovan, and Tolbert 1998;
Bowler and Donovan 1998; Gerber 1999; Magleby 1984; Mendolsohn and
Parkin 2001.
14. Norris 2001, 96; Toffler 1995.
15. For a more general discussion, see Norris 2001, 97-98 and 95-111.
16. Rheingold 1993; Budge 1996; Hague and Loader 1999; Grossman
1995.
17. Initiative and Referenda Institute 2002.
18. Gibson 2002; Alvarez and Nagler 2002.
19. Alvarez and Nagler 2002; Wilhelm 2000; Margolis and Resnick
2000; Putnam 2000, 166-80. According to critics, there are other draw¬
backs to online politics as well. For example, some argue that the Internet
will narrow the focus of attention by fostering selective exposure to political
information consistent with individual preferences and interests. Reduced
exposure to conflicting views may reduce citizen political tolerance (Sunstein
2001).
20. Wolfinger and Rosenstone 1980.
21. Norris 2001, 98; Putnam 2000, 174-75. See also Davis and Owen
1998; Davis 1999; McChesney 1999; Wilhelm 2000.
22. Bimber 2001.
23. Tolbert and McNeal 2003.

24. Weber and Bergman 2001.
25. Shah, Kwak, and Holbert 2001. *
26. Rosenstone and Hansen 1993; Verba, Scblozman, and Brady 1995.
27. Bimber 1999.
28. Gibson 2002.
29. Solop 2000.
30. Alvarez and Nagler 2002.
31. King 1997.
32. Multivariate regression analysis suggests elderly, nonwhite, unem¬
ployed and rural residents were also statistically less likely to engage in Inter¬
net voting, controlling for other factors (Alvarez and Nagler 2002).
33. With only fifteen counties in the state of Arizona, the estimates of
minority and white voters are so aggregated that they may not be represen¬
tative of actual statewide voting patterns. Such a small number of cases vio¬
lates one of the assumptions of multiple regression. The central limit
theorem requires that there be a minimum of fifty cases in order for the
dependent variable to be normally distributed. With such a small number
of cases, the research runs the risk of incorrect inferences.
34. Clift 2000.
35. Markle Foundation 1999; Norris, Fletcher, and Holden 2001.
36. West 2000, 2. See also Chadwick 2001; Clift 2000; Norris 2001; Tap-
scott 1997.
37. Chadwick 2001. See www.firstgov.gov.
38. See Stowers 1999 and Norris, Fletcher, and Holden 2001.
39. Seifert and Petersen 2002; Stowers 1999.
40. Fountain 2001, 6; Seifert and Petersen 2002.
41. West 2001.
42. Gartner Group 2000.
43. Chadwick 2001; Fountain 2001,4; West, forthcoming; but on e-gov-
ernment at the local level, see Ho 2002.
44. National Performance Review 1993; for similar arguments about
reinventing government, see Osborne and Gaebler 1992.
45. Seifert and Petersen 2002.
46. Layne and Lee (2001) delineate a four-stage model of e-government
evolution (cataloging, transaction, vertical integration, and horizontal inte¬
gration). The availability of transactions on the web represents advancement
to at least the second stage of implementation. Others define the evolution
of e-government by the four stages of presence, interaction, transaction, and
transformation. While an example of “presence” is a basic website that lists

cursory information about an agency, hours of operation, mailing address,
or phone numbers but has no interactive capabilities, “interactive” web-
based initiatives offer enhanced capabilities, including instructions for
obtaining services or downloadable forms to be printed and mailed back to
an agency. “Transaction” allows clients to complete entire tasks electronically
through self-service operations such as license renewals, paying taxes and
fees, and submitting bids for procurement contracts. Transformation is the
highest order of evolution for e-government initiatives, including robust
customer relationship management capabilities required to handle a full
range of questions, problems and needs (Seifert and Peterson 2002).
Although there are currently few examples of this type of initiative, some
suggest that, at its most advanced level, e-government could potentially reor¬
ganize, combine, and or eliminate existing government agencies and replace
them with virtual ones.
47. West, forthcoming.
48. Stowers 1999.
49. Fagan and Fagan 2001; Fountain 2001, 21-22; Stowers 1999.
50. Norris 2001, 128. See also Clift 2000; Melitski 2001; Tapscott 1997.
51. West, forthcoming; Clift 2000.
52. West, forthcoming.
53. Seifert and Petersen 2002.
54. See West 2000.
55. McNeal et al. 2003.
56. Kettl 2000.
57. West 2001; Chadwick 2001; McNeal et al. 2003.
58. Pew Internet and American Life Project 2002.
59. West 2001; Pew Internet and American Life Project 2002. The pub¬
lic opinion poll reported in West (2001) also shows that men are more likely
than women to have used e-government. The poll cited by West is the Hart-
Teeter for the Council for Excellence in Government national survey taken
in August 2000.
60. Pew Internet and American Life Project 2002; West 2001.
61. Because the dependent variables are binary, logistic regression was
used to analyze the data. See table A5.1 for the results.
62. See Bimber 1999.
63. The result showing that Asian Americans are less likely than whites to
support e-government is statistically significant at .10, which is borderline.
The sample for Asian Americans was also small, so we do not have sufficient
confidence in the results for Asian Americans to report them as significant

findings.
64. See King, Tomz, and Wittenberg 2000. We calculate the changfe in
the probability of access and support for online participation caused by mov¬
ing from a variable’s high to low value while simultaneously keeping all other
variables set to their mean (or 0 or 1 category for dichotomous variables).
The change in the probability of support for online participation caused by
moving from low to high values of the independent variables allows for effec¬
tive substantive comparisons across independent variables.
65. Table A5.2 shows the results of a multivariate regression using com¬
bined scales for attitudes about online political participation, attitudes about
online participation in a public place, and actual experience with online par¬
ticipation. Since the dependent variables are measured on an ordinal scale,
our estimates are based on ordered logistic regression.
66. See the last column in table A5.2 for the results of the multivariate
regression analysis.
67. We created a dummy variable for the 2000 elections, coding
responses as 1 for voting and 0 for not voting.
68. Alvarez and Nagler 2002.
69. Solop 2000.
70. Both city officials and community activists noted this problem of a
participatory bottleneck.
71. U.S. Census Bureau 2000.
72. The software developers involved in Moveon.org and the ActionFo-
rum are Berkeley residents and creators of the Broederbon Software “flying
toaster” screensaver and the popular computer quiz game “You Don’t Know
Jack.”
73. The collaboration between the city and Moveon.org was beneficial to
both parties. Organizations outside city government have more discretion
in monitoring a website for slanderous content or in controlling content
through ranking or other methods. (First amendment issues prevent city
governments from censoring material in any way.) Moveon.org valued the
feedback it received from the city, and the city saw this as an important
experiment with a new venue for participation. The first amendment issue
is likely to have a broader influence in the implementation of officially spon¬
sored online forums. Other legal issues specific to California hampered the
exchange between officials and citizens through the forum. The city attor¬
neys office ruled that council members and members of commissions were
not allowed to participate in the online forum, based on California state law
The Democratic Divide I 15
called the Brown Act” that prohibits elected officials from gathering to dis¬
cuss issues or make decisions outside of public meetings.
74. For a demonstration and the archived comments from the Berkeley
general plan, go to www.actionforum.com.
75. Davis 1999.
76. Brookings Institution 2000.
77. Tolbert and McNeal 2003.
78. Norris 2001.
79. James 2001.
80. Schattschneider I960, 34—35.
Chapter i
Beyond the Divides: Toward

Opportunity and Equity
with Lisa Dotterweich
ur analysis has allowed us to achieve several goals: to move

beyond a narrow definition of the “digital divide,” to better
describe patterns of access and skill, and to describe and pre¬
dict the impacts of these disparities for economic opportunity and
democratic participation. We accomplished this by expanding the
definition of the problem so it includes the range of public policy con¬
cerns on the issue and by gathering data on skills, attitudes, and expe¬
riences relevant to these concerns. We have provided more accurate
evidence than existing studies have done, through a large sample of
low-income and minority respondents and the use of statistical con¬
trols. In short, we have presented a number of arguments about the
ways in which the policy “map” of the digital divide must be altered
to more closely resemble the reality of information technology-
related disparities. In this chapter, we briefly trace the new geography
of the divides and their policy significance. We then highlight pat¬
terns that cut across the four divides and discuss current and future
public policy.
Measuring the Access Divide
Despite arguments to the contrary made by policymakers and schol¬

ars, the access divide that appeared during the 1990s persists. The evi-
Beyond the Divides:Toward Opportunity and Equity I 17
dence has sometimes been muddled by weak research methods and

unsophisticated analyses. The debate has been confused by a number
of claims that we hope to lay to rest.
First, it is erroneous to allege that the access divide is dead. Endur¬
ing, and statistically significant, gaps remain in terms of income, race,
ethnicity, education, and age, even as more Americans join the ranks
of those online. These differences appeared in our analysis of Inter¬
net access since the mid-1990s, and they still matter today for Inter¬
net access, e-mail addresses, and computer ownership. The gender
gap has been more transient, but we cannot make the assumption that
the other disparities are vanishing as well. The momentum gathering
behind diffusion does not erase ail differences.
Second, some studies have suggested that race and ethnicity do not
affect the access divide. Our data, with its large sample of minorities,
demonstrate that African Americans and Latinos have significantly
lower rates of home computer ownership, e-mail use, and Internet
access than whites, even when we control for differences in age, edu¬
cation, and income.
Factors such as education, income, and age, however, influence
access to an even greater extent than race and ethnicity. The survey data
we examined from 1996 onward showed that income played a more
complex role in the early days of the Internet but that better-educated
and younger individuals have been enthusiastic users throughout. This
suggests an array of factors are involved in the access divide—the afford¬
ability of technology, knowledge of technology and the skill to use it,
and exposure to, and attitudes toward, technology use.
Uncovering the Skills Divide
It is clear that a skills divide also exists and that it closely follows the
contours of the access divide. Those who are most likely to need assis¬
tance with computers are older, less-educated, low-income, African
American, and Latino. Age and education make the greatest differ¬
ence. Both dimensions of skill—technical competence and informa¬
tion literacy—echoed the same disparities as the access divide.
The existence of a parallel skills divide, while not surprising, has a
number of troubling implications. To us, the question of skills tran¬
scends the issue of computer ownership and home Internet access, for
it is technology skills that most clearly enable individuals to pursue

opportunities for economic advancement or political participation,
as well as other goals. Nevertheless, more than a fifth of our sample
reported needing help with the simplest of all tasks—using a mouse
and keyboard—indicating that they were completely devoid of any
technical competence. Approximately one-third needed help locating
information using computers. This may demonstrate a lack of famil¬
iarity with computerized data bases and the Internet, but it could also
indicate more serious problems with basic literacy and with naviga¬
tion through the proliferating sources of information in modern soci¬
ety. Our data cannot address this latter question directly, but it is
certainly a topic worth pursuing in further research.
The access and skills divides matter for public policy because of their
potential to influence economic opportunity and civic participation.
Pronounced inequalities in these areas, of course, already exist. The pol¬
icy question is whether the use of technology promises to exacerbate or
ameliorate current inequities. To explore this question in the areas of
economic opportunity and political participation, we measured the
possible impact of the access and skills divides through the attitudes of
respondents as well as their experiences. We reason that, even where
experience is limited, attitudes give us some evidence of the interests
and aspirations of our respondents. We can therefore better define the
problem—do people lacking access and skills also lack an interest in
information technology and its uses? With experimental or rare appli¬
cations such as Internet voting and electronic town meetings, we can
find out how receptive the general public is to such innovations and
whether attitudes differ among various groups in the population.
Assessing the Economic Opportunity Divide
Computer use on the job is prevalent and growing and has con¬
tributed toward rising skills demands in the work force. We found
that just under half of our respondents with a high school education
or less reported actually using the computer at work at least once dur¬
ing the past month. About one-quarter of these less-educated work¬
ers used the Internet.
Most Americans identify computer skills with economic opportu¬
nity, and the poor and less educated are no different. Attitudes about
Beyond the Divides:Toward Opportunity and Equity I I9
technology and economic opportunity do differ, however, by age,

race and ethnicity, employment status, and, to a modest extent, gen¬
der. Older respondents are less interested in technology and economic
opportunity. African Americans, Latinos, the unemployed, and
women, however, seem particularly sensitive to the potential connec¬
tion between technology and opportunity. This suggests that, in low-
income and minority communities, the task is not so much to
convince individuals that these skills are necessary, but rather to make
technology and instruction available and relevant to job search, career
development, and small business needs. Experience with the devel¬
oping area of online job search supports this contention. African
Americans are more likely than whites to have actually used the Inter¬
net to search for a job or to take a course online, even though African
Americans are statistically less likely to have access. We conclude that,
although the access and skills divide may in fact diminish economic
opportunities for those who are already disadvantaged, substantial
interest in information technology and economic opportunity exists
among many disadvantaged groups.
Mapping the Democratic Divide
Technology promises to have an increased impact on the way in which

individuals interact with government and participate in politics, as
debates over Internet voting and the advent of online registration and
absentee voting show. Some scholars predict that minorities and the
poor will be disenfranchised by these developments, although the
empirical record is based on limited use of these innovations. Our data
on attitudes regarding various forms of online participation, therefore,
open a unique window to the future, amid competing claims, hype,
and confusion over the impact of the Internet in the public sphere.
We found that use of government websites for information about
services was more popular among respondents than was use for online
political participation. More than three-quarters of Americans were
willing to look up government information online, but slightly less
than half were willing to vote online or participate in an online town
meeting. Support for online voter registration was considerably
higher, and support for Internet voting increased 10 percentage points
if it occurred in a public place where access and security issues are less
relevant. Some partisan differences appeared, with Democrats gener¬

ally more supportive of online registration and voting but Republicans
more supportive of online voting if it occurred in a public place. Our
survey results and our case study of an online forum in Berkeley, Cal¬
ifornia, indicate that widespread use of the Internet for voting and
town meetings is not on the immediate horizon, but it may be in the
future if difficulties such as legal issues and security are resolved.
Attitudes about politics and government online differ among
groups, but they promise to reinforce most existing disparities in
political participation. Those who are most likely to support use of
the Internet for voting, registering, looking up government informa¬
tion, and participating in town meetings are generally those who par¬
ticipate now (the educated and the affluent, and those who voted in
the prior election). The major way in which technology augurs
change in patterns of political participation is through its potential to
mobilize the young. The young were consistently more likely to
express interest in online political participation across the range of
questions we asked.
The Internet may constitute a double-edged sword for political
participation—mobilizing some groups that are currently less likely
to participate in politics (the young) while perpetuating or perhaps
even magnifying disparities based on education and income. Our
findings agree with the well-established body of research on political
participation, that lower-income and less-educated individuals are
least likely to participate. Race and ethnicity did not matter for most
questions about political engagement in our survey (except that Lati¬
nos were less supportive of e-government than whites). Traditionally,
African Americans are more likely to vote, controlling for income and
education, but this did not appear to be true in regard to online vot¬
ing or other forms of participation.
Across the Four Divides
What is our prognosis for the policy landscape of the future, in light
of our assessment of these four divides? We find both barriers and
resources for bridging the divides.
The access divide currently presents a gloomy picture, with per¬
sistent gaps in home access based on age, income, race, ethnicity, and
Beyond the Divides:Toward Opportunity and Equity 121
WHAT MATTERS
What Matters across the Four Divides
Barriers Resources
Access
Disparities by income, education. General growth in computer and
race, ethnicity, and age Internet access; high likelihood of
access for young
Skills
Disparities that mirror access divide; Generally positive attitudes toward
poor and old less interested in learning and public access, especially
learning skills among the young and African
Americans, sometimes Latinos
Economic Opportunity
Lack of computer skills, reading, and Generally positive attitudes about
math for economic mobility and technology for economic advance-
opportunity ment; especially among the young and
African Americans, sometimes Lati¬
nos, women, the unemployed
Democratic Participation
Continued or increased disparities Positive attitudes among young, who
based on education, income, interest are currently underrepresented
(measured by previous participation)
education. The skills divide is mostly ominous as well, for findings

on the skills front generally reinforce the findings on access. The
access and skills divides seem closely linked, perhaps in a vicious cir¬
cle—those without skills have little need to use computers, and those
without frequent availability have little chance to develop the skills
that they need through trial and error and practice. Frequent com¬
puter use occurs at home or work, underscoring the relevance of
home access for developing acumen. Yet the skills divide is more com¬
plex than either the access debates or needs for technical competence
suggest. Reading skills and information literacy are needed both to
promote further diffusion of information technology and to allow
individuals to capture the full benefits of technology. The barriers
are more substantial than indicated in previous political discourse on

the issue.
There is cause for some optimism, however, in the favorable opih-
ions that most Americans hold regarding technology and their will¬
ingness to use it for many purposes. African Americans, in particular,
express positive attitudes about public access, learning new skills, and
using technology for economic advancement. The results on the
democratic divide give rise to both optimism, because of technology’s
potential to involve younger citizens, and pessimism, because the
Internet promises to perpetuate and perhaps widen disparities in
political participation currently based on income and education.
More generally, how will differences based on age, gender, income,
race, ethnicity, and education develop in the near future?
Younger individuals, who have enthusiastically embraced the
Internet, will continue to lead the way in information technology use.
Disparities in access and skills based on age will disappear over time.
Public policy may encourage older individuals to gain new computer
skills at work and at public access sites such as libraries and commu¬
nity centers. For the workforce, however, it is those who have a num¬
ber of years before retirement who are most likely to represent a policy
problem, if they lack the skills to enjoy some choice and mobility in
the labor market. Training and continuing education are concerns for
this segment of the population. Our examination of technology use
for political participation suggests that widespread use for voting is
somewhat distant. Steps such as training to use newly digitized vot-
ing machines, assistance at public access sites, and alternatives such
as mail ballots may help to ease the transition toward greater use of
technology for voting. Public access may also be helpful in encour-
aging older individuals to take advantage of information and services
available through government websites.
Most gender gaps in access and skills have vanished over the years,
and the gender differences that remain primarily involve attitudes
toward learning (preferences for more hands-on instruction) or use
of technology (less enthusiasm for Internet job search or participa¬
tion in an online town meeting). This suggests a difference in inter¬
ests rather than an inability to participate economically or politically.
One reason that the gender-based access and skills divides have been
temporary, we hypothesize, is that women have similar educational
opportunities as men. As computers and the Internet became more
Beyond the DividesrToward Opportunity and Equity 123
widespread, and the range of applications increased, women had the

ability to master the new technology. Disparities based on income,
race, and ethnicity may prove more persistent, for the quality of edu¬
cation in the United States varies greatly along lines of class and race.
Even if we control for educational attainment, our measures tell us
little about the content of that education and how well it has prepared
individuals to learn new skills and to adapt to social change.
Income was significant for predicting disparities in access and
skills, which suggests that access and skills are intertwined. As men¬
tioned above, lower-income individuals may lack the educational
experiences necessary to prepare them for technology use, even if we
control for educational attainment, and so may be less likely to invest
in computers. But the cost of computers and Internet access may also
represent a hurdle for low-income households. Even as prices fall,
computer purchases compete with other needs, especially for the
poorest. As we have seen, frequent computer and Internet use occurs
at home or at work, so those without home access may have fewer
chances to develop skills.
Low-income respondents were less likely to be interested in using
the Internet for e-government and online town meetings, but they
were not less supportive of using the Internet for voting or register¬
ing to vote. Disinterest in e-government is of particular concern,
because, unlike Internet voting or online forums, e-government has
been widely adopted by local, state, and federal governments. It is the
most popular use of technology in our survey, with even more wide¬
spread support than online job search or distance learning. Apathy or
negative attitudes about e-government among the poor distinguish
them as markedly different from the rest of the population. The
potential benefits of e-government, such as greater transparency in
government and easier access to public services, will largely bypass the
poor. Because low-income people often depend upon public services
for their daily needs, this finding is all the more disconcerting.
The poor did not differ from other Americans in their favorable
attitudes toward technology as a tool for achieving economic oppor¬
tunity, and the unemployed were actually more likely than the
employed to think that computer skills are needed to get ahead. Dis¬
interest in e-government and online forums may signal disengage¬
ment from government rather than apathy toward technology. The
problem of limited political participation among the poor predates
the Internet, and the answer to this particular problem lies outside of
the realm of technology.
Race and ethnicity have been prominent in debates over the digi¬
tal divide. Our data show that race matters, but in complex and some¬
times puzzling ways. Income and education alone are not sufficient
to explain the gap between African American and white access and
skills. Some commentators have argued that African Americans do
not see the content of the Internet as relevant and that there is a cul¬
tural divide. Yet African Americans (and, less consistently, Latinos)
had more-positive attitudes than whites about public access, learning
new skills, and using technology in pursuit of economic opportunity.
This means that we need to look elsewhere for explanations. Racial
segregation, concentrated poverty, and a lack of exposure to technol¬
ogy within poor African American neighborhoods may have some
effect over and above the influence of family income and individual
educational attainment in these areas. In low-income minority com¬
munities, institutions such as schools and libraries may lack the
resources to provide adequate educational preparation, access, and
computer training. Personal networks may not include friends and
relatives who are themselves “plugged in” and able to provide encour¬
agement and mentoring. We cannot provide direct evidence for these
conjectures, or some of the earlier hypotheses about the quality of
educational experiences, but they make sense in view of the data.
The positive attitudes expressed by African American respondents
also provide some ammunition for those who are worried about racial
stereotypes being unintentionally perpetuated by attention to a racial
dimension of the access and skills divides. The problem is not that
African Americans do not “get” technology. In fact, they hear the
message more clearly than most and may be even more motivated to
learn when given a chance. Despite lower rates of access and skills,
African Americans are more likely than whites to have used the Inter¬
net for online job search and distance learning.
Finally, the issue of education looms large across the four divides
and, to us, presents the most urgent need for preparing workers and
citizens for the information age. The results of national literacy sur¬
veys cast doubt on the ability of many Americans to cope with the
demands of technology, work, and political participation. Education
already influences economic advancement and political participation,
apart from the issue of information technology. With the advent of
the Internet, basic skills such as reading, the ability to locate infor¬
mation, to use it to solve problems, and to evaluate its appropriate¬
ness are all necessary to use information technology to its full
potential. Similar skills have been identified more generally as work¬
place competencies in the new economy, and the Internet raises the
literacy bar for political participation to some extent. The significance
of education is clear if we examine who has the most experience using
(and is presumably most able to use) the Internet in a variety of ways.
Digital Experience and Education:

Evidence from the Survey
As another way of summarizing the results of the survey, we created

an index of seven questions regarding applications of technology to
measure what we call “digital experience” in key areas: whether an
individual (1) can locate information on the web, (2) has searched for
political information online, (3) has looked up information on gov¬
ernment services online, (4) has searched for or applied for a job
online, (5) has taken a class online, (6) has used the computer to do
homework, and (7) has used the computer to find books in the library
without assistance. Together, these responses not only measure expe¬
rience with information technology, but also imply the ability to use
technology to gather information and perform complex tasks. Posi¬
tive responses to each question were coded 1 (0 for negative re¬
sponses), and then the seven questions were summed to create an
index ranging from 0 to 7.
A frequency graph (histogram, shown in figure 6.1) of the digital
experience index reveals a relatively bell-shaped curve, which would
be expected with a “normal” distribution, where the majority of
people fall around the average. The curve is somewhat skewed,
however, at the lower end. The spike at the far left shows that a num¬
ber of respondents scored 0, having no experience in using informa¬
tion technology for any of these purposes. The fact that these
respondents were not able to even locate information on the Inter¬
net suggests that many of these individuals lack rudimentary skills
as well as experience. The mean score was 3.1, indicating that the
average respondent in the survey answered positively to three of the
seven questions. About 37 percent of respondents could be classified
Proficiency—Number of Tasks Respondent Can Complete
Nofe: Responses to seven questions were used to create the index of digital experience,
which approximates a normal distribution. Each column represents the number of sur¬
vey respondents who could complete that number of online tasks. The graph is slightly
skewed to the left, reflecting the low-income sample.
Figure 6.1 Distribution of Digital Experience in the United States
as having low digital experience, with scores ranging from 0 to 2 on

the index. At the same time, 28 percent could be defined as highly
experienced, with scores ranging from 5 to 7 on the index. Thirty-
five percent had moderate experience, with scores of 3 or 4.
What factors explain varying levels of digital experience among
survey respondents? Using the index as the dependent variable, or the
result to be explained, we examined the influence of the demographic
factors used throughout the book as well as several measures of access
to information technology: computer ownership, an e-mail address,
and home access to the Internet^ Our analysis reveals that digital
experience was related to home Internet access and e-mail access (but
not computer ownership). Apparently home Internet access is more
conducive to digital experience than access that is limited to either

work or public access venues. When controlling for access, digital
experience, however, is not significantly related to race, ethnicity, or
income, but it is clearly linked to education. Digital experience is to
some extent a result of interest in the particular activities we included
in the index as well. Men, the young, the employed, and the educated
have more digital experience, consistent with the patterns of interest
in online politics, job search, and distance learning found in earlier
chapters.
The policy issues that emerge, then, are access, skills, and education.
What resources currently exist for closing the access and skills gaps?
Where do our data lead us in terms of policy recommendations?
The Current Environment
A number of government and nonprofit initiatives attempt to

broaden access to technology. By one recent count, more than 20,000
programs exist.2 This flurry of activity to some extent conceals the
small scale and tenuous existence of many of these efforts, their patch-
work coverage of poor communities, and a lack of research on the
effectiveness of existing programs. Nevertheless, some progress has
been made in terms of providing more hardware in schools and
libraries. The challenge for the future will be to maintain and expand
these resources and to use them effectively to develop skills and to
meet the needs of the communities they serve.
The major programs currently in place address public access in
schools, libraries, and community technology centers (CTCs) rather
than home access. Government activity in this area is primarily fed¬
eral, although at least six states have passed legislation to either fund
infrastructure or study the issue.3 The largest federal effort is the E-
Rate program, which expended over $2.25 billion in fiscal year 2001.
This is a modest expenditure, however, in a $3.5 trillion federal bud¬
get and less than the $2.5 billion the Navy will be spending for just
two LPD-17 amphibious ships by 2003.4The E-Rate was established
through the Telecommunications Act of 1996 (PL 104-104), and it
offers considerable discounts on telecommunications technologies to
libraries and schools (ranging from 20 percent to 90 percent). The
depth of the discount is calculated on the basis of economic need,
with some additional priority given to rural locations. In the first two
years of its operation, E-Rate benefited nearly 80,000 schools and
libraries. The lion’s share of the funds—84 percent—went to public
schools. The program has apparently targeted poor communities, as
intended. The E-Rate program, however, has limited purposes, fund¬
ing wiring, phone and Internet access, rather than computer equip¬
ment, staff support, or staff training.5
The other two main federal programs provide a broader range of
services, but to fewer beneficiaries. The Technologies Opportunities
Program (TOP) awards matching grants for model projects that
exhibit innovative use of technology. TOP addresses issues involved
in the access and skills divides, but is stretched across a variety of
objectives, including meeting the technology needs of government
agencies. CTCs provide access to technology, training, and some¬
times other services, such as job search, after-school enrichment, and
adult and continuing education.6 Programs may be located in
libraries, churches, public housing, the facilities of community-based
organizations, or a variety of places in low-income urban and rural
areas. Nearly 400 centers have been funded, but like the TOP pro¬
gram, the thrust of the CTC program is to support model projects
over the short-term rather than to provide a comprehensive or sus¬
tained solution. Together, the TOP and CTC programs totaled about
$110 million in fiscal year 2001. The Bush administration has
attempted to eliminate both programs.
Resistance from civil rights groups and congressional support have
stalled such cuts. Pressure to continue a federal role in the promotion
of technology use has come from sources closer to the administration
as well. The President’s Council of Advisors on Science and Technol¬
ogy identified the extensive diffusion of broadband technology as
potentially contributing $500 billion annually to gross domestic
product over the next ten years. In response, some technology indus¬
try representatives have called for government support of broadband
in underserved areas. Apparently such arguments have been difficult
to ignore entirely, although the White House has not proposed any
specific policy. Despite the administration’s stance on the CTC and
TOP programs, President Bush declared in August 2002 that bring-
mg the promise of broadband technology to millions of Americans”
was needed to make sure the economy grows” and to “stay on the
cutting edge of innovation.”7
Some private giving has also helped to extend technology access.

The Bill and Melinda Gates Foundation committed more than $250
million in equipment, and the foundation also trains library staff
in the latest technology. A score of other foundations have some
involvement in the issue: Annie E. Casey, Kellogg, the National
Cristina Foundation, and corporate foundations such as AOL/Time
Warner, AT&T, Microsoft Giving, BellSouth, Cisco, IBM, Marco-
Polo, and SBC. The Boys and Girls Clubs of America, the Girl Scouts
of America, and the National Urban League have digital divide ini¬
tiatives. This interest on the part of many donors and volunteer
organizations offers important resources, but efforts are fragmented
and do not necessarily replace a steady commitment of government
support.
Some increased public access is evident, as a result of federal, state,
local, and private spending. According to the American Library Asso¬
ciation, 95 percent of public libraries now offer Internet access to the
public, in comparison to 1998, when 76 percent of libraries offered
the service. Nearly half of public libraries in the U.S. provide some
training for patrons in computer use or the evaluation of information
on the web.8
Few data exist, though, on how well the supply of hardware, soft¬
ware, and assistance matches the demand from library patrons, espe¬
cially in poor communities. Public access sites may require users to
wait long periods for their turn, impose strict time limits on com¬
puter sessions, filter out useful websites deemed to have controversial
content, and provide users with little privacy. Libraries in poor urban
neighborhoods or rural areas may have only a few terminals and insuf¬
ficient numbers of staff to help patrons or to provide more systematic
training. Computer users at a library or a community center also may
not have time and leisure to explore and use the computer in the same
way that someone with an Internet connection at home would. In
many communities, access through CTCs or nonprofit programs is
sporadic or nonexistent.9
The greatest effort on the part of federal, state, and local govern¬
ments in closing the access and skills divide has been in providing com¬
puters and Internet connectivity in the schools. Schools in low-income
communities now have 5.3 students per computer in comparison to
the national average of 4.9 students per computer.10 Most schools are
now wired for the Internet, as a result of the E-Rate program.11 Still,
there are some significant holes in school access and substantial

inequality in the use of technology. Schools with a high concentra¬
tion of African American students have less computer access than
other poor schools, although that gap seems to be narrowing.12 Sim¬
ple ratios also obscure differences in the quality and use of equipment.
Anecdotal reports paint a grim portrait. Financially strapped districts
often scrape by with aging and outdated technology. Schools in low-
income areas often lack the technical support staff and teacher train¬
ing needed to integrate technology throughout the curriculum, so
students get little practice in the varied ways in which they might
apply technology and even less experience searching for or critically
evaluating information on the web. Several studies have shown that
lower-achieving students enjoy less frequent use of computers during
the school day and have more experience using computers for reme¬
dial drills rather than exploring the Internet for assignments. This
tendency may be even more pronounced in poor urban schools.
Teachers in such schools may have less experience using technology.
Computer use can also absorb precious class time for instructing stu¬
dents who lack basic computer skills, and fewer students have home
computers or Internet access to complete assignments later.13
Public Policy for the Future
Despite tangible progress toward closing technology gaps, more needs

to be done. Technology in K—12 schools is an important part of the
solution, for what happens in schools today will determine the dimen¬
sions of the access and skills divides in the future. Our data, however,
describe the experiences and attitudes of adults twenty-one years of age
and older. No matter what advances are made in improving access and
skills within public school systems, significant challenges will persist
for adults who have not acquired important technical and information
skills—and who will remain within the workforce, the community,
and the electorate for many years to come.
The primary argument against government action on this issue is
that the market is effectively closing the access gap and that public
intervention is a waste of public dollars or unwarranted distortion of
market forces. Both Adam Thierer, of the conservative think tank the
Heritage Foundation, and Benjamin Compaine, the editor of a volume
of essays on the digital divide, contend that this is the case. They point
to the rapid diffusion of the Internet and falling prices of computers
and Internet access services.14 Compaine also argues that digital tech-
nology is becoming easier to use, reducing skills barriers as well as
cost.1 ^ Thierers critique is loaded with rhetoric about big government,
competition and choice in the free market, and the possibility that gov¬
ernment subsidies will entail more government regulation of the tech-
nology industry. Although Compaine shares many of Thierers
concerns—for example, that subsidies may lock in less efficient and
more expensive technologies—he concedes that government has facil¬
itated the diffusion of some technological advances in the past and that
there may be some role to be played by government in this case. Cross¬
subsidies for universal access assisted in the diffusion of the telephone,
just as federal highway support subsidized the use of automobiles.
Compaine argues that the proper role of government is less than clear
in the case of information technology, though, and that it is advisable
to wait to find out how necessary computers will be in the future and
whether the market will close the gap on its own.16
We disagree with Thierer and Compaine on several fronts. First,
the preceding chapters demonstrate that information technology use
has current and potential benefits for society and that technology
access and skills are public goods that have spillover benefits beyond
the individuals concerned. Disparities in information technology
merit not just continued, but greater, policy attention. The social
costs of ignoring technology disparities are two-fold: failure to fully
realize the potential offered by technology and deepening inequali¬
ties in economic opportunity and democratic citizenship.
Productivity gains during the late 1990s resulted from the use of
technology in a broad range of industries and uses. Widespread tech¬
nical skills and information literacy are important resources for
increasing efficiency and innovation throughout the economy. Geo¬
graphic inequality in the distribution of technology skills may also
restrict the ability of communities to compete for business invest¬
ment, particularly for industries requiring a skilled workforce.
For individuals, deficient technology skills and limited education
constrain the possibilities for economic mobility. Continued techno¬
logical diffusion and change portend a future where even more jobs
will depend upon information technology use. Although jobs requir¬
ing few or no computer skills will continue to be available, many of
them will offer low wages and few benefits. Americans clearly recog¬
nize this new reality in the labor market, as our survey shows, and
women, racial and ethnic minorities, and the unemployed are even
more likely than others to believe that a level playing field requires
computer skills.
Prevalent skills and access are also needed if society is to capitalize
on technology’s potential to make political information and oppor¬
tunities for participation more accessible. Currently, the use of tech¬
nology is most developed as a means of communicating political
information rather than as a mechanism for direct participation
through voting or community forums. The proliferation of political
and government websites, however, represents an important step
toward enhancing civic engagement by increasing information access.
Furthermore, information technology may yet provide a means of
slowing or reversing declining levels of political participation, by
making acts such as voting or commenting on community issues eas¬
ier for many. The interest of the young in technology may have a gen¬
erational effect, raising levels of participation in the long term. These
are important potential benefits for society.
At the same time, both attitudes toward online participation and
evidence regarding access and skills show that the hurdles for politi¬
cal participation may be increased for the less educated and the poor.
Because African Americans and Latinos disproportionately fill the
ranks of the poor and less educated, the overall effect may well be to
diminish the political participation of minorities as well. The princi¬
ple of political equality demands that we must not leave whole seg¬
ments of the population unable to participate as the use of technology
evolves.
Second, both Thierer and Compaine wrongly focus the question
on access alone. Inexpensive computers, discount Internet providers,
WebTV, and the used computer market may fill some of the current
gap in access, but our data demonstrate that affordability is not the
only problem. Compaine s contention that skills requirements are
negligible is also clearly incorrect. Demands for technical competence
are indeed lower than in past decades—PC users no longer need to
know programming languages such as Fortran. The Internet, how¬
ever, has emphasized the need for literacy and information skills as
well as technical competence. One study of CTCs, for example, doc¬
umented the problems that individuals with low literacy skills had in
understanding content on the Internet.17 The market is not likely to

resolve this problem, which affects economic opportunities and polit¬
ical participation as well. There is an important public role in sup¬
porting the acquisition of information literacy, as well as technical
competence, in order to capture the social benefits of technology and
promote equal chances for political and economic participation.
What more, then, needs to be done?
On the basis of our survey and our review of the social signifi¬
cance of information technology, we put forward a few general
recommendations.
1. Attention to skills development in public access sites. One contri¬
bution that we hope to make through this study is to focus more atten¬
tion on the skills needed for technical competence and information
literacy. The data in our survey demonstrate that more than one-fifth
of the population lacks even the most basic technical skills (the ability
to use a mouse and keyboard) and that a third of the population needs
help finding information using the Internet and data bases. Public pol¬
icy should not only encourage public access, but also provide more sys¬
tematic training and assistance in public places such as libraries,
community centers, and public schools after hours. Printed manuals
and online assistance are less accessible to inexperienced users and less-
educated individuals, and our survey showed that hands-on personal
assistance and group instruction were the most widely preferred meth¬
ods of help. We found that respondents who already had a computer
or Internet access at home were as likely to use public access at libraries
as individuals without home access. This indicates that public access
sites are more than places to boot up and log on—that the availability
of assistance attracts users as well. This is an important role that can
likely be improved upon in the future.
CTCs and public institutions such as libraries and schools (partic¬
ularly adult education programs) can provide assistance with distance
education and locating information on jobs or government services
as well as providing technical support and general assistance in navi¬
gating the web. Public and nonprofit institutions have a role to play
in exploiting the “public goods” nature of the technology and the
advantages that accrue to society in terms of workforce development
and community participation.
For these reasons, we see public access as an important element in
any strategy to address the access and skills divides. Our data, how-
ever, leave us with nagging questions about the current effectiveness

of public access. Low rates of use are problematic. Only 13 percent
of those without a home computer, and only 9 percent without home
Internet access, use computers or the Internet at a library. Those with¬
out computers and the Internet are not more likely to use public
access than those who do have home access. This is true, even though
most respondents say in the abstract that they are willing to use pub¬
lic access and that they overwhelmingly favor schools and libraries as
access venues. These are currently the most common places for pub¬
lic use. Poor and minority respondents are more likely than others to
have positive images of libraries as community gathering places, indi¬
cating that the setting itself is not an issue. More needs to be known
about the actual facilities and services available in public access
venues, especially in poor communities.
There are also inherent limitations to public access as the sole
venue for computer use. The ideal situation is broader home access
to computers and the Internet, as well as continued public access for
the development of skills, information about Internet uses and soft¬
ware, and increased availability of more expensive options such as
broadband.
2. Limited experimentation with an educational technology subsidy.
Most government and nonprofit programs currently support public
access rather than increased home computer and Internet access.
Some nonprofit programs offer low-income participants special dis¬
counts, loans, or refurbished equipment, but the scale of these initia¬
tives is small in comparison to public access programs. Should public
dollars provide subsidies for low-income families, such as vouchers?
Some proponents have raised this idea—for example, a group of
Silicon Valley executives, including the chief executive of Novell, Eric
Schmidt. The proposal fielded by this group, called ClickStart, fea¬
tured a federal voucher that would cover both computer hardware
and Internet access. Eligible individuals would pay $5 dollars a
month and get a $ 10 monthly voucher that would be honored by par¬
ticipating vendors.18 Some other proposals have been put forward,
for example, for a flat $300 federal voucher.19
Survey data: Opinions on vouchers. We asked our survey respondents
for their opinion on vouchers. Our survey included questions on sup¬
port for government programs to provide computers to low-income
families and government programs to provide Internet access to low-

income families and rural residents.20 When the question was posed
in this general way, a slight majority of respondents was in favor of
such government intervention. About 56 percent of respondents sup¬
ported government funding for computers for low-income house¬
holds, and 52 percent supported a government program to subsidize
Internet access. This indicates a fair amount of support for a policy
such as vouchers, although the specific details—the size of the sub¬
sidy and the definition of “low-income”—would likely affect public
opinion regarding a concrete proposal.
We were interested in knowing who was most likely to support
funding such as vouchers, so we analyzed the survey responses using
multivariate regression. The same explanatory variables used in the
other chapters are included here—income, age, education, race, eth¬
nicity, and gender—as well as partisanship and voting in the 2000
election. Controlling for other factors, the individuals most likely to
support vouchers to provide computers and Internet access are the
young, Democrats, Latinos, African Americans, and the poor. These
results are not surprising. Except for younger respondents, those most
likely to support government programs to provide computers and
Internet access are the least likely to have access to this new technol¬
ogy. Such groups therefore express interest in computers and the
Internet as a policy issue. They are also groups that traditionally sup¬
port a more active role for government in tackling social issues.
Vouchers enjoy the support of a little over half the population, with
disadvantaged groups most likely to favor subsidies. Yet we foresee
some potential problems with vouchers as public policy, as well as
benefits, and we recommend a cautious approach.
Our recommendation. We propose a limited experiment with what
we call an “educational technology subsidy,” or a voucher that is linked
to the condition that the recipient is enrolled in some type of further
education. Our data lead us to take the position that home access is
desirable because of its flexibility and convenience. Our survey shows
that interest in many activities, particularly distance learning, decreases
when use in a public place is specified. Home Internet and e-mail access
are associated with greater levels of digital experience and may play
an enabling role for skills development, economic opportunity, and
political participation. Home access is associated with more intensive
use (and presumably more familiarity with the technology). Ample

opportunities for exploration and practice are needed.
For these reasons, we favor experimentation with voucher programs.
We hesitate to support handing out vouchers, however, without some
provisions for skills development. Our research indicates that issues
such as information literacy and even basic literacy limit the use of
information technology. Connecting the voucher to educational
advancement would also emphasize the public goods potential of infor¬
mation technology, addressing the educational issues that surfaced in
both the economic opportunity divide and the democratic divide.
We are in favor of small-scale experimentation only, to test the
effectiveness of a voucher program. Despite its limitations, public
access may still benefit a larger number of individuals, foster more
skills development, and couple technology assistance with informa¬
tion about jobs, education, public services, and community events.
The market is also more likely to provide some cheaper solutions for
access at home than to solve problems of skills development and lit¬
eracy. Concern for the latter leads to our final recommendation.
3. Equal educational opportunity and public investment in lifelong
learning. Our data explored four dimensions of the information tech¬
nology divide rather than education, but educational issues cropped up
at every turn. The ability to use information technology, and to learn
and adapt in a changing world, rests upon a general educational foun¬
dation. Basic literacy and the ability to locate and evaluate information
are academic skills. Education cut across all four divides as the most sig¬
nificant factor other than age, and it is clearly implicated in disparities
in economic opportunity and political participation.
Technology promises to magnify some of these inequalities, but it
has not created them. As Jennifer Light has argued, Americans have
a tendency to seek technological fixes for complex social problems.21
There are certainly technological aspects at stake, but it is disingenu¬
ous to reduce disparities across the four divides to simple questions
of access or a few hours of computer training. The underlying prob¬
lems are two-fold.
First, the quality of American education varies greatly, and poor
communities are disadvantaged not only in a relative way, but also,
perhaps absolutely, in terms of their ability to prepare students for
economic survival and democratic citizenship. Proposals for reform¬
ing urban education are rampant, ranging from the equalization of
school funding, broad community involvement, business involve¬

ment, organizational restructuring, greater accountability through
mayoral control, school empowerment, magnet schools, charter
schools, vouchers, and proficiency testing. Urban schools have been
the focus of a larger educational debate about the quality of public
schools, and other low-income communities face some of the same
challenges as urban school systems. We hesitate to step into this quag¬
mire of debate and to recommend any specific proposal for improv¬
ing education, particularly because our research did not directly
involve educational issues. Nevertheless, we cannot ignore the fact
that our four divides are manifestations of a larger problem of edu¬
cational opportunity. Society will not truly close any of these gaps
without addressing educational inequities.
Second, adults now and in the future will need the ability to secure
further education, either to compensate for inadequate preparation
in the past or to keep pace with evolving requirements. Rapid tech¬
nological change, shifts in the labor market, rising skills demands,
and instability in employment have made lifelong learning an issue
in the economic sphere. Education has always been important for
democratic participation, as Thomas Jefferson declared. But the
involvement of government in complex and technical issues today far
outstrips anything Jefferson could have imagined. Informed partici¬
pation has become more demanding, because of the growth of ini¬
tiatives and referenda in the political process, the decline of
partisanship, candidate-centered politics, and the proliferation of
new sources of political information through the Internet. Lifelong
learning may have some advantages in the political sphere as well, if
it imparts general skills such as literacy, the ability to analyze and eval¬
uate information, and critical thinking.
Lifelong learning, of course, is a catchall phrase. Depending upon
an individual’s needs, it may embrace anything from remedial read¬
ing classes at a community center to doctoral study. Postsecondary
institutions have experienced a surge of nontraditional students
returning to the classroom over the past few decades, so trends toward
lifelong learning are well underway already. For individuals with lim¬
ited financial means, however, access to continuing education may be
difficult, particularly for vocational or postsecondary training, where
tuition is a consideration. American employers offer relatively little
job training in comparison with their European counterparts, plac-
ing more of the burden on individual workers. Tuition costs have

been escalating, as many states seek to balance their beleaguered bud¬
gets by cutting subsidies for higher education. Public investment, at
all levels of government, can increase the quality and the availability
of lifelong learning by supporting adult education and postsecondary
institutions. Issues of affordability can be addressed in many ways—
for example, through scholarships, loans, grants, tax credits, current
funding for job training programs, and government subsidies that
hold down general tuition costs. Different solutions are likely needed
for different types of continuing education. By investing in lifelong
learning, however, we increase the capacity of society to respond to
continued change in technology, the economy, and political needs.
The comprehensive high school was an American innovation that
responded to the needs of industrialization and urbanization by mak¬
ing education beyond the “3 Rs” available on a widespread basis dur¬
ing the twentieth century.22 The task for the twenty-first century may
well be to insure that all have access to that basic foundation, even on
a “second-chance” basis, and that a large number can choose further
education according to their interests and qualifications, whether it
is specialized vocational training, continuing studies, or postsec¬
ondary degrees.
The role of technology should not be viewed in isolation from
other challenges to full participation in society. As computers, data¬
bases, and the Internet have transformed processes of production and
the dissemination of information, they have replicated—and, in
some cases, exacerbated—long-standing inequalities. Computers and
Internet access will not remedy problems of racism, segregation,
unequal education, unequal political participation, and economic
inequality, but they represent one dimension of the problem of pro¬
viding equal opportunity in a democratic society. Public policy that
promotes access, skills, and the empowering potential of technology
should represent one dimension of the solution as well.
Notes
1. Ordinary least squares multivariate regression is used with the digital

literacy index as the dependent variable. The independent variables are
income, education, age, gender, ethnicity, race, employment status, home
computer ownership, e-mail address, and home Internet access. See table
A6.1.
2. Trotter 2001.
3. National Conference of State Legislatures 2001.
4. See description of Department of Defense budget at the Office of
Management and Budget website [www.whitehouse.gov/omb/budget/
fy2003/budget/html].
5. Carvin, Conte, and Gilbert 2001, 223; Puma, Chaplin, and Pape
2000. For the distribution of E-Rate dollars, economic need is measured by
the percentage of students served by the school or library who participate in
the federal school lunch program. For more information about the E-Rate
and an initial program evaluation, see especially Puma, Chaplin, and Pape
2000.
6. U.S. Department of Education 2002.
7. Flolsendolph 2002; Phillips 2002.
8. Trotter 2001.
9. Trotter 2001.
10. Education Week on the Web 2001.
11. Jesdanun 2002, B1, 4.
12. Hess and Leal 2001.
13. Bushweller 2001; Manzo 2001.
14. Both reference “Moores Law,” developed by Gordon Moore, one of
the founders of Intel. Moore predicted that the power of microprocessors
would double roughly every eighteen months, creating more powerful com¬
puters at a lower cost. Moore’s predictions so far have been largely correct.
SeeThierer 2000 and Compaine 2001, 320.
15. Compaine 2001, xv.
16. Ibid., 115-16.
17. Penuel and Kim 2000.
18. Lacey 2000.
19. Thierer 2000.
20. Both variables are coded as dummy variables, where 1 indicates sup¬
port for vouchers and 0 otherwise. Because the dependent variables are
binary, logistic regression is used. See table A6.2.
21. Light 2001.
22. Goldin 2001.
Appendix I
Multivariate Regression Tables
Table A2.2
Respondent Has Internet Access
1996 1998 2000
Variable P (se) P> P (se) P> P (se) P>

M M M
Political factors:
Strong Democrat .21 (.19) .24 .09 (.21) .68 -.42 (.18) .02
Strong Republican -.16 (.21) .45 .58 (.24) .02 -.01 (.22) .95
Pure Independent .01 (.25) .96 .02 (.24) .95 -.46 (.22) .03
Social factors:
Log age -1.0 (.18) .00 -1.95 (.20) .00 -2.06 (.20) .00
Female .14 (.13) .30 -.29 (.14) .04 .11 (.15) .45
Latino .08 (.23) .74 -.35 (.24) .13 -.29 (.27) .27
African American -.04 (.22) .87 -.64 (.24) .01 -.97 (.22) .00
Asian American .32 (.56) .57 -.33 (.61) .58 .10 (.62) .87
Education .02 (.04) .58 .52 (.05) .00 .54 (.05) .00
Square income .01 (.00) .00 .01 (.00) .00 -.01 (.00) .08
Poor -.15 (.05) .00 -.14 (.04) .00 .25 (.06) .00
Constant 2.07 (.77) .01 4.66 (.73) .00 5.31 (.76) .00
Pseudo-/?2 .1159 .2631 .2536

Log-reduced %2 (11) 184.59 .00 431.45 .00 451.38 .00
N 1,380 1,203 1,346
Source-. American National Election Survey, postelection study, 1996, 1998, and 2000.
Notes: Unstandardized logistic regression coefficients are given. Standard errors are in parenthe¬
ses. Reported probabilities are based on two-tailed test. Statistically significant coefficients at a
confidence interval greater than 90 percent are in bold. For 1996, African Americans constituted
11.9 percent of those surveyed, Asian Americans 1.4 percent, and Latinos 8.7 percent. For 1998,
African Americans constituted 11.9 percent, Asian Americans 1.2 percent, and Latinos 10.7 per¬
cent. For 2000, African Americans constituted 11.6 percent, Asian Americans 1.8 percent, and
Latinos 7.6 percent.
142 Multivariate Regression Tables
Table A2.3
The Access Divide: Respondent Has Access
to Information Technology
Internet Access
Variable at Home E-Mail Address Home Computer'
P (se) ?■> P (se) P >lzl P (se) P >lzl

lZl
Poor -.97 (.13) .00 -.89 (.14) .00 -1.07 (.13) .00
Education .36 (.05) .00 .55 (.06) .00 .37 (.05) .00
Age -.02 (.00) .00 -.03 (.00) .00 -.02 (.00) .00
Male .19 (.12) .12 .26 (.13) .04 .06 (.13) .61
Democrat -.13 (.16) .00 -.01 (.17) .94 -.15 (.16) .35
Republican .27 (.17) .12 .30 (.18) .09 .37 (.18) .03
Latino -.52 (.22) .01 -.75 (.23) .00 -.79 (.22) .00
African -.69 (.17) .00 -.66 (.17) .00 -.40 (.17) .01
American
Asian .06 (.82) .93 .32 (.82) .68
American
Constant .82 (.29) .00 .68 (.30) .02 1.02 (.30) .00
N 1,309 1,311 1,319
Log-reduced 277.16 .00 351.63 .00 275.6 .00
X2 (12)
Pseudo R2 .1539 .1981 .1570
Source-. Tolbert, Stansbury, and Mossberger 2001; N = 1,837.
Notes: Unstandardized logistic regression coefficients. Standard errors are in parentheses.

Probabilities are based on two-tailed test. Statistically significant coefficients at a confi¬
dence interval greater than 90 percent are in bold. Survey questions were as follows:
“Do you personally have a home computer?” “Do you have an e-mail address with
which you can send or receive e-mail?” “Do you have access to the Internet from
home?” Responses were coded 1 for agree and 0 for disagree.
Table A3.1
Technical Competence and Information Literacy
Need More Need More

Variable Information Literacy Technical Competence
(3 (se) p >lzl P (se) P >|z|

Latino .49 (.20) .01 .39 (.18) .03
African American .71 (.14) .00 .50 (.13) .00
Asian American .43 (.54) .42 .54 (.48) .26
Poor .57 (.12) .00 .55 (.11) .00
Education -.51 (.05) .00 -.53 (.05) .00
Male -.38 (.12) .11 -.17 (.10) .00
Age .05 (.00) .00 .05 (.00) .00
Frequency library -.23 (.06) .00 -.18 (.06) .00
patronage
N 1,545 1,379
Log-reduced y2 (10) 547.92 .00 499.13 .00
Pseudo-/?2 .1390 .1521
Source-. Tolbert, Stansbury, and Mossberger 2001; N= 1,837.

Notes: Unstandardized ordered logistic regression coefficients. Standard errors in paren¬
theses. Probabilities are based on two-tailed test. Statistically significant coefficients at
more then a 90 percent confidence interval in bold. To assess computer literacy, we cre¬
ated an index of responses to the following three questions: Do you need computer
assistance (1) locating information on the web, (2) doing homework, or (3) findings
books in a library? Variable range is 0-3. To assess the need of respondents for addi¬
tional technological competence, we created an index of responses to the following
three questions: Do you need computer assistance (1) using a mouse or typing, (2)
using e-mail, or (3) using word processing or spreadsheet programs. Variable range is
0-3.
Table A3.2
Computer Skills
Need
Variable Computer Skills
P (se) p >|z|
Poor .55 (.11) .00
Education -.53 (.05) .00
Age .05 (.00) .00
Male -.38 (.11) .00
Latino .31 (.18) .08
African American .52 (.13) .00
Asian American -.02 (.48) .97
Frequency library patronage -.02 (.06) .00
N 1,311
Log-reduced X2 (10) 545.93 .00
Pseudo R2 .1112
Source: Tolbert, Stansbury, and Mossberger 2001; N= 1,837.
Notes: Unstandardized ordered logistic regression coefficients. Standard errors in paren¬

theses. Probabilities are based on two-tailed test. Statistically significant coefficients at a
confidence interval greater than 90 percent are in bold. This table was used to generate
the probabilities reported in chapter 3. To assess the need of respondents for computer
skills, we created an index of responses to the following six questions: Do you need
computer assistance (1) using a mouse and typing, (2) using e-mail, (3) locating infor¬
mation on the web, (4) using word processing/spreadsheet programs, (5) doing home¬
work, or (6) finding books in a library? Variable range is 0-6.
Table A3.3
Attitudes about Library Use
Consider the Library Support Public

a Community Access for
Variable Meeting Place Computers and Internet
P (se) P>\A P (se) P> N

Poor .29 (.12) .01 -.21 (.11) .04
Education .01 (.05) .81 .10 (.04) .03
Age .00 (.00) .15 -.00 (.00) .69
Male -.21 (.11) .05 -.09 (.10) .35
Latino .55 (.19) .00 -.26 (.17) .12
Asian American -.25 (.44) .56 -1.12 (.36) .00
Frequency library .22 (.06) .00
patronage
1V 1,549 1,399
Log-reduced %2 (10) 56.09 .00 53.52 .00
Pseudo-i?2 .0261 .0111
Source: Tolbert, Stansbury, and Mossberger 2001; N = 1,837.

Notes: Model 1 unstandardized logistic regression coefficients, standard errors in paren¬
theses; probabilities are based on two-tailed test. Model 2 unstandardized ordered logis¬
tic coefficients, standard errors in parentheses; probabilities are based on two-tailed test.
Statistically significant coefficients at a confidence interval greater than 90 percent are
in bold. To assess respondents’ support for public access to computers and the Internet,
we created an index of responses to six questions. Would you use a computer or access
the Internet at a (1) recreation center, (2) senior center, (3) local church, (4) govern¬
ment office, (5) public library, or (6) public school after hours? Variable range is 0-6.
Table A3.4
Opposed to Opposed to Opposed to Opposed to

One-on-One Group Online Printed
Variable Instruction Instruction Instruction Manuals
P (se) p >|z| P (se) p >|z| P (se) pAA P (se) pAA

Poor .17 .13 .15 .16 .19 .08 -.04 .73
(.11) (.11) (.11) (.11)
Education -.08 .07 -.09 .05 -.12 .01 .05 .30
(.05) (.05) (.05) (.05)
Age .01 .04 .01 .00 .01 .00 .01 .03
(.00) (.00) (.00) (.00)
Male .24 .02 .15 .15 -.27 .01 -.11 .26
(.10) (.10) (.10) (.10)
Latino .04 .80 -.30 .09 -.02 .93 -.01 .97
(.18) (.18) U7) (•17)
African .09 .49 -.44 .00 -.39 .00 -.31 .02
American (.13) (.13) (.13) (.13)
Asian .12 .76 .87 .03 .30 .44 -.62 .13
American (.40) (.40) (.39) (.40)
Frequency -.10 .07 -.06 .28 -.00 .95 .01 .84
library (.06) (.06) (.05) (.05)
patronage
N 1,563 1,561 1,545 1,558
Log-reduced 24.78 .00 43.47 .00 45.80 .00 17.37 .03
X2 (10)
Pseudo-i?2 .0074 .0118 .0113 .0044
Source: Tolbert, Stansbury, and Mossberger 2001; N= 1,837.
Notes: Unstandardized ordered logistic regression coefficients, standard errors in parenthe¬

ses; probabilities are based on two-tailed test. Statistically significant coefficients at a confi¬
dence interval greater than 90 percent are in bold. Dependent variables were measured on a
five-point ordinal scale: 1 = strongly agree, 2 = agree, 3 = neutral, 4 = disagree, and 5 =
strongly disagree.
Table A4.I
Attitudes about Computers and Economic Opportunity
Believe Using Computer Need to Learn

Necessary to Keep New Computer Skills
Variable Up with the Times for Career Advancement
P (se) p> M P (se) p> M

Poor -.03 (.13) .80 .02 (.11) .86
Education .05 (.05) .33 -.02 (.04) .73
Age -.01 (.00) .22 -.01 (.00) .01
Male -.11 (.12) .33 -.20 (.10) .04
Latino .75 (.23) .00 .15 (.17) .36
Asian American .84 (.56) .13 .44 (.41) .29
Unemployed .24 (.15) .11 .38 (.12) .00
Constant .69 (.26) .00
N 1,512 1,543
Log-reduced y2 (8) 32.93 .00 38.77 .00
Pseudo-i?2 .02 .01 —
Source: Tolbert, Stansbury, and Mossberger 2001; N = 1,837.
Notes: Coefficients reported in model 1 are unstandardized logistic regression estimates;

standard errors are in parentheses. Coefficients reported in model 2 are unstandardized
ordered logistic regression estimates; standard errors are in parentheses. Statistically sig¬
nificant coefficients at a confidence interval greater than 90 percent are in bold. Con¬
stant is not estimated in ordered logit models. Probabilities are based on two-tailed test.
We measured responses to the following questions: These days, do you think it is neces¬
sary for people to use the Internet to keep up with the times? (agree coded 1, and dis¬
agree coded 0) and Do you believe you need to learn new computer skills to get a job?
Do you believe you need to learn computer skills to get a higher paying job? Do you
believe you need to learn computer skills to get a promotion? Do you believe you need
to learn new computer skills to start a small business? Variable range is 0-4.
Table A4.2
Support Use of Internet for Employment and Education
Variable Would Search for a Job Online Would Take a Class Online
P (se) p>\z\ P (se) P >M

Latino -.06 (.23) .78 -.32 (.20) .00
Asian American .30 (.66) .64 -1.30 (.44) .00
Poor -.04 (.14) .78 -.49 (.13) .00
Education .31 (.06) .00 .24 (.06) .00
Male .39 (.13) .00 -.03 (.12) .80
Age -.04 (.00) .00 -.02 (.00) .00
Unemployed -.30 (.15) .05 -.33 (.15) .02
Constant 1.63 (.28) .00 1.07 (.27) .00
N 1,434 1,412
Log-reduced yj (8) 220.60 .00 112.56 .00
Pseudo-i?2 .13 .06 —
Source: Tolbert, Stansbury, and Mossberger 2001; N - 1,837.

Notes: Unstandardized logistic regression coefficients; standard errors are in parentheses;
probabilities are based on two-tailed test. Statistically significant coefficients at a confi¬
dence interval greater than 90 percent are in bold. We measured responses to the fol¬
lowing questions: How do you feel about searching or applying for a job online? How
do you feel about taking a class online? Positive responses were coded 1, and negative
responses were coded 0.
Table A4.3
Support Use of Internet for Employment
and Education in a Public Place
Would Search for a Job Would Take a Class

Variable Online in a Public Place Online in a Public Place
P (se) P> N P (se) p>\4

Latino .16 (.21) .45 -.05 (.19) .79
Asian American -.33 (.48) .48 -.60 (.44) .17
Poor -.06 (.13) .64 -.05 (.12) .65
Education .15(.05) .00 -.02 (.04) .56
Male .45 (.12) .00 .26 (.11) .02
Age -.03 (.00) .00 -.00 (.00) .00
Unemployed -.02 (.14) .88 -.08 (.13) .51
Constant 1.36 (.26) .00 .55 (.24) .02
N 1,417 1,418
Log-reduced yj (8) 140.85 .00 25.97 .00
Pseudo-#2 .07 .01 —
Source-. Tolbert, Stansbury, and Mossberger 2001; N = 1,837.

Notes: Unstandardized logistic regression coefficients; standard errors in parentheses;
lowing questions: How do you feel about searching or applying for a job online in a
public place? How do you feel about taking a class online in a public place? Positive
responses were coded 1, and negative responses were coded 0.
Table A4.4
Respondents’ Internet Use for Employment and Education *
Variable Searched for a Job Online Taken a Class Online
P (se) P >lzl P (se) P>\A

Latino -.03 (.22) .89 .19 (.30) .51
Asian American .57 (.50) .25 .85 (.51) .10
Poor .03 (.15) .85 -.31 (.20) .12
Education .44(.06) .00 .20 (.08) .01
Male .40 (.13) .00 .26 (.17) .12
Age -.06 (.01) .00 -.02 (.00) .00
Unemployed -.42 (.13) .02 -.66 (.26) .00
Constant .27 (.29) .35 -1.86 (.40) .00
N 1,523 1,525
Log-reduced %2 (8) 369.98 .00 81.01 .00
Pseudo-i?2 .1965 .07 —
Source-. Tolbert, Stansbury, and Mossberger 2001; N- 1,837.

lowing questions: Have you searched or applied for a job online? Have you taken a class
online? Agree was coded 1, and disagree was coded 0.
Table A5.1
Support for Digital Democracy and E-Government
Looking Up
Voter Government Town
Variable Voting Registration Information Meetings
P (se) P>\A P (se) P> M P (se) p>|z| P (se) P> N

Poor -.11 (.14) .40 -.10 (.11) .48 -.32 (.17) .05 -.30 (.14) .03
Education .33 (.05) .00 .40 (.05) .00 .42 (.07) .00 .22 (.05) .00
Age -.02 (.00) .00 -.03 (.00) .00 -.02 (.00) .00 -.01 (.00) .00
Male .14 (.12) .25 .28 (.13) .03 .16 (.16) .29 .26 (.12) .04
Democrat .46 (.16) .00 .34 (.16) .04 .41 (.20) .03 .10 (.16) .51
Republican .06 (.17) .69 .07 (.17) .67 .27 (.21) .19 .07 (.17) .65
Latino -.10 (.23) .66 -.27 (.23) .24 -.53 (.26) .04 .11 (.23) .61
African -.25 (.17) .14 -.22 (.17) .20 -.10 (.21) .61 .01 (.17) .93
American
Asian -.18 (.67) .78 -.37 (.72) .60 -1.17 (.72) .10 -.51 (.66) .43
American
Voter .32 (.16) .04 .38 (.16) .02 .62 (.19) .00 .52 (.16) .00
Constant -.41 (.30) .16 .29 (.30) .33 .73 (.36) .04 -.40 (.30) .19
N 1,167 1,227 1,210 1,118
Log-reduced 107.15 .00 162.15 .00 111.17 .00 73.39 .00
X2 (12)
Pseudo-/?2 .0663 .0995 .0935 .0474

Notes: Unstandardized logistic regression coefficients; standard errors are in parentheses; proba¬
bilities are based on two-tailed test. Statistically significant coefficients at a confidence interval
greater than 90 percent are in bold. We measured responses to the following questions: How
do you feel about voting in a government election online? How do you feel about registering to
vote online? How do you feel about looking up government information online? and How do
you feel about participating in an online town meeting? Positive responses were coded 1, and
negative responses were coded 0.
Table A5.2
Support for Digital Democracy vs. Actual Experience
Support for
Support for Online Experience with
Online Participation in Online Political
Variable Participation Public Place Participation
P (se) P >lzl P (se) P>\4 . P (se) P >lzl

Poor -.22 (.13) .09 -.36 (.12) .00 -.28 (.12) .02
Education .37 (.05) .00 .17 (.05) .00 .40 (.05) .00
Age -.02 (.00) .00 -.01 (.00) .00 -.03 (.00) .00
Male .27 (.12) .02 .20 (.11) .07 .43 (.11) .00
Democrat .35 (.15) .02 .35 (.14) .15 .09 (.14) .50
Republican .08 (.16) .59 .28 (.15) .06 -.11 (.15) .45
Latino -.30 (.22) .17 .07 (.20) .72 -.06 (.20) .75
African -.14 (.16) .36 .04 (.15) .78 -.15 (.15) .33
American
Asian -.52 (.55) .34 -.71 (.54) .19 -.79 (.56) .15
American
Participate .48 (.15) .00 .36 (.14) .01 .60 (.14) .00
N 963 1,185 1,251
Log-reduced 151.65 .00 77.13 .00 230.04 .00
X2 (10)
Pseudo-./?2 .0505 .0262 .0741
Source-. Tolbert, Stansbury, and Mossberger 2001; N- 1,837.
Notes: Unstandardized ordered logistic regression coefficients; standard errors are in

parentheses; probabilities are based on two-tailed test. Statistically significant coeffi¬
cients at a confidence interval greater than 90 percent are in bold. To measure support
for online political participation, we created an index of responses to the following four
questions: How do you feel about (1) voting in a government election online, (2) regis¬
tering to vote online, (3) looking up government information online, and (4) partici¬
pating in an online town meeting? Variable range is 0-4. To measure support for public
online political participation, we created an index of responses to the following three
questions: How do you feel about (1) voting in a government election online in a pub¬
lic place, (2) registering to vote online in a public place, and (3) looking up government
information online in a public place? Variable range is 0—3. To measure experience with
online political participation, we created an index of responses to the following three
questions: Have you (1) searched for political information online, (2) looked up infor¬
mation on government services online, and (3) seen an online political advertisement?
Variable range is 0-3.
Table A6.I
Digital Experience
Variable Digital Experience
P (se) P> M
Poor -.00 (.09) .99
Education .34 (.03) .00
Age -.04 (.00) .00
Male .40 (.08) .00
Home Computer .14 (.13) .29
E-mail address 1.38 (.12) .00
Home Internet Access .47 (.14) .00
Latino .02 (.14) .84
African American .15 (.10) .13
Asian American -.27 (.32) .40
Unemployed -.26 (.10) .01
Constant 2.30 (.19) .00
N 1,454
F 124.006 .00
Adjusted R2 .482
Notes: Unstandardized OLS regression coefficients; standard errors are in parentheses;

probabilities are based on two-tailed test. This variable measures experience using com¬
puters and the Internet for various transactions and is an index of responses to seven
questions. Has the respondent (1) located information on the web, (2) searched for
political information online, (3) looked up information on government service online,
(4) applied for or searched for a job online, (5) taken a class online, (6) used the com¬
puter to do homework, or (7) used the computer to find books in the library without
assistance? Variable range is 0-7.
Table A6.2
Support for Vouchers *
Support Vouchers to Support Vouchers to

Low-Income Persons Low-Income Persons
Variable for Computers for Internet Access
P (se) P >lzl P (se) P>\A

Poor .39 (.16) .01 .31 (.15) .04
Education -.09 (.06) .16 -.01'(.06) .83
Age -.01 (.00) .00 -.00 (.00) .07
Male -.16 (.14) .23 .09 (.13) .46
Democrat .56(.18) .00 .57 (.17) .00
Republican -.64 (.18) .00 -.58 (.17) .00
Latino .79 (.28) .01 .84 (.27) .00
Asian American .22 (.86) .79 1.26 (1.11) .25
Participate -.10 (.18) .56 -.21 (.17) .21
Constant 1.53 (.34) .00 .61 (.31) .05
N 1,075 1,095
Log-reduced %2 (10) 161.38 .00 121.07 .00
Pseudo- R2 .1181 .0833

dence interval greater than 90 percent are in bold. To measure support for a voucher
system to assist low-income persons to purchase computers, we created an index based
on responses to the question, How do you feel about a government program to provide
computers for the low income? To measure support for a voucher system to assist low-
income persons gain access to the Internet, we created an index based on responses to
the question. How do you feel about a government program to fund Internet access for
low income or rural residents? For both measures, positive responses are coded 1, and
negative responses are coded 0.
Appendix I
Survey Questionnaire
Question first
[dial telephone number and press any key to continue]
[use FI THROUGHOUT THE SURVEY TO MAKE NOTES CONCERN¬

ING THE INTERVIEW]
[par] Hi, this is [name] from Kent State. We started an interview

about computers and Internet use. We would really appreciate it if
you could finish the interview now. Do you have time? [press any
key or quit]
[ref] Hello, this is [name], a student calling from the Department

of Sociology at Kent State University. Earlier this (wk/mo) you or
someone in your household declined to be interviewed in a survey
concerning computer and Internet use. We really would like your
opinion for our study. Do you have time to answer a few questions?
[press any key or quit]
Question introduction
Hi, my name is [interviewer name] and I’m a student at Kent

State University in Kent, Ohio. I’m helping researchers at Kent State
investigate issues about computers and Internet use. The interview
only takes about 8 minutes. Will you help us?
1. CONTINUE
2. REFUSAL
156 Survey Questionnaire
3. CALLBACK
4. BUSINESS
*
5. COMMUNICATION BARRIERS
6. PHONE ISSUES
7. NO ADULT AT HOME
8. ALREADY INTERVIEWED
Question phone
Enter the correct phone disposition
1. NO ANSWER
2. BUSY
3. ANSWERING MACHINE (iS DEFINITELY A HOUSEHOLD)
4. ANSWERING MACHINE (UNSURE WHETHER IT IS A HOUSE¬
HOLD OR BUSINESS)
5. fax/data LINE
6. pager/cell phone
7. disconnected/nonworking number
8. NUMBER CHANGED
9. PRIVACY MANAGER
10. CALL FORWARDING
11. TECHNICAL PHONE ISSUES
Question commbarr
Enter the correct communication disposition
1. LANGUAGE (CANNOT UNDERSTAND PERSON ON PHONE)
2. PHYSICALLY OR MENTALLY UNABLE/lNCOMPETENT
3. MISCELLANEOUS
Question callback
Is there a better day or time I could call back?
1. QUALIFIED RESPONDENT NOT IDENTIFIED
2. QUALIFIED RESPONDENT IDENTIFIED

Question refusal
1. HANG-UP AT THE INTRODUCTION SCREEN

2. REFUSAL AT THE INTRODUCTION SCREEN
**used persuaders/qualified RESPONDENT NOT IDENTIFIED
Question consent
Drs. Tolbert, Stansbury, and Mossberger at Kent State University are

conducting a study about computers and Internet use. This interview
is completely voluntary, and you may terminate the interview at any
time. All information that you share with us is strictly anonymous. This
means that there will be no way to associate you with the information
you give. Also, if you do not want to answer any particular question
just tell me and I’ll skip to the next one. This study has been approved
by Kent State University. If you have further questions that you would
like addressed before you participate in the study, you may call Dr. Tol¬
bert at 330-672-2060, or they can call you back at your convenience.
[press any key to continue]
Question screen
Are you 21 years or older?
1. Yes
2. No
3. Don’t know/not applicable
Question qualification
Can I speak to someone in the household who is 21 years or older?
1. Yes
2. No
Question Q1
Do you personally have a home computer?
1. Yes
2. No
Question Q2
In the last month, how often did you personally use a computer
at. . .
work
home
school
a library or community center
a friend’s or relative’s house
... for any reason.
[read through answer choices]
1. 0 times
2. 1—10 times
3. 11—30 times
4. 31-100 times
5. More than 100 times
Question Q3
Do you have an e-mail address through which you can send or receive
e-mail by computer?
1. Yes
2. No
Question Q4
In the last month, how often did you access the Internet from . . .
work
home
school
a library or community center
a friend’s or relative’s house
... for any reason.
[read through answer choices]
1. 0 times
2. 1—10 times
3. 11—30 times
4. 31—100 times
Question Q5
How do you access the Internet or an online service from home?
1. WebTV
2. Telephone line and modem
3. DSL
4. Broadband or cable modem
5. None
Question Q6
These days, do you think it is necessary for people to use the Inter¬
net if they want to keep up with the times?
[yes—necessary; no—not necessary]
1. Yes
2. No
i 60 Survey Questionnaire
Question Q7
Have you ever been turned down for a job because you needed to
know more about computers?
1. Yes
2. No
Question Q8
Have your computer skills ever helped you get a job or a promotion?
[yes—helped; no—haven’t helped]
1. Yes
2. No
Question Q9
Do you believe you need to learn new computer skills to get a . . .

(Name all that apply)
1. Job
2. Promotion
3. Higher-paying job
4. Start a small business
5. None
Question Q10
I am going to read you a list of activities people can do online. For

each activity, please tell me if you have used the Internet in this way?
Searched for political information

Looked for information on government services or contact a govern¬
ment official
Seen a political campaign ad

Searched or applied for a job
Taken a class
[yes-USED INTERNET; NO-DID NOT USE INTERNET]
1. Yes
2. No
Question Ql 1
How do you feel about the following statements?

(Answer agree, neutral, or disagree)
I would . . .
vote in a government election online

register to vote online
look up government information online
search or apply for a job online
take a class online
participate in an online town meeting
1. Agree
2. Neutral
3. Disagree
4. Don’t know/don’t care/other
Question Q12
Would you use a computer located in a public place to ... ?

(Read the following response categories)
1. Vote in an election
2. Register to vote
3. Look up information about government services
4. Search or apply for a job
5. Take a class online
Question Q13
I am going to read you a list of possible places to use computers and

access the Internet. For each location, please tell me if you would be
willing to go there?
(Read the following response categories)
Recreation center
Senior center
Local church
Government office
Library
Public schools after hours
Don’t know/not applicable/other
Question Ql4
Please tell me if you would need assistance doing the following com¬
puter task:
Using the mouse and typing

Using e-mail
Locating information on the Internet
Using word processing/spreadsheet programs
Taking a class online
Doing homework
Finding books
1. Yes
2. No
Question Ql 3
If you need to learn a new computer or Internet skill, how would you
prefer to be taught that skill?
Personal instruction
Take a class
Use online help or tutorials
Use printed manuals
1. Agree
2. Neutral
3. Disagree
Question Q16
In the last month, how frequently have you visited a library for any
reason?
1. 0 times
2. 1-3 times
3. 6—10 times
4. 11—15 times
Question Q17
Do you think of the library as a community gathering place?
1. Yes
2. No
Question Q18
There should be a government program to partially fund the cost of

a home computer for low-income individuals.
1. Agree
2. Neutral
3. Disagree
4. Don’t know/don’t care/other v
Question Q19
There should be a government program to partially fund the cost of

Internet access for low-income individuals or those living in rural
areas.
1. Agree
2. Neutral
3. Disagree
Question Q20
Some people are registered to vote and others are not. Are you regis¬
tered to vote in the precinct or election district in which you now live?
1. Yes
2. No
Question Q21
Did you vote in the 2000 presidential elections?
1. Yes
2. No
Question Q22
Generally speaking, do you usually consider yourself a . . .
1. Republican
2. Democrat
3. Independent
4. Don’t know/not appiicable/other
Question Q23
How would you describe your views on most political matters? Gen¬
erally do you think of yourself as . . .
1. Liberal
2. Moderate
3. Conservative
4. Don’t know/not applicable/other
Question Q24
How many people live in your household?

1. Actual number of people
2. Refused
Question Q25
What was the last grade in school you completed?

1. Not a high school graduate
2. High school graduate
3. Some college
4. College graduate
3. Postgraduate work or degree (Master’s, M.B.A., J.D., M.D., and
Ph.D.)
Question Q26
How old are you?

1. Actual age between 18 and 98
2. Refused
Question Q27
Are you . . .
1. White or Caucasian
2. Black or African American
3. Asian
4. Other (specify)
5. Refused
Question Q28
Are you of Latino origin or descent?

1. Yes
2. No
Question Q29
Was your total family income in 2000 over $30,000?

1. Under
2. Over
3. Refused
Question Q30 [income]
Would you say . . .

$25,000-$30,000
$20,000-$25,000
$15,000-$20,000
$10,000-$13,000
$5,000-$10,000
Under $5,000
Refused
Don’t know/not applicable
Question Q31
At this time do you receive any form of public assistance, such as food
stamps, section 8 housing vouchers, TANF, Medicaid, SSI disability?
1. Yes
2. No
Question Q32 [income]
Would you say . . .

$30,000-$50,000
$30,000- $73,000
$75,000-$100,000
$100,000-$ 125,000
Over $125,000
Refused
Don’t know/not applicable
Question Q33
What is your current employment status?
1. Currently employed
2. Temporarily out of work
3. Not in the market for work
4. Full-time student
5. Retired
Question Q34
This concludes the interview. Thank you for your time.

[do not ask this question, determine gender]
1. Male
2. Female
Question thank
Thank you for your time.
[press any key to continue]

'
Glossary
ANES American National Election Survey

CATI Computer-Assisted Telephone Interviewing laboratory at
Kent State University that conducted the telephone survey used
in this study
CPS Current Population Survey, U.S. Census Bureau
CTC Community technology center, a federal program to assist
communities with technology needs, such as hardware, soft¬
ware, content development, and training
digital divide Disparities in information technology based on
demographic factors such as race, ethnicity, income, education,
and gender
distance learning Education or training delivered through use of
the Internet or other technology, such as the television
DSL form of high-speed Internet access called a “digital subscriber
line”
e-commerce Commercial business transactions, such as purchases,
conducted over the Internet
e-democracy Civic engagement using e-mail, online discussion
lists or chat rooms, and websites for political mobilization and
participation
e-government The delivery of information and services online via
the Internet or other digital means
E-Rate program Established through the Telecommunications Act
of 1996, it offers considerable discounts on telecommunications
technologies to libraries and schools (ranging from 20 percent
to 90 percent). The depth of the discount is calculated on the
basis of economic need, with some additional priority given to
rural locations. Funded by taxes on local and long distance tele¬
phone carriers, provides schools and libraries in poor commu¬
nities with discounted rates for Internet access, high-speed data
connections, phone service, and wiring
170 Glossary
GED General education development credential

NALS National Adult Literacy Survey *
NTIA National Telecommunications and Information
Administration
PC Personal computer
SCANS Secretary’s Commission on Achieving Necessary Skills
TOP Technologies Opportunities Program is a federal program to
assist communities with technology needs, .such as hardware,
software, content development, and training
References
Abramson, P. R. 1983. Political attitudes in America. San Francisco: Freeman.

ActionForum. 2000. Berkeley general plan ActionForum. Available [on¬
line]: www.actionforum.com/forum/index.htmEforum_idMO [30
March 2003].
Alvarez. R. M., and J. Nagler. 2002. The likely consequences of Internet vot¬
ing for political representation. Loyola of Los Angeles Law Review 34 (3):
1115-53.
American Association of School Librarians (with the Association for Edu¬
cational Communications and Technology). 1998. Information power:
Building partnerships for student learning. Chicago: American Library
Association.
American Library Association. 1989. Presidential committee on information
literacy. Available [online]: www.infolit.org/documents/89Report.htm
[23 May 2002],
American National Election Survey. 1996. Available [online]: www.icpsr.
umich.edu [10 January 2002],
-. 1998. Available [online]: www.icpsr.umich.edu [10 January 2002],
-. 2000. Available [online]: www.icpsr.umich.edu [10 January 2002],
Appelbaum, E., and P. Albin. 1998. Computer rationalization and the trans¬
formation of work: Lessons from the insurance industry. In The chang¬
ing nature of work, edited by F. Ackerman, N. R. Goodwin, L.
Dougherty, and K. Gallagher, 153-56. Washington, D.C.: Island Press.
Association of College and Research Libraries. 2000. Information literacy
competency standards for higher education. Available [online]:
www.ala.org/acrl/ ilcomstan.html [30 March 2002].
Atkinson, R. D., and R. H. Court. 1998. The new economy index: Under¬
standing America’s economic transformation and new economy project.
Available [online]: www.neweconomyindex.org [8 March 2002].
Baldassare, M. 2000. California in the new millennium: The changing social
and political landscape. Berkeley: University of California Press.
Barber, B. 1984. Strong democracy. Berkeley: University of California Press.
Barrington, L. 2000. Does a rising tide lift all boats? America’s full-time work¬
ing poor reap limited gains in the new economy. Conference Board
Report 1271-00-RR. Washington, D.C.: Conference Board.
172 References
Benner, J. 2002. Bush plan “digital distortion.” Wired News (February 7).
Available [online]: www. wired, com/news/print/0,1294,50279,00. bpnl
[15 May 2002],
Benton Foundation. 1996. Buildings, books, and bytes: Libraries and com¬
munities in the digital age, a report on the publics opinion of library lead¬
ers’ visions for the future. Washington, D.C.: Benton Foundation.
Bernhardt, A., M. Morris, M. S. Handcock, and M. A. Scott. 2001. Diver¬
gent paths: Economic mobility in the new American labor market. New
York: Russell Sage Foundation.
Bimber, B. 1999. The Internet and citizen communication with govern¬
ment: Does the medium matter? Political Communication 16 (4):
409-28.
-. 2001. Information and political engagement in America: The
search for effects of information technology at the individual level.
Political Research Quarterly 54 (1): 53—67.
Bowler, S., andT. Donovan. 1998. Demanding choices. Ann Arbor: Univer¬
sity of Michigan Press.
Bowler, S., T. Donovan, and C. Tolbert, eds. 1998. Citizens as legislators:
Direct democracy in the United States. Columbus: Ohio State University
Press.
Brookings Institution. 2002. The future of Internet voting. A symposium
cosponsored by the Brookings Institution and Systems, Inc., January
20. Available [online]: www.brookings.org/comm/transcripts/
20000120.htm [15 May 2002],
Bucy, E. P. 2000. Social access to the Internet. Harvard International Jour¬
nal of the Press-Politics 5(1): 50-61.
Budge, I. 1996. The new challenge of direct democracy. Cambridge, Mass.:
Blackwell.
Bushweller, Kevin. 2001. Beyond machines. Education Week on the Web.
Available [online]: www.edweek.org/sreports/tcO 1/tco/article.cfm?
slug=35pittsburgh.h20 [10 July 2002],
Butler, D., and Ranney, A. 1994. Referendums around the world: The grow¬
ing use of direct democracy. Washington, D.C.: AEI Press.
Campbell, A., P. E. Converse, W. E. Miller, and D. E. Stokes. I960. The
American voter. Chicago: University of Chicago Press.
Cancian, M„ and D. R. Meyer. 2000. Work after welfare: Women’s work
effort, occupation, and economic well being. Social 97ork Research 24
(2): 69-86.
Carnevale, A. P., and S. J. Rose. 2001. Low earners: Who are they? Do they
have a way out? In Low-wage workers in the new economy, edited by
R. Kazis and M. C. Miller, 45-66. Washington, D.C.: Urban Institute
Press.
References 173
Carvin, A., with C. Conte and A. Gilbert. 2001. The E-rate in America:
A tale of four cities. In The digital divide: Facing a crisis or creating a
myth? edited by B. M. Compaine, 223-42. Cambridge, Mass.: MIT
Press.
Castells, M. 2000. The rise of the network society. 2d ed. Oxford: Blackwell.
Chadwick, A., with Christopher May. 2001. Interaction between states and
citizens in the age of the Internet: E-government in the United States,
Britain, and the European Union. Presented at the annual meeting of
the American Political Science Association, 30 August-2 September,
San Francisco.
Chew, F. 1994. The relationship of information needs to issue relevance and
media use. Journalism Quarterly 71: 676-88.
Chow, C., J. Ellis, J. Mark, and B. Wise. 1998. Impact of CTCNet affiliates:
Findings from a national survey of users of community technology centers.
Available [online]: www.ctcnet.org/impact98.htm [25 March 2003].
Civille, R. 1995. The Internet and the poor. In Public access to the Internet,
edited by B. Kahin and J. Keller, 175-207. Cambridge, Mass.: MIT
Press.
Clarke, S. E., and G. L. Gaile. 1998. The work of cities. Minneapolis: Uni¬
versity of Minnesota Press.
Cliff, S. 2000. The e-democracy e-book: Democracy is online 2.0. Available
[online]: www.e-democracy.org.do [12 January 2002],
Compaine, B. M. , ed. 2001. The digital divide: Facing a crisis or myth? Cam¬
bridge, Mass.: MIT Press.
Conway, M. 1991. Political participation in the United States. 2d ed. Wash¬
ington, D.C.: CQ Press.
Davis, R. 1999. The web of politics: The Internet’s impact on the American
political system. New York: Oxford University Press.
Davis, R., and D. Owen. 1998. New media and American politics. New York:
Oxford University Press.
Dionne, E. J. 1996. They only look dead: Why progressives will dominate the
new political era. New York: Simon & Schuster.
Dryzek, J. S. 1990. Discursive democracy: Politics, policy and political science.
Cambridge: Cambridge University Press.
Dutton, W. H. 1999. Society on the line: Information politics in the digital
age. Oxford: Oxford University Press.
Education Week on the Web. 2001. The new divides, technology counts 2001.
Available [online]: www.edweek.org/sreports/tc01article.cfm?slug=
35execsum.h20 [10 July 2002],
Ellwood, D. T. 2000. Winners and losers in America: Taking the measure of
the new economic realities. In A working nation: Workers, work, and
government in the new economy, edited by D. T. Ellwood, R. M. Blank,
174 References
J. Blasi, D. Kruse, W. A. Niskanen, and K. Lynn-Dyson, 1—41. New

York: Russell Sage Foundation. *
Evelyn, J. 2002. Nontraditional students dominate undergraduate enroll¬
ments, U. S. study finds. Chronicle of Higher Education. Available
[online]: http://chronicle.com/free/2002/2002060402n.htm [ 10 June
2002],
Fagan, J. C., and B. D. Fagan. 2001. Citizens’ access to on-line state leg¬
islative documents. Government Information Quarterly 14 (2): 173—89.
Fishkin, J. 1993. Democracy and deliberation. New Flaven, Conn.: Yale Uni¬
versity Press.
Fountain, J. E. 2001. Building the virtual state: Information technology and
institutional change. Washington, D.C.: Brookings Institution.
Gartner Group. 2000. Key issues in e-government strategy and manage¬
ment. Research notes, key issues. Stanford, Calif.: Gartner Group.
Gerber, E. R. 1999. The populist paradox. Princeton, N.J.: Princeton Uni¬
versity Press.
Gibson, R. 2002. Elections online: Assessing internet voting in light of the
Arizona Democratic Primary. Political Science Quarterly (winter):
561-83.
Ginsburg, L., J. Sabatini, and D. A. Wagner. 2000. Basic skills in adult
education and the digital divide. In Learning to bridge the digital
divide, 77-89. Paris: Organization for Economic Cooperation and
Development.
Gladden, T., and C. Taber. 2000. Wage progression among less skilled work¬
ers. In Finding jobs: Work and welfare reform, edited by D. Card and
R. M. Blank, 160—92. New York: Russell Sage Foundation.
Goldin, C. 2001. The human capital century and American leadership: Virtues
of the past. National Bureau of Economic Research working paper
8239. Available [online]: www.nber.org/papers/w8239 [10 May
2002],
Greengard, S. 1998. Going for the distance: As the technology advances,
more and more employees are reaping the benefits of distance learning.
Industry Week 247 (9): 22-23.
Grossman, L. 1995. The electronic commonwealth. New York: Penguin.
Hague, B., and B. Loader. 1999. Digital democracy: Discourse and decision¬
making in the information age. London: Routledge.
Hall, D. T„ and P. Mirvis. 1995. Careers as lifelong learning. In The chang¬
ing nature of work, edited by A. Howard, 323-64. San Francisco:
Jossey-Bass.
Hart-Teeter. 2000. E-government: The next American revolution. Prepared
for the Council for Excellence in Government. Available [online]:
www.excelgov.org [5 December 2001],
References 175
Heldrich (John J.) Center for Workforce Development. 2002. One-stop

innovations: Leading change under the WLA one-stop system. Pre¬
pared for the U.S. Department of Labor/Employment and Training
Administration/Office of Workforce Security. Available [online]: www.
heldrich.rutgers.edu/Resources/Publication/85/
PromisingPracticesFullReport.pdf [25 May 2002],
Herzenberg, S., J. Alic, and H. Wial. 1998. New rules for a new economy:
Employment and opportunity in postindustrial America. Ithaca, N.Y.:
Cornell University Press.
Hess, F. M., and D. L. Leal. 2001. A shrinking digital divide? The provision
of classroom computers across urban school systems. Social Science
Quarterly 82 (4): 765-78.
Ho, A. T. 2002. Reinventing local governments and the e-government ini¬
tiative. Public Administration Review 62 (4): 434—44.
Hochschild, J. L. 1995. Facing up to the American dream: Race, class, and the
soul of the nation. Princeton, N.J.: Princeton University Press.
Hochschild, J. L., and N. Scovronick. 2000. Democratic education and the
American Dream. In Rediscovering the democratic purposes of education,
edited by L. M. McDonnell, P. M. Timpane, and R. Benjamin,
209-42. Lawrence: University Press of Kansas.
Hoffman, D. L., T. P. Novak, and A. Schlosser. 2000. The evolution of the
digital divide: How gaps in internet access may impact electronic com¬
merce. Journal of Computer Mediated Communication 5 (3). Available
[online]: www.ascusc.org/jcmc/vol5/issue3/hoffman.html [15 Febru¬
ary 2001].
Hoffman, D. L., T. P. Novak, and A. E. Schlosser. 2001. The evolution of the
digital divide: Examining the relationship of race to Internet access and
usage over time. In The digital divide: Facing a crisis or creating a myth?
edited by B. M. Compaine, 47-98. Cambridge, Mass.: MIT Press.
Holsendolph, E. 2002. U.S. sees future, and it is broadband. Atlanta Jour¬
nal and Constitution, 29 September, 6P.
Holzer, H. J. 1996. What employers want: Job prospects for less-educated work¬
ers. New York: Russell Sage Foundation.
Horrigan, J. B., and L. Rainie. 2002. Getting serious online (Pew Internet
and American Life Project). Available [online]: www.pewinternet.org/
reports/pdfs/ PIP_Getting_Serious_Online3ng.pdf [10 March 2002],
Initiative and Referenda Institute. 2002. Ballotwatch. Available [online:
www.iandrinstitute.org [8 June 2002],
International Technology Education Association. 2002. Executive summary
of standards for technological literacy. In Standards for technological
literacy: Content for the study of technology, 2d ed. Available [online]:
www.iteawww.org/TAA/STLexesum.htm [20 March 2003].
176 References
James, E. 2001. Learning to bridge the digital divide. OECD Observer 224:
43-45. v
Jefferson, T. [1816] 1988. A letter to Samuel Kercheval. In Political thought
in America: An anthology, edited by M. B. Levy, 160—62. Chicago:
Dorsey Press.
Jesdanun, A. 2002. Battle over digital divide moves to homefront: Internet
use up, but minorities and poor less likely to have home hookup. Akron
Beacon-Journal, 18 March, B1.
Johnson, J. H., Jr. 2002. A conceptual model for enhancing community
competitiveness in the new economy. Urban Affairs Review 37 (6):
763-79.
Kaestle, C. E, A. Campbell, J. D. Finn, S. T. Johnson, and L. J. Mickulecky.
2001. Adult literacy and education in America: Four studies based on the
National Adult Literacy Survey. NCES publication number 2001534.
Washington, D.C.: U.S. Department of Education, National Center
for Education Statistics. Available [online]: http://nces.ed.gov/
pubsearch/pubsinfo.asp?pubid=2001534 [23 May 2002].
Kasarda, J. D. 1990. City jobs and residents on a collision course: The urban
underclass dilemma. Economic Development Quarterly A (4): 286—307.
Kettl, D. 2000. The global public management revolution: A report on the trans¬
formation of governance. Washington, D.C.: Brookings Institution.
King, G. 1997. A solution to the ecological inference problem. Cambridge,
Mass: Harvard University Press.
King, G., M. Tomz, and Jason Wittenberg. 2000. Making the most of sta¬
tistical analysis: Improving interpretation and presentation. American
Journal of Political Science 44: 347-361.
Kirsch, I. S., A. Jungeblut, L. Jenkins, and A. Kolstad. 2002. Adult literacy
in America: A first look at the findings of the National Adult Literacy Sur¬
vey. 3d ed. NCES publication number 1993-275. Washington, D.C.:
U.S. Department of Education, National Center for Education Statis¬
tics. Available [online]: nces.ed.gov/pubs93/93275.pdf [20 March
2003].
Kruse, D., and J. Blasi. 2000. The new employer-employee relationship. In
A working nation: Workers, work, and government in the new economy,
edited by D. T. Ellwood, R. M. Blank, J. Blasi, D. Kruse, W. A. Niska-
nen, and K. Lynn-Dyson, 42—91. New York: Russell Sage Foundation.
Lacey, M. 2000. Clinton to seek U.S. subsidies to help the poor get online.
New York Times, 22 January, A7.
Layne, K., and J. W. Lee. 2001. Developing fully functional e-government:
A four-stage model. Government Information Quarterly 18 (2): 122-36.
Lewis, M. M. 2001. Next: The future just happened. New York: W. W.
Norton.
References 177
Light) J. S, 2001. Rethinking the digital divide. Harvard Educational Review

71 (4): 709-33.
Magleby, D. 1984. Direct legislation: Voting on ballot propositions in the
United States. Baltimore: Johns Hopkins University Press.
Mann, H. [1839] 1988. The necessity of education in a republican govern¬
ment. In Political thought in America: An anthology, edited by M. B.
Levy, 209—16. Chicago: Dorsey Press.
Manzo, K. K. 2001. Academic record. Education Week on the Web. Avail¬
able [online]: www.edweek.org/sreports/tc01/tc01article.cfm?slug=
35academic.h20 [10 July 2002].
Margolis, M., and D. Resnick. 2000. Politics as usual: The cyberspace “revo¬
lution. ” Thousand Oaks, Calif.: Sage.
Markle Foundation. 1999. Government programs involving citizen access
to Internet services. Available [online]: http://fdncenter.org/pnd/
19990727/webreview.html [2 August 2002],
McChesney, R. 1999. Rich media, poor democracy: Communication policy in
dubious times. Urbana: University of Illinois Press.
McGuckin, R. H., and B. Van Ark. 2001. Making the most of the informa¬
tion age: Productivity and structural reform in the new economy. Perspec¬
tives on a Global Economy. Conference Board report 1301-07-RR.
Washington, D.C.: Conference Board.
McMahon, C., and C. Bruce. 2002. Information literacy needs of local staff
in cross-cultural development projects. Journal of International Devel¬
opment 14: 113—27.
McNeal, R., C. Tolbert, K. Mossberger, and L. Dotterweich. 2003. “Inno¬
vating in digital government in the American states.” Social Science
Quarterly 84 (1). In press.
Melitski, J. 2001. The world of e-government and e-governance. Available
[online]: www.aspanet.org [5 January 2002],
Mendolsohn, M., and A. Parkin, eds. 2001. Referendum democracy: Citizens,
elites, and deliberation in referendum campaigns. New York: MacMillan.
Mickelson, R. A. 1996. Opportunity and danger: Understanding the busi¬
ness contribution to public education. In Implementing educational
reform: Sociological perspectives on educational policy, edited by K. M.
Borman, P. W. Cookson, Jr., A. R. Sadounk, and J. Z. Spade, 245-72.
Norwood, N.J.: Ablex.
Mishel, L., J. Bernstein, andj. Schmitt. 2001. The state of working America:
2000-2001. Ithaca, N.Y.: Cornell University Press/ILR Press.
Moss, P., and C. Tilly. 2001. Stories employers tell: Race, skill, and hiring in
America. New York: Russell Sage Foundation.
Mossberger, K. 2000. The politics of ideas and the spread of enterprise zones.
Washington, D.C.: Georgetown University Press.
178 References
National Academy of Sciences. 1996. Summary and recommendations. In

Measuring poverty: A new approach. Available [online]: www.nap.
edu/readingroom/books/poverty/summary.html [8 September 2002].
National Center for Education Statistics (NCES). (n.d.). Defining and mea¬
suring literacy. Available [online]: http://nces.ed.gov/naal/defining/
defining.asp [23 May 2002].
National Conference of State Legislatures. 2001. Information technology
and Internet laws. Available [online]: www.ncsl.org/programs/lis/
legislation/ITlawsO 1 .htm [ 15 July 2002].
National Performance Review. 1993. From red tape to results: Creating a
government that works better and costs less. Washington, D.C.: Gov¬
ernment Printing Office.
Neu, C. R., R. H. Anderson, andT. K. Bikson. 1999. Sending your govern¬
ment a message: E-mail communication between citizens and government.
Santa Monica, Calif.: RAND.
Nie, N., and L. Erbring. 2000. Internet and society: A preliminary report.
Stanford, Calif.: Stanford Institute for the Quantative Study of Soci¬
ety, Stanford University. Available [online]: www.stanford.edu/
group.siqss/press_release/preliminary_report.pdf [10 February 2002],
Niskanen, W. A. 2000. Creating good jobs and good wages. In A working
nation: Workers, work, and government in the new economy, edited by D.
T. Ellwood, R. M. Blank, J. Blasi, D. Kruse, W. A. Niskanen, and K.
Lynn-Dyson, 92—104. New York: Russell Sage Foundation.
Norris, D. F., P. D. Fletcher, and S. Holden. 2001. Is your local government
plugged in? Highlights of the 2000 Electronic Government Survey. Pre¬
pared for the International City/County Management Association and
Public Technology, Inc. Available [online]: www.umbc.edu/mipar/
fmal_draft/PDFs/e-gov.icma.final-4-25-01 .pdf [ 10 June 2002].
Norris, P. 1999. Critical citizens: Global support for democratic governance.
Oxford: Oxford University Press.
-. 2001. Digital divide: Civic engagement, information poverty, and the
Internet worldwide. New York: Cambridge University Press.
Novak, T. P., D. L. Hoffman, and A. Venkatesh. 1997. Diversity on the
Internet: The relationship of race to access and usage. Presented at the
Aspen Institute Forum on Diversity and the Media, October 1997.
Office of Management and Budget. 2002. United States budget fiscal year
2003. Available [online]: www.whitehouse.gov/omb/budget/fy2003/
budget.html [10 July 2002],
Organization for Economic Cooperation and Development. 2000. Learn¬
ing to bridge the digital divide. Paris: Organization for Economic Coop¬
eration and Development.
References 179
Osborne, D., and T. Gaebler. 1992. Reinventing government: How the entre¬
preneurial spirit is transforming the public sector. Reading, Mass. Addi-
son-Wesley.
Osterman, P. 1999. Securing prosperity: The American labor market: How it
has changed and what to do about it. Princeton, N.J.: Princeton Uni¬
versity Press.
-. 2001. Employers in the low-wage/low-skill labor market. In Low-
wage workers in the new economy, edited by R. Kazis and M. S. Miller,
67—87. Washington, D.C.: Urban Institute Press.
Pateman, C. 1970. Participation and democratic theory. New York: Cam¬
bridge University Press.
Pear, R. 2002. Number of people living in poverty increases in U.S. New
York Times, 25 September, Al.
Pearson, G., andT. A. Young, eds. 2002. Technically speaking: Why all Amer¬
icans need to know more about technology. Washington, D.C.: National
Academy of Engineering Committee on Technological Literacy.
Penuel, W., and D. Kim. 2000. Promising practices and organizational chal¬
lenges in community technology centers. Available [online]: www.
sri. com/policy/ctl/assets/images/vStreets_Promising_Practices. pdf.
Peters, B. G. 1996. The Future of governing: Four emerging models. Lawrence:
University Press of Kansas.
Pew Internet and American Life Project. 2000. African-Americans and the
Internet by T. Spooner and L. Rainie. Available [online]: www.
pewinternet.org/reports/pdfs/PIP_African_Americans_Report.pdf [5
January 2002],
-. 2002. The rise of the e-citizen: How people use government agen¬
cies’ web sites. By E. Larsen and L. Rainie. Pew Internet and American
Life Project. Available [online]: www.pewinternet.org/reports/
pdfs/PIP_Govt_Website_Rpt.pdf [15 August 2002],
Pew Research Center. 2000. Campaign and Internet survey. October
10-November 25. Available [online]: www.pewinternet.org/datasets/
index.asp [5 January 2002].
Phillips, H. L 2002. Making a case for net access. San Jose Mercury News, 1
October, 3C.
Piven, F. F., and R. A. Cloward. 1988. Why Americans don’t vote. New York:
Pantheon.
Puma, M. J., D. D. Chaplin, and A. D. Pape. 2000. E-rate and the digital
divide: A preliminary analysis from the integrated studies of educational
technology. Washington, D.C.: Urban Institute. Available [online]:
www.urban.org/education/erate.html [10 July 2002].
Putnam, R. 2000. Bowling alone: The collapse and revival of American com¬
munity. New York: Simon & Schuster.
180 References
Rangarajan, A., P. Schochet, and D. Chu. 1998. Employment experiences of

welfare recipients who find jobs: Is targeting possible? Princeton, NtJ.:
Mathematica Policy Research, Inc. Available [online]: www.mathinc.
com [ 10 January 2002].
Rheingold, H. 1993. The virtual community: Homesteading on the electronic
frontier. Reading, Mass.: Addison-Wesley.
Rosenstone, S. J., and M. Hansen. 1993. Mobilization, participation, and
democracy in America. New York: Macmillan.
Rosenthal, A. 1997. The decline of representative democracy. Washington,
D.C.: CQ Press.
Salamon, L. M. 1991. Overview: Why human capital? Why now? In An eco¬
nomic strategy for the 90s: Human capital and America’s future, edited by
D. W. Hornbeck and L. M. Salamon, 1-39. Baltimore: Johns Hopkins
University Press.
Saulny, S. 2001. New jobless centers offer more than a benefit check. New
York Times, 5 September, Al, A21.
Schattschneider, E. E. 1960. The semi-sovereign people. New York: Holt,
Rinehart, and Winston.
Scheufele, D., and D. Shah. 2000. Personality strength and social capital:
The role of dispositional and informational variables in the production
of civic participation. Communication Research 21 (2): 107-31.
Secretary’s Commission on Achieving Necessary Skills. 1992. Learning a liv¬
ing: A blueprint for high performance: a SCANS report for America
2000. Available [online]: http://wdr.doleta.gov/SCANS/lal/lal.htm [5
January 2002],
Seifert, J. and E. Petersen. 2002. The promise of all things E? Expectations
and challenges of emergent electronic government. Perspectives on
Global Development and Technology 1 (2): 193-212.
Servon, L. J., and M. K. Nelson. 2001. Community technology centers:
Narrowing the digital divide in low-income, urban communities. Jour¬
nal of Urban Affairs 23: 279-90.
Shadid, A. 2001. Report shows narrowing digital divide: GAO finds net use
up in rural areas, among minorities. Boston Globe, 22 February, C1.
Shah, D., N. Kwak, and R. Holbert. 2001. “Connecting” and “disconnect¬
ing with civic life: Patterns of Internet use and the production of social
capital. Political Communication 18: 141-62.
Sipple, J. W., C. G. Miskel, T. M. Matheney, and C. P. Kearney. 1997. The
creation and development of an interest group: Life at the intersection
of big business and education reform. Educational Administration
Quarterly 33: 440-73.
Smith, E. 1989. The unchanging American voter. Berkeley: University of Cal¬
ifornia Press.
References 181
Solop, F. I. 2000. Digital democracy comes of age in Arizona: Participation

and politics in the first binding Internet election. Presented at Annual
Meeting of the American Political Science Association, 31 August-3
September, Washington, D.C.
Stiglitz, J. E., P. R. Orszag, and J. M. Orszag. 2000. The role of government
in a digital age. Commissioned by the Computer and Communica¬
tions Industry Association. Available [online]: www.ccianet.org/
digital_age/report.pdf [20 August 2002].
Stone, D. 2002. Policy paradox: The art of political decision making. Rev. ed.
New York: Norton.
Stowers, G. N. L. 1999. Becoming cyberactive: State and local governments
on the World Wide Web. Government Information Quarterly 16 (2):
111-27.
Strawn, J., and K. Martinson. 2001. Promoting access to better jobs: Lessons
for job advancement from welfare reform. In Low-wage workers in the
new economy, edited by R. Kazis and M. S. Miller, 111-34. Washing¬
ton, D.C.: Urban Institute Press.
Sunstein, C. 2001. Freedom of expression in the United States: The future.
In The boundaries of freedom of expression and order in American demo¬
cracy, edited byT. Hensley, 319-47. Kent, Ohio: Kent State University
Press.
Swaim, P. L., R. M. Gibbs, and R. Teixeira. 1998. Introduction to Rural edu¬
cation and training in the new economy: The myth of the rural skills gap,
edited by R. M. Gibbs, P. L. Swaim, and R. Teixeira, 3-22. Ames: Iowa
State University Press.
Tapscott, D. 1997. The digital media and the reinvention of government.
Canadian Public Administration 40 (2): 328—45.
Teixeira, R., and D. A. McGranahan. 1998. Rural employer demand and
worker skills. In Rural education and training in the new economy: The
myth of the rural skills gap, edited by R. M. Gibbs, P. L. Swaim, and R.
Teixeira, 115—30. Ames: Iowa State University Press.
Thierer, A. D. 2000. How free computers are filling the digital divide. Report
no. 1361. Heritage Foundation Backgrounder. Available [online]:
www.heritage.org/library/backgrounder/bgl36l.html [5 January
2002],
Thurow, L. C. 1999. Building wealth: The new rules for individuals, compa¬
nies, and nations in a knowledge-based economy. New York: Harper-
Collins.
Toffier, A. 1995. Creating a new civilization: The politics of the third wave.
Kansas City, Mo.: Turner.
Tolbert, C., and R. McNeal. 2003. Does the Internet increase voter turnout
in elections? Political Research Quarterly 56:2.
182 References
Tolbert, C., M. Stansbury, and K. Mossberger. 2001. Defining the digital

divide: Survey. Available from authors [[email protected]]. v
Trotter, A. 2001. Closing the digital divide. Education Week on the Web.
Available [online]: www.edweek.org/sreports/tcO 1 article.cfm?slug=35
solutions.h20 [10 July 2002].
Uchitelle, L. 1999. Devising new math to define poverty: Millions more
would be poor in fresher census formula. New York Times, 18 October,
Al, A14.
U.S. Census Bureau. 2000. Quick Tables. QT-PE, Race and Hispanic or
Latino: 2000. Data Set: Census 2000 Summary File 1 (SF 1) 100-Per¬
cent Data; Geographic Area: Berkeley city, California. Available
[online]: http://factfinder.census.gov/bf/_lang=e?_SFl_U_QTP3_
goe_id= 16000US0606000.htm [10 June 2002].
-. 2001. Income 2000. Table 1. Available [online]:
www.census.gov/hhes/income/incomeOO/inctabl.html [30 March
2003].
-. 2002a. Poverty 2001. Available [online]: www.census.gov/
hhes/poverty/threshld/thresh01.html [30 March 2003].
-. 2002b. Income 2001. Table 1. Available [online]: www.
census.gov.hhes/income/income01/inctabl.html [30 March 2003].
U.S. Department of Commerce. 1995. National Telecommunication and
Information Administration. Falling through the net: A survey of the
“Have Nots” in rural and urban America. Available [online]: www.
ntia.doc.gov/ntiahome/fallingthru.html [18 March 2002],
-. 1998. National Telecommunication and Information Administra¬
tion. Falling through the net II: New data on the digital divide. Available
[online]: www.ntia.doc.gov/ntiahome/net2/falling.html [18 March
2002],
-. 1999. National Telecommunication and Information Administra¬
tion. Falling through the net: Defining the digital divide. Available
[online]: www.ntia.doc.gov/ntiahome/fttn99/FTTN.pdf [18 March
2002],
-. 2000a. Background for digital economy 2000 report.
Available [online]: http://osecnt 13-osec.doc.gov/public.nsf/ docs/
5E4B499384543E86852568F5006F1AA4 [2 March 2002],
-. 2000b. National Telecommunications and Information Adminis¬
tration. Falling through the net: Toward digital inclusion. Available
[online]: www.esa.doc.gov/fftnOO.htm [18 March 2002],
-. 2002. National Telecommunication and Information Admin¬
istration. A nation online: How Americans are expanding their use of the
Internet. Available [online] :www.ntia. doc.gov/ntiahome/dn/
anationonline2.pdf [18 March 2002],
References 183
U.S. Department of Education. 2002. ED programs that help bridge the

digital divide. Available [online]: www.ed.gov/Technology/digdiv.html
[27 August 2002].
U.S. Department of Labor. 1999. Futurework: Trends and challenges for work
in the 21st century. Available [online]: www.dol.gov/dol/asp/public/
futurework/report/chapter2/main.htm [10 February 2000].
-. 2002. Employment and Training Adminstration. Workforce invest¬
ment act; Lower living standard income level. Available [online]:
http://wsdc.doleta.gov/llsil01.asp [July 10, 2002].
Verba, S., K. L. Schlozman, and H. E. Brady. 1995. Voice and equality: Civic
voluntarism in American politics. Cambridge, Mass.: Harvard Univer¬
sity Press.
Verba, S., and N. H. Nie. 1972. Participation in America. New York: Harper
& Row.
Walsh, E. O, with M. E. Gazala and C. Ham. 2001. The truth about the
digital divide. In The digital divide: Facing a crisis or creating a myth?
edited by B. M. Compaine, 279-84. Cambridge, Mass.: MIT Press.
Weber, L. M., and J. Bergman. 2001. Who participates and how? A com¬
parison of citizens “online” and the mass public. Presented at the
Annual Meeting of the Western Political Science Association, 15-17
March, Las Vegas.
West, D. M. 2000. Assessing e-government: The Internet, democracy, and ser¬
vice delivery by state and federal governments. Washington, D.C.: World
Bank. Available [online]: wwwl.worldbank.org/publicsector/egov/
EgovReportUSOO.htm [21 June 2002],
-. 2001. E-government and the transformation of public sector ser¬
vice delivery. Presented at the annual meeting of the American Politi¬
cal Science Association, August 30-September 2, San Francisco.
-. Forthcoming. E-government and the transformation of service
delivery and citizen attitudes. Public Administration Review.
Wilhelm, A. 2000. Democracy in the digital age: Challenges to political life in
cyberspace. New York: Routledge.
Wolfinger, R., and S. J. Rosenstone. 1980. Who votes? New Haven, Conn.:
Yale University Press.
Wright, G. 2002. Groups, lawmakers protest Bush plan to cut programs to
bridge digital divide. Gannett News Service (May 26). Available
[online]: www.detnews.corn/2002/technology/0205/15/technology-
489537.htm [10 June 2002],
I,.
Index
A age differences
access divide, 15—37 computer skills, attitudes towards, 71
ANES survey data, 26—29 courses online, 77
barriers for bridging the, 121 Internet use study, 21
e-mail address, 142 table job search online, 73
federal policy, 4, 15—16, 34, 35 low-income survey, 29, 32
Internet access at home, 142 table research measures in years, 27
low-income survey, 24—26, 29—33 skills index, 46
measuring the, 116—17 voting and registration online,
multivariate regression table, 142 101-2
NTIA survey, 34—35 Alvarez, R. M., Internet voting, 92, 104
redefining the digital divide, 1—2 American Association of School Librar¬
research methodology, 9, 23—24 ians, information literacy
research studies on, 11—12, 16—23 standards, 57n. 2
respondent has Internet access, 141 American Council of the Blind, 15
table American Library Association (ALA),
survey results, 33—35 information literacy standards, 41,
ActionForum, Berkeley online experi¬ 57n. 7
ment, 106, 107, ll4n. 72 American National Election Studies
African Americans (ANES)
computer skills, attitudes towards, description of, 24
70 findings from data analysis, 16,
economic opportunity, 80—81 28-29, 34
education gap, 17, 63 Internet access and voter turnout, 90
e-government, 96, 101 patterns of access, 26—28
e-mail address, 30 Anderson, R. H., Internet access and
home computers, 18, 30 e-mail use study, 20—21
instructional preferences, 54 ANES. See American National Election
Internet access, 16, 18, 19, 28, 30, Studies (ANES)
Arizona, Internet voting case study, 89,
33, 34
Internet use, 22 92-93, 104-5, 109, 112n. 33
job search online, 73—75 Asian Americans
online courses, 78—79 e-mail use, 21
instructional preferences, 54
and public access sites, 50, 51
Internet access, 21, 30—31, 32
technical competence, 47
“African-Americans and the Internet” and public access sites, 50—51
(Pew Internet and American Life public participation online, 101,
113-l4n. 63
Project), 20
186 Index
B as public access sites, 50—52

Barber, Benjamin, an advocate of direct skills developmefit, 39, 47—50
democracy, 88 Compaine, Benjamin, edited volume,
basic literacy, 42-44 23
adult literacy survey, 42-43 computer skills
document literacy, 42, 43 attitudes about, 70—71, 147 table

and economic opportunity, 66-67 as job requirement, 68-71
prose literacy, 42, 43 multivariate regression table, 144
Benton Foundation, public library’s table
role in the community, 48 See also information literacy; skills
Bergman, J., Internet access and polit¬ divide; technical competence
ical activities, 91 computer usage
Bikson, T. K., Internet access and friend or relative’s computer, 25-26
e-mail use study, 20-21 home computers, 25-26
Bimber, B., Internet access and voter at public libraries, 45, 58-59n. 22,
participation, 90-91, 101 59n. 34, 145 table
Bruce, C., information literacy needs workplace, 25-26, 64, 65, 83-84nn.
study, 57-58n. 11 19, 22, 23
Bureau of Labor Statistics, fastest- See also Internet use
growing occupations, 64 Conference Board, on information
Bush administration, digital divide technology, 60
policy, 4, 15, 35 CTCNet Research and Evaluation
Team, 49
c
California D
Berkeley online e-government exper¬ Davis, R., impact of Internet on
iment, 11, 95, 106-8, ll4-15n. politics, 87, 108
73 democratic divide, 86—115
Internet voting ballot initiative, 89 barriers to bridging the, 121
Santa Monica Public Electronic Berkeley, California, online experi¬
Network, 95 ment, 11,95, 106-8, 114—15n.
careers. See economic opportunity 73
divide; job issues and digital democracy theories,
Castells, Manuel, on “information- 88-89
alism,” 60 measuring/mapping the, 119-20
Cheskin Research, 22, 23 origin of the term, xii, 13n. 1
Clinton administration political participation research
digital divide policy, 2-3, 15 studies, 90—91
e-government initiatives, 94 and public participation, 13
Commerce, Department of. See redefining the digital divide, 2, 5
National Telecommunications and research methodology, 9
Information Administration survey findings, 97-99
(NTIA) survey results, 109-10
Community Technology Centers See also e-government; Internet
(CTCs), 3-4, 15 voting
course offerings, 49 Democrats. See partisanship
Index 187
digital democracy skills requirements and, 66—67

and e-government, 151 survey results, 79-82
multivariate regression table, 151 ways in which technology enhances,
support for, 99—105 61
theories of, 88-89 See also low-income survey;
vs. actual experience, 152 workplace
digital divide education gap
e-government, 151 and the digital divide, 124—25
federal policy, 3—4, 15-16, 34, 35 Internet use study, 21
future projections, 120-25 low-income survey, 31-32
origin of the term, 3, I4n. 4 NTIA survey, 34—35
redefining as multiple divides, xii, race/ethnicity, 17, 18, 20, 63
1-14 skills divide, 46—47
term defined, 1 See also learning
See also access divide; democratic educational attainment
divide; economic opportunity and computer use at work, 64—66
divide; skills divide and economic opportunity, 62
Digital Divide Network, 3-4 and job search online, 73-74
Digital Empowerment Campaign, race/ethnicity gap, 17, 18, 20, 63
15-16 research measures for education
digital experience, 125-26, 153 table variable, 27, 29, 69
direct democracy, 88 voting and registration online,
101-2
E e-government
economic opportunity divide, 60-85 for business and economic develop¬
attitudes about computers and, 147 ment, 94
table case study (Berkeley, California), 11,
barriers for bridging the, 121 95,106-8, 114—15n. 73
computer skills as job requirement, defined, 93, 94
68-71 for driver’s licenses, 94, 97
computers and less-educated evolutionary stages of, 112—13n. 46
workers, 64—66 for fishing, hunting, and recreation
computers in the workplace, 25-26, licenses, 94, 97
64 for government information, 97, 98
and democratic citizenship, 6 multivariate regression table, 151
and education gap, 62-63 online forums/town meetings, 11,
and information networks, 71-73 95, 99, 100, 106-8, 151
job search online, 71-75, 148 table, Pew Internet and American Life
149 table, 150 table Project study, 96
lifelong learning, 75-76 and political participation, 87
measuring/assessing the, 118—19 state and local government websites,
and the “new economy,” 62 93,94-95
online courses, 76—79, 148 table, support for, 99-105
149 table, 150 table for tax filing, 94
and redefining the digital divide, 2, 5 theories of, 93-97
research methodology, 9 for vehicle registration, 94, 97
188 Index
e-mail address computer skills, attitudes towards, 70

and access divide, 142 table courses online in public places, 78
race/ethnicity data, 30 as dummy variable, 27
those most least likely to have, 30 education gap, 63
e-mail use and income inequality, 63
patterns of access, 24, 34 and information literacy, 46
race/ethnicity, 20 instructional learning preferences,
English-language skills, 49 53, 54, 56
E-Rate program, 3,15 Internet access,, 28, 34, 46
Erbring, L., Internet access study, 21 Internet voting and, 101
ethnicity gap. See race/ethnicity gap job search online, 74
skills divide, 122-23
F voter registration, 102
Falling through the Net: A Survey of the Georgia, absentee ballots online, 93
“Have Nots” in Rural and Urban Gore, Al, and the information super¬
America (NTLA), 3, 17 highway, 2
Falling through the Net: Defining the government information online, 97, 98
Digital Divide (NTIA), 18
Falling through the Net: New Data on H
the Digital Divide (NTIA), 18 Hoffman, D. L., Internet access study,
Falling through the Net: Toward Digital 20
Inclusion (NTIA), 18-19, 34-35 home computers
Federal Communications Commission, and Internet access at home, 32,
and digital divide issue, 5 37n. 26, 142 table
federal policy number in U.S., 1
on access divide, 3-4, 15-16, 34 race/ethnicity data, 30
on skills divide, 47—50 support for vouchers, 154 table
federal programs those least likely to have, 30
Community Technology Centers usage, 25-26
(CTCs), 3-4, 15 Horatio Alger myth, 6
E-Rate program, 3, 15
Technologies Opportunities Program i
(TOP), 3-4, 15 income gap
Forrester Research Inc., 22, 23 economic opportunity divide factor,
Franklin, Benjamin, and public 80
libraries, 6 income inequality in U.S., 62-63
friend or relative’s computer, usage, and Internet access, 16, 18, 123-24
25-26 low-income survey data, 31
NTIA survey, 35
G research measures, 27
Gates Foundation, 4 skills divide, 123—24
GED' (general education development, See also economic opportunity
for high school equivalency) divide; low-income survey
credential, 49, 75 information literacy
gender differences case studies on needs assessment, 42,
access divide, 122—23 57-58n. 11
Index 189
defined, 38, 41 computer skills requirement, 68-71

and economic opportunity, 67 dislocation rate, 71
low-income survey results, 45 economic advancement, and
movement, 57n. 9 computer skills, 12-13, 61
multivariate regression table, 143 information networks and economic
table opportunity, 71-73
and online courses, 76—77 job search online, 61, 72—75, 94,
research studies on, 12, 42 148 table, 149 table, 150 table
and skills divide, 12, 41—42, 45, 121 networking online, 72-73
standards, 41, 57nn. 7, 10 occupations, fastest-growing, 64,
information “superhighway,” xi, 2 83n. 21
International Technology Education See also workplace
Association (ITEA), technological
literacy standards, 56-57n. 2 L
Internet use Labor, Department of, income
defined, 19 measurements, 10
for e-mail, 40 Latinos
for health information, 40 computer skills, attitude towards, 70
high-poverty tracts, 7—8 economic opportunity and, 81
home computer access, 1, 142 table education gap, 17, 63
multivariate regression tables, 141 e-mail addresses, 30
table home computers, 18, 30
for online purchases, 40 instructional preferences, 54
patterns of access (by Internet access, 16, 18, 19, 30, 32,
location/venue), 24—26 34
for product and service information, Internet use, 22
40 political participation online, 101
Internet voting and public access sites, 51
Arizona case study, 89, 92-93, technical competence, 47
104-5, 109, 112n. 33 learning
attitudes about, 98 distance learning programs, 76
California ballot initiative defeated, GED credential, 49, 75
89 instructional preferences, 53, 54, 56,
multivariate regression table, 151 146 table
table lifelong learning, 61, 75-76
partisanship, 100-105 online courses, 76-79, 148 table,
research studies, 92—93 150 table
support for, 98, 99-105 libraries. See public libraries
survey findings, 99—105 library patronage. See public libraries,
ITEA, 56-57n. 2 usage/library patronage
lifelong learning, 61, 75-76
i literacy. See basic literacy; information

literacy
Jefferson, Thomas, and public
education, 6 low-income survey
job issues access divide, 35
career mobility research, 63, 72 data analysis, 29—33
190 Index
low-income survey (continued) National Information Infrastructure, 2

patterns of access and use, 24-26, National Telecommunications and ^
37n. 26 Information Administration
survey questionnaire, 155—67 (NTLA), Internet access surveys,
See also economic opportunity divide 3-4, 15-20, 34-35, 37n.26
National Urban League, 15
M Neu, C. R., Internet access and e-mail
Mann, Horace, common schools use study, 20—21
system, 6 Next: The Future Just Happened, 42
McMahon, C., information literacy Nie, N., Internet access study, 21
needs study, 57-58n. 11 Niskanen, William, on information
McNeal, R., Internet access and voter technology as “third industrial
turnout, 90 revolution,” 60
men and women. See gender differ¬ Norris, Pippa
ences on the democratic divide, xii
Mikulski, Barbara, Sen. (D-Md.), 15 on political participation, 13n. 1
Monte Carlo simulation technique, 11, Novak, T. P., Internet access study, 20
I4n. 22, 24,31,46 NTIA. See National Telecommunica¬
Mossberger, Karen, xii tions and Information Adminis¬
Moveon.org, 106, 107, ll4-15nn. tration (NTIA)
72-73
multivariate regression tables, 141-54 o
tables occupations. See job issues
multivariate regression technique, Organization for Economic Coopera¬
9-10, 24, 59n. 35 tion and Development (OECD)
distance learning programs, 76
N and the “new economy,” 62
Nagler, ]., Internet voting, 92, 104
NALS. See National Adult Literacy P
Survey (NALS) partisanship
NAS. See National Academy of ANES survey data, 28
Sciences (NAS) identification of, as dummy variable,
A Nation Online: How Americans Are 27-28
Expanding Their Use of the Internet and Internet voting, 101
(NTLA), 4, 15, 16, 17, 19,27 low-income survey, 29, 31,33
National Academy of Sciences (NAS) voter registration, 102-3
and digital divide research, 10 voting online, 100-105
and technological literacy, 39-40, Pew Charitable Trust, 17
56n. 2 Pew Internet and American Life
National Adult Literacy Survey Project, 20, 40, 96
(NALS), 42-43 Pew Research Center, Internet access
National Congress of American studies, 19-20, 35
Indians, 15 political participation
National Employer Survey, 83n. 19, experience with, 152 table
83-84n. 23 impact of Internet technology on,
National Geographic Interactive, 91 86-87
Index 191
and Internet access, 6, 13, 13n. 1 job search online, 73—75

Internet access by party, 28 low-income survey, 29, 32-33
partisan/party identification, 27-28 NTIA survey, 34, 35
and reading about politics, 86-87 political participation online, 101
research methodology, 114nn. 64, research measures for race/ethnicity,
65 27
research studies, 90-91 skills divide, 56, 124
support for, 152 table See also African Americans; Asian
See also democratic divide Americans; Latinos
positive externalities, 5-6 recreation centers, as public access site,
Powell, Michael, on digital divide, 5 50
public access sites Republicans. See partisanship
for computers and Internet, 145 research methodology, 7—11
table Computer Assisted Telephone Inter¬
job search online, 148 table viewing (CATI), 7—8
and online education, 77, 78, 149 data analysis, 9—11, 23—24
table data collection, 7—8
for online voting, 103 dependent variable (Internet access),
survey results, 50—52 27
See also Community Technology dummy variables, 27-28
Centers (CTCs); public libraries; explanatory variables, 69
schools income measures, 10, 27
public good independent variables, 27-28, 59n.
information technology skills and 34
access as a, 5 measuring the divides, 9, 27-28
and public education, 6 Monte Carlo simulation technique,
public libraries 11, l4n. 22, 24,31,46
as “community gathering” places, multivariate regression technique,
51, 145 table 9-10, 24, 59n. 35
establishment of in U.S., 6 survey questionnaire, 155—67
federal policy and resources, 4 research studies, 11—12, 16—23
Internet use, 25—26 academic studies, 20—21
as public access site, 50 Internet voting, 92-93
public support for, 145 table low-income survey, 24—26
skills development, 39, 47-50 market research surveys, 22-23
usage/library patronage, 45, 58-59n. NTIA Internet access surveys,
22, 59n. 34, 145 table 17-19
Putman, Robert, on “cyberapartheid,” Pew Research Center surveys,
6 19-20
on political participation, 90-91
R
race/ethnicity gap s
access divide, 12, 117, 124 SCANS, 67
computer skills, attitudes towards, 70 Schattschneider, E. E., 110
and income inequality, 63 Schlosser, A. E„ Internet access study,
instructional preferences, 54 20
192 Index
schools research studies on, 40

federal resources, 4 technological literacy
Internet use, 25—26 ITEA standards for, 56-57n. 2
as public access site, 50 NAS definition of, 39^10, 56n. 2
Secretary’s Commission on Achieving Technologies Opportunities Program
Necessary Skills (SCANS), 67 (TOP), 3-4, 15
senior centers, as public access site, 50 Telecommunications Act of 1996, 3
skills divide, 38-59 Thurow, Lester, on knowledge as “basis
barriers for bridging the, 121 for wealth,” 82
basic computer tasks, 44, 55 Tolbert, Caroline
computer skills as job requirement, Internet access and voter turnout, 90
68-71 profile of, xii
and economic opportunity, 66—67 Tomas Rivera Policy Institute (Clare¬
and information literacy, 12, 41—42, mont College), public access case
45, 121 study, 49
instructional preferences, 52-54, town meetings/forums online, 11, 95,
146 table 99, 100, 106-8, 151 table
low-income survey results, 44—47
measuring/uncovering the, 117—18 ¥
and political participation, 6 voter registration
and public access, 29, 47—50 gender differences, 102
public access sites, 50—52 multivariate regression table, 151
public policy on, 3, 47-50 table
in redefining the digital divide, 2-3, online in some states, 93
4 partisanship, 102—3
research methodology, 9 and political participation, 89
survey results, 54—56 support for online, 98, 99-105
technical competence, 38, 40, 44-47 voter turnout, 86
Solop, F. I., Arizona Internet voting voting online. See Internet voting
case study, 92
Standard Assessment of Information w
Literacy Skills (SAILS), 57n. 10 Weber, L. M., Internet access and polit¬
Stansbury, Mary, xii ical activities, 91
Stone, Deborah, on policymaking, xi Who’s Not Online (Pew Research
Strickland, Ted, Rep. (D-Ohio), 15 Center), 20
survey questionnaire, 155-67 Wilhelm, A., Internet access study, 20,
21, 36n. 20
T women and men. See gender differ¬
technical competence ences
basic computer tasks, 44, 55 workplace
defined, 38, 40 computer use, 25-26, 64, 65,
and economic opportunity, 67 83-84nn. 19, 22, 23
low-income survey results, 44—47 See also job issues
multivariate regression table, 143
table
Public Policy / Information Technology
VIRTUALINEQUALITY
Beyond the Digital Divide
Karen Mossberger, Caroline J. Tolbert,
and Mary Stansbury
"Who should read Virtual Inequality? "Learning how to understand the impact of
Students, teachers, journalists, policy ana¬ the Internet on civic life is one of the most
lysts, government decision makers, communi¬ important challenges facing us today. Anyone
ty leaders, and especially ordinary citizens interested in how technology is reshaping
should read it. The authors ... present and democracy should read this book."
explain a much bigger, more complex, critical —Darrell West, professor of political
set of problems and issues regarding how the science and public policy, and director of
allocation of information and knowledge, the Taubman Center for Public Policy at
political power and opportunity is defining the Brown University
evolving Information Society of the 21st cen¬
"Eminently readable, Virtual Inequality is
tury. Beyond the term 'digital' divide are
chockfull of insightful data, analysis, and
issues about quality of life, democracy, exclu¬
recommendations."
sion, and who gets what, when, and under
—Amanda B. Lenhart, research specialist,
what circumstances."
Pew Internet and American Life Project
—Nolan A. Bowie, senior fellow and
adjunct lecturer in public policy. Harvard
University KAREN MOSSBERGER is an associate profes¬
sor in the Department of Political Science at
" [The authors] draw on both a wealth of
Kent State University.
important data and rigorous analysis to
demonstrate convincingly that persistent dis¬
CAROLINE J. TOLBERT is an assistant profes¬
parities in technology access and skill matter
sor in the Department of Political Science at
for economic opportunity and political partici¬
Kent State University.
pation, and that governments will have to
fashion broader solutions to take full advan¬
MARY STANSBURY is an assistant professor
tage of technology's social potential."
in the School of Library and Information
—Jane E. Fountain, associate professor of
Science at Kent State University.
public policy, and director of the National
Center for Digital Government, John F.
Kennedy School of Government, Harvard American Governance and Public Policy
University
Barry Rabe, series editor
Georgetown University Press

Washington, D.C.
www.press.georgbtown.edu

Virtual Inequality Beyond The Digital Divide

Uploaded by

Copyright:

Available Formats

Virtual Inequality Beyond The Digital Divide

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Virtual Inequality Beyond The Digital Divide

Uploaded by

Copyright:

Available Formats

VIRTUAL

After Disaster: Agenda Setting, Public Policy, and Focusing Events

Budgeting Entitlements: The Politics of Food Stamps

Collaborative Public Management: New Strategies for Local Governments

Conflict Amid Consensus in American Trade Policy

Controlling Technocracy: Citizen Rationality and the NIMBY Syndrome

Dam Politics: Restoring America’s Rivers

Expert Advice for Policy Choice: Analysis and Discourse

Federalism and Environmental Policy: Trust and the Politics of Implementation

Federalism in the Forest: National versus State Natural Resource Policy

Fenced Off: The Suburbanization of American Politics

Green Politics and Global Trade: NAFTA and the Future

Hazardous Waste Siting and Democratic Choice

Improving Governance: A New Logic for Empirical Research

Lobbying Together: Interest Group Coalitions in Legislative Politics

Pluralism by the Rules: Conflict and Cooperation in Environmental Regulation

Policy Entrepreneurs and School Choice

The Political Economy of Special-Purpose Government

The Politics of Automobile Insurance Reform: Ideas, Institutions,

The Politics of Ideas and the Spread of Enterprise Zones

The Politics of Unfunded Mandates: Whither Federalism''

Rethinking Health Care Policy: The New Politics of State Regulation

Taking the Initiative: Leadership Agendas in Congress and the

Virtual Inequality: Beyond the Digital Divide

Welfare Policymaking in the States: The Devil in Devolution

Georgetown University Press

This book is printed on acid-free recycled paper meeting the requirements

Library of Congress Cataloging-in-Publication Data

List of Tables and Figure IX

Tables listed below are in Appendix 1.

A2.2 Respondent Has Internet Access 141

A2.3 The Access Divide: Respondent Has Access

Technical Competence and Information Literacy 143

Computer Skills 144

Attitudes about Library Use 143

Instructional Preferences 146

A4.1 Attitudes about Computers and Economic

A4.2 Support Use of Internet for Employment

A4.3 Support Use of Internet for Employment and

A4.4 Respondents’ Internet Use for Employment

Support for Digital Democracy and E-Government 151

A5.2 Support for Digital Democracy vs. Actual Experience f52

A6.1 Digital Experience 153

A6.2 Support for Vouchers 154

6.1 Distribution of Digital Experience in the United States 126

T he issue of the “digital divide”—or disparities in information

to move “beyond the digital divide” as an issue of access and to pro¬

1. Stone 2002, 176.

A project such as this one—a book based on a national survey—

and we are grateful to be working with a publishing team of George¬

Redefining the Digital Divide

I nformation technology is now an integral part of the workplace

research evidence on skills. Moreover, little is known about the expe¬

Issue Evolution: The Access Divide and Beyond

Once the province of scientists, the Internet’s potential as mass

administration. Also during 1995, the first of several studies under¬

with a few notable exceptions, such as the Gates Foundation’s assis¬

dition to qualify as a policy issue rather than a personal concern, there

Likewise, there are positive externalities to be found in the politi¬