Download as pdf or txt
Download as pdf or txt
You are on page 1of 15

Journal of Information Technology & Politics

ISSN: 1933-1681 (Print) 1933-169X (Online) Journal homepage: https://www.tandfonline.com/loi/witp20

From the Digital Divide to the Democratic Divide:


Internet Skills, Political Interest, and the Second-
Level Digital Divide in Political Internet Use

Seong-Jae Min

To cite this article: Seong-Jae Min (2010) From the Digital Divide to the Democratic Divide:
Internet Skills, Political Interest, and the Second-Level Digital Divide in Political Internet Use,
Journal of Information Technology & Politics, 7:1, 22-35, DOI: 10.1080/19331680903109402

To link to this article: https://doi.org/10.1080/19331680903109402

Published online: 03 Feb 2010.

Submit your article to this journal

Article views: 9220

View related articles

Citing articles: 96 View citing articles

Full Terms & Conditions of access and use can be found at


https://www.tandfonline.com/action/journalInformation?journalCode=witp20
Journal of Information Technology & Politics, 7:22–35, 2010
Copyright © Taylor & Francis Group, LLC
ISSN: 1933-1681 print/1933-169X online
DOI: 10.1080/19331680903109402

From the Digital Divide to the Democratic Divide: Internet


WITP

Skills, Political Interest, and the Second-Level Digital


Divide in Political Internet Use
Seong-Jae Min
Min

ABSTRACT. Digital divide research is now focused on the so-called second-level divide, which
concerns Internet “usage” divides. This article suggests that while the first-level divide was associated
with sociodemographic factors, the second-level divide is associated with factors such as motivations
and Internet skills. It then illustrates an example of the second-level digital divide—the democratic
divide. The democratic divide concerns the differences between those who actively use the Web for
politics and those who do not. Analysis of General Social Survey data shows there is a democratic
divide where political Internet users are individuals with high Internet skills and political interest.

KEYWORDS. Democratic divide, digital divide, e-democracy, Internet politics, Internet skills,
second-level divide

There is a concern that information and com- disadvantages for the already marginalized
munication technologies (ICTs), which are groups in society. With the persistence of the
expected to contribute to the development of all digital divide, will the old, the poor, and other
humans, actually widen the inequalities social minorities who lack access to ICTs fall
between the developed world and the underde- behind in their ability to exploit the many
veloped world, the rich and poor, whites and opportunities brought about by the digital
blacks, the educated and less-educated, etc., revolution?
creating the so-called “digital divide” (Van The present study proposes that the digital
Dijk, 2005; Warschauer, 2003).1 Some studies divide will not likely disappear. It is argued that
have shown that access to ICTs is unequal the current Internet access divide will persist in
along lines of socioeconomic status, gender, the form of “usage” divides. One of the Internet
age, race, and geography (e.g., Mossberger, usage divides explored in this study is what
Tolbert, & Gilbert, 2006; National Telecommu- Norris (2001) called the “democratic divide,”
nications and Information Administration which concerns people’s differential usage of
[NTIA], 1995, 1999, 2000). The digital divide the Internet for political purposes. The demo-
raises an important social question, because cratic divide, if any, raises a critical social ques-
unequal access to ICTs may cause additional tion since it suggests that there may be

Seong-Jae Min is Assistant Professor of Communication Studies at Pace University in New York City.
His research focuses on political communication, new technologies, and deliberative democracy. This
research was supported by the Survey Research Fellowship program at The Ohio State University’s College
of Social and Behavioral Sciences, where the author received his Ph.D. degree.
Address correspondence to: Seong-Jae Min, Department of Communication Studies, 5th floor, 41 Park
Row, New York, NY 10038 (E-mail: sjminn@yahoo.com).

22
Min 23

politically marginalized people in the digital divide will eventually fade away as ICTs
world. If the Internet, which is much touted as a become more pervasive and inexpensive. How-
democracy-promoting medium, is mainly used ever, there are many reasons to believe that the
by a certain segment of the population, then its digital divide will persist or even widen in the
democratic potential will be greatly under- future. First, their analogy between the diffusion
mined. This study explores the existence of the of ICTs and the diffusion of other technologies,
democratic divide, and in so doing, it seeks to such as the automobile or telephone, fails.
identify what types of social and individual ICTs, especially computer and Internet technol-
factors cause the divide. ogies, are different from other technologies in
that they are much more complex, multifunc-
tional, and are considered to be “platform”
DIGITAL DIVIDE: IS THERE technologies for information and knowledge
AN ISSUE? (Van Dijk, 1999; Van Dijk & Hacker, 2003).
Unlike users of radios, television, and automo-
Despite the concern about the digital divide, biles, users of computers and the Internet must
some analysts deny either the existence or the actively upgrade the hardware, software, and
severity of the phenomenon (see, e.g., Compaine, individual skills to use them. ICTs become out-
1988, 2001; NTIA, 2004; Thierer, 2000). Their dated much faster than any other technologies,
main argument is that the digital divide will and users always have to catch up with the latest
eventually disappear, as have other technologi- technologies so as to not lag behind. For example,
cal divides in history. Compaine (1988, 2001) users must keep upgrading software and Inter-
argues that technological gaps or divides net connection speeds to remain productive.
among people would hardly occur, because, as Furthermore, the Internet is different from other
the world’s workforce becomes wealthier and technologies in terms of the amount of informa-
as technology costs decline, differences in all tion and knowledge it generates. Facing infor-
aspects of living standards decrease. Compaine mation and knowledge that increase every
cites the history of various technologies (e.g., second, Internet users are asked to constantly
electricity, radio, telephone, and automobiles) improve critical skills, such as the ability to
that were innovations in the past century and search, select, process, and apply information
have followed a similar developmental path. from a superabundance of sources (Van Dijk &
This path involved starting on a small scale at a Hacker, 2003, p. 316).
high price and use only by wealthier people. This suggests it is possible that those who
However, as the volume of use increased, the have been exposed to computers and the Internet
cost of providing the product decreased, allow- from the earliest stages are better situated in the
ing use of the new technology to diffuse rapidly current information society, because they have
through society, thus removing the gap. Like already gathered a lot of information and
Compaine, Thierer (2000) makes a claim that it knowledge and can use these tools to find addi-
is premature to act to defeat the digital divide, tional resources and to upgrade the skills
because society needs time to see how the tech- required maintain their productivity. In other
nology will adjust itself to the natural market. words, unlike users of traditional media, ICT
Thierer thus suggests that government involve- users will likely have much different levels of
ment in the digital divide issue will not be efficiency and experience in using the new
necessary; if any intervention is required, it technologies. This observation, in essence,
would be to remove tax and regulatory road- represents the heart of the knowledge gap
blocks that discourage companies in the free hypothesis (Tichenor, Donohue, & Olien, 1970)
market from offering consumers the new prod- and the “Sesame Street effect” (Cook, Appleton,
ucts and services that they demand. Conner, Shaffer, Tamkin, & Weber, 1975),
If one accepts the argument by Compaine which asserts that, even when everyone has
(1988, 2001) and Thierer (2000), then it is not equal access to media and technologies, the
necessary to worry about the digital divide. The information gap between the haves and have-nots
24 JOURNAL OF INFORMATION TECHNOLOGY & POLITICS

will not decrease because the haves typically divide to ICTs may eventually disappear.
make better use of media and technologies. For Indeed, as of 2008, the Internet penetration rate
example, in studying the Sesame Street effect, in the United States is as high as 74 percent
researchers showed that children with a higher (Pew Internet and American Life Project,
socioeconomic status achieve more educationally 2008), and the figure will continue to increase
than children with a lower socioeconomic to include an even larger population. However,
status, despite similar educational settings Compaine and Thierer fail to acknowledge that
(Attewell & Battle, 1999). Similarly, diffusion the digital divide is a much more complicated
theorists (Rogers, 1995; Ryan & Gross, 1943) and multifaceted phenomenon that goes beyond
showed that, without successful policy initia- the mere issue of access.
tives, the existing differences in socioeconomic Recent research has focused on what some
status may structure the diffusion of technological have termed the “second-level” digital divide
innovations. According to Rogers, early adopters (Hargittai, 2002a, 2002b), which is a divide that
of technological innovations, when compared concerns “multiple layers of access and use” of
with laggards, had a higher socioeconomic status. ICTs (Norris, 2001; Van Dijk, 1999). The
An important point of Rogers’ theory is that the results of such research suggests that individuals
adoption of successful new technologies often have diverse ways of accessing and using ICTs
reinforces the existing socioeconomic status; and that these multiple layers of access and use
thus, the rich get richer, and the poor remain are often determined by a variety of factors that
poor or even become poorer. include not only socioeconomic and demo-
While Compaine (1988, 2001) and Thierer graphic elements, but also physical, psychological,
(2000) predict that the diffusion of ICTs will cultural, and ecological factors. Hargittai
gradually become normalized over time, as emphasizes that what matters in the study of the
other technologies historically have, Rogers digital divide is the user’s skill level in using
(1995) argues that diffusion of innovations is ICTs. She demonstrated that there is a skill
often structured by the social environment sur- divide among Internet users, in that highly
rounding the technologies. At which end of the skilled users make better use of the Internet.
spectrum—between normal diffusion and struc- Others (Adams, Stubbs, & Woods, 2005; Stanley,
tured diffusion—will ICTs be placed? Consid- 2003) have focused on psychological variables
ering the intrinsic characteristics of ICTs—that of individuals and have demonstrated that there
they are complex, fast-evolving, and information are individual psychological differences moti-
generating—they are more likely to be placed vating the access to and use of ICTs. From an
at Rogers’ end. This notion has actually been ecological perspective, Ball-Rokeach’s Meta-
supported by some empirical research (Atkin & morphosis Project and its Internet Connected-
Jeffres, 1998; Lin, 1998). For instance, Atkin ness Index (ICI) also address the aspects of
and Jeffres showed that the early Internet multifaceted Internet use (Jung, Qiu, & Kim,
adopters’ profiles roughly matched with those 2001; Matei & Ball-Rokeach, 2003). The ICI
predicted by Rogers’ diffusion theory. attempts to capture the scope and centrality of
Internet incorporation into people’s everyday
lives, rather than identifying simple access
THE SECOND-LEVEL DIGITAL divide.
DIVIDE Selwyn (2004) suggests that researchers
should approach the issue of the digital divide
The argument by Compaine (1988, 2001) more comprehensively; accordingly, one
and Thierer (2000) stems from their conceptu- should consider the dimensions of “access,”
alizations of the digital divide solely in terms of “use,” and “consequences.” Access to ICTs
“access” to ICTs; under such a definition of the does not simply mean the binary distinction of
digital divide, they may be correct. As computers whether the population has ICT access or not.
and the Internet become more and more wide- The issue involves both the quality of access,
spread in society, it is possible that the access including speed, as well as the ease of access.
Min 25

In the use dimension, how people make use of 1993; Schwartz, 1996). Many studies (e.g., Hill
ICTs and what factors influence the different & Sen, 2005; Ott & Rosser, 2000) have shown
uses will have to be investigated. Finally, Selwyn how citizens use computers and the Internet for
argues that we need to examine the conse- enhanced political and democratic initiatives.
quences of engaging meaningfully with ICTs, For the so-called cyber pessimists, however, the
studying the impact of the use of ICTs on the Internet is a digital replica of the real world
various dimensions of citizens’ participation in where one observes politics as usual (Margolis
society. Similarly, DiMaggio and Hargittai & Resnick, 2000; Wilhelm, 2000).
(2001) suggest five dimensions along which the Norris (2001) adds to the discussion of Internet
digital divide may exist: “technical means” that politics by proposing the democratic divide
concern software, hardware, and connectivity hypothesis. She suggests that the democratic
quality; “autonomy of use” that concerns the divide signifies “the differences between those
location and quality of access; “use patterns” who do, and do not, use the panoply of digital
that concern the types of Internet use; “social resources to engage, mobilize, and participate
support networks” that concern the availability in public life” (p. 4). The existence of the dem-
of others one can turn to for assistance with use ocratic divide, to any extent, poses an important
of ICTs; and “skills” that involve one’s ability social question. Under the concept of the demo-
to use the medium effectively. These research- cratic divide, according to Norris, the Internet
ers provide a useful analytical framework, one mainly serves to reinforce the activism of the
that guides the present study to investigate the activists, facilitating participation for those who
complicated phenomenon of the digital divide are already interested in politics; whereas those
at the “second” level. who are disengaged from the politics of the real
world may further lag behind in the digital
world. A similar concern was raised, with
THE DEMOCRATIC DIVIDE empirical support, when Shelley, Thrane, and
Shulman (2006) investigated generational dif-
Following the theoretical framework devel- ferences in Internet use. They argued:
oped by recent digital divide researchers, as
described above, this study attempts to look at By permitting some citizens to conduct
the multifaceted aspects of the digital divide their routine business with the govern-
beyond the simple issue of access. Of particular ment more easily, information technology
interest is the probing of individuals’ different appears to be widening the gap between
usage patterns of the Internet, because as the the IT literate and those without basic
Internet spreads widely, what matters is not navigation skills. As society becomes
access to the Internet, but how people actually increasingly dependent on e-government,
use it. In particular, the study focuses on indi- social barriers will be compounded if
viduals’ “political” use of the Internet due to non-electronic voices are marginalized
the medium’s ever-increasing importance for from political participation. (p. 48)
political participation.
One area where the Internet is bringing new As such, the issue of the democratic divide pro-
kinds of social interaction is the realm of poli- vides an important caveat for the future of
tics. The Internet at present is characterized as e-democracy.
being, among other things, multimodal, interac- The democratic divide explained here is a
tive, horizontal, low-cost, and nonterritorial. rather broad concept that concerns many
These characteristics may provide high hopes aspects of digital civic life. As mentioned in the
for the future of democracy. Hence, Internet previous section that Selwyn (2004) suggests,
enthusiasts have argued that the Internet can the democratic divide may occur along the lines
contribute to democracy by bonding people, of ICT access, use, and consequence. In the
regardless of territory, and by creating public present study, however, the democratic divide
spheres and new social movements (Rheingold, will be defined operationally as “the divide
26 JOURNAL OF INFORMATION TECHNOLOGY & POLITICS

pertaining to individuals’ differential use of the skill levels and knowledge of the Internet, it
Internet for politics.” This is because the “use” would be difficult for citizens to engage in such
dimension is perhaps the most noticeable aspect digital political initiatives as online donation,
of the democratic divide that easily warrants signing electronic petitions, and discussing pol-
empirical investigation, and the Internet is the itics online. Internet skills have several dimen-
representative medium of all ICTs today. sions and can be measured by behavioral and
self-reported measurements (Hargittai, 2005).
These include knowledge of and experience
SKILLS, MOTIVATIONS, AND with the medium, practical ability, and confi-
POLITICAL INTERNET USE dence in using it. The level of Internet skills
may be related to schooling, because people
If there is a divide in individuals’ Internet with higher levels of education are likely to
use for politics, what would drive such a have had more exposure to the Internet for
divide? Conventional studies typically associate educational purposes. Yet formal schooling
socioeconomic and demographic factors with does not necessarily guarantee good Internet
the divide. Previous empirical studies have skills, because Internet skills also include many
shown that those who are active in Internet practical Internet abilities used in everyday life.
civic engagements are usually people with Therefore, good Internet skills, independent of
higher socioeconomic status (Hill & Hughes, level of education, may serve as a predictor of
1998; Norris, 1999). In those studies, basic online political actions.
demographic and socioeconomic factors such Motivations, or interest, are another important
as gender, age, educational level, income, and factor that may predict citizens’ online political
race were considered to be major determinants engagements. As Norris (2001) points out, the
of differential Internet use for politics. Specifi- Internet is a medium of par excellence and a la
cally, the studies demonstrated that males, carte. That is, Internet users tend to select and
whites, and those with higher education and customize information for their own interests.
income are more likely to be active in using the People determine which emails they respond to
Internet for political affairs. and which listservs they join. In this way,
In addition to basic demographic and socio- according to Norris, those who have high political
economic factors, however, factors such as indi- interests and motivation are more likely to
viduals’ Internet skills and political motivations engage in public activities that are available
should matter as well. Indeed, skills and moti- online, thereby enhancing the scope of their
vations may be the two most important factors political influence. This argument is in line with
that would explain individuals’ differential Inter- the famous uses and gratifications approach in
net use for politics. As the Internet becomes media studies, which avers that people use
more and more widespread in society and as new media in ways that will satisfy their needs and
political opportunities available online increase, interests. Due to its high potential for customi-
the importance of Internet skills and political zation, the Internet provides many good oppor-
interest will matter even more, if one wants to tunities to satisfy specific individual interests.
use the Internet meaningfully for political pur- Katz and Rice’s (2002) extensive empirical
poses. At the same time, we may observe a study suggests that people use the Internet
decrease in the importance of socioeconomic differently so that it fits with the important
and demographic factors as the Internet is more aspects of their lives.
and more evenly accessed and used across the Thus, Internet skills and political interest are
population. key factors that may cause the democratic
Internet skills, or Internet literacy, have divide, especially when it concerns the political
received some attention in recent years. As Har- use of the Internet. In a classic study of traditional
gittai (2002b, 2005) argues, individuals’ online political participation such as voting and com-
behavior is, in part, a reflection of their online munity engagements, Verba, Schlozman, and
skills. This makes sense because without relevant Brady (1995) showed that capacity and motivation
Min 27

matter in determining the level of participation. Kim, Koch, & Park, 2006). Although it is a few
This observation may hold true in the online years old, the 2004 GSS contains extensive
environment as well: not just the simple existence questions on Americans’ political behavior,
of new technology, but human interest and attitudes, and Internet use. In particular, the
skills, will lead to meaningful political partici- 2004 GSS contains a special Information Society
pation using the technology. Therefore, in addition topical module that allows detailed assessment
to basic demographic and socioeconomic fac- of the respondents’ Internet skills and political
tors, this study analyzes the role of political Internet use patterns. Because of these advan-
interest and Internet skills in the use of the tages, the 2004 GSS was adopted for in-depth
Internet for politics. empirical investigation.
The GSS is a face-to-face, full-probability
sample survey of adults living in households in
HYPOTHESES AND RESEARCH the United States. The 2004 GSS had a
QUESTIONS response rate of 70.4 percent out of a 3,628 per-
son net sample. For the analysis of Internet use
This section formulates some hypotheses and divide, only the Information Society module
research questions based on the rationale pre- was selected. This yielded a subsample of 684
sented thus far. First, as suggested in the adults who said they used the Internet.
review, it is likely that there is a democratic
divide in cyberspace such that individuals Measurements
differentially use the Internet for politics. Who
uses the Internet for politics and who does not? Political Internet Use
This question should be addressed first. Therefore,
This was the dependent variable, which
operationally represented the concept of the
RQ1: What are the profiles of Internet users
democratic divide. Political Internet use con-
for politics?
sisted of two dimensions: online political infor-
Second, it has been argued that Internet skills mation seeking and online political discussion.
and political interest are key factors that predict Online political information seeking was a
differential Internet use for politics, because dichotomous variable, where those who said
political Internet use requires a certain level of they visited a political Web site in the past month
Internet skills and political motivations. Therefore, were coded 1 and those who said they had not
were coded 0 (variable POL30 in the GSS ques-
H1: Internet skills will be positively associ- tionnaire. For actual wording, see Appendix).
ated with political Internet use. Online political discussion (INTERPOL) was also
H2: Political interest will be positively asso- a dichotomous variable, where those who joined
ciated with political Internet use. online political discussion were re-coded 1.

Internet Skills
METHOD
The GSS dataset had comprehensive measures
The Dataset of Internet skills. These included self-reported
ability to use the Web, knowledge of Internet
The main goal of this study is to look at the terms, and practical knowledge of how to
existence of the democratic divide in terms of exchange files using the Internet. A composite,
Internet use among individuals. To this end, the general Internet skills index was constructed
2004 General Social Survey (GSS) dataset was considering the three dimensions above: self-
analyzed. The GSS is perhaps one of the most reported Web ability, knowledge of Internet
authoritative datasets in social science survey terms, and practical skill. Self-reported Web
research today and has been considered repre- ability in the dataset (WEBABLE) was reverse-
sentative and methodologically rigorous (Smith, coded so that it could range from “very poor” to
28 JOURNAL OF INFORMATION TECHNOLOGY & POLITICS

“excellent” (M = 3.88, SD = .98 on a five-point on a 23-point level, ranging from less than
scale of 1 to 5). The knowledge score variable $1,000 to more than $110,000. The median
was built from the three questions (ADVSRCH/ income fell between $50,000 and $59,999.
PREFSETS/NEWSGRPS) that measured the Income was also used as a continuous variable.
respondents’ familiarity with Internet terms. Age was used as a continuous variable (M =
Respondents who said they were either “very 42.69, SD = 14.48). Race was reorganized into
familiar” or “somewhat familiar” with the three a dichotomous variable: Whites and Asian
terms “Newsgroups,” “Advanced Search,” and Americans were coded 0 and African Ameri-
“Preference Settings” received three points. cans, Hispanics, and other races were coded 1.
Respondents received two points for familiarity The rationale for this rather unorthodox coding
with two terms, one point for one term, and 0 was that previous research shows that whites
points for no familiarity with any of the terms and Asian Americans tend to have a high level
(M = 2.13, SD = 1.03 on a four-point scale of 0 of Internet access and use, whereas African
to 3). Practical skill was measured by the variable Americans’ and Hispanics’ access and use fall
(UPLOAD/DOWNLOAD), which asked much behind those of whites and Asian Ameri-
whether the respondents knew how to upload cans (NTIA, 1995, 1999, 2000). In other words,
and download files using the Internet. Respon- whites and Asians can be regarded as a major-
dents received two points if they knew how to ity, whereas other races can be considered a
both upload and download files, one point if they minority in terms of Internet use.
knew how to either upload or download, and 0
points for no practical skill of file exchange (M = Analytic Strategy
1.54, SD = .76 on a three-point scale of 0 to 2).
Each of the three skill scores—self-reported skill, To probe individuals’ differential Internet
knowledge, and practical skill—was standardized use for politics, first of all, the profiles of politi-
and then added together to make a composite cal versus nonpolitical Internet users were
index (Cronbach’s a = .81, M = 7.55, SD = 2.37 descriptively analyzed based on socioeco-
on a ten-point scale of 1 to 10). nomic and demographic factors. Second, to
make a statistical inference to the larger Ameri-
Political Interest can Internet population, a logistic regression
analysis was performed. Here the dependent
The 2004 GSS had a standard political interest variables were two: online political information
item (POLINT1) that asked the respondents how seeking and online political discussion. The
much they were interested in politics. The independent variables were socioeconomic and
responses were reverse-coded so that a higher demographic variables, Internet skills, and
number meant higher interest (1 meant “not inter- political interest.
ested at all” and 4 meant “very interested.” M = Lastly, a structural equation model was con-
2.86, SD = .82 on a four-point scale of 1 to 4). structed to further assess the effects among the
Socioeconomic and Demographic variables. One of the advantages of using struc-
Variables tural equation modeling here was that it could
increase the reliability of the measurement. For
Previous research (e.g., Norris, 1999) sug- example, the Internet skills variable in this
gests that individuals’ Internet use for politics study had three dimensions, and it would have
may be different based on gender, income, edu- been difficult to find out in conventional regres-
cation, age, and race. Therefore, these variables sion analysis whether these three would reliably
were used as independent variables. Gender represent the Internet skills variable. In the
was coded with female being 0 and male being structural equation model, however, the three
1 (57.7 percent females). Education, measured were not simply added together, but were
as the highest year of school completed, was regarded as multiple indicators of a latent vari-
used as a continuous variable (M = 14.56, SD = able. This also held true to the political Internet
2.57). Total household income was measured use variable, which had two dimensions. This
Min 29

type of analysis allows for assessment of the nonpolitical users, political users of the Internet
coefficients between the indicators and their in this sample tended to be male, highly edu-
corresponding latent variable with measure- cated, and have high income: there were more
ment errors, which can contribute to the reli- females (63.4 percent) than males (36.6 per-
ability of the measurement. cent) among nonpolitical Internet users; of all
In the structural equation model, socioeco- political Internet users, 49.0 percent had at least
nomic and demographic variables, Internet college degrees, whereas only 33.2 percent of
skills, and political interest were used as exoge- nonpolitical Internet users had college or higher
nous variables. The Internet skills variable was degrees; of political Internet users, 38.0 percent
constructed as a latent exogenous variable that had a family income of $75,000 or higher,
was measured by the three observed variables whereas only 27.1 percent of nonpolitical Inter-
(dimensions) of self-reported skill, knowledge, net users had an income of $75,000 or higher.
and practical skill. Political interest, gender, Pearson chi-square tests suggest that the differ-
race, income, education, and age each involved ences in political Internet use were statistically
a single observed variable. Political Internet use significant along the lines of gender (c2 =
was the latent endogenous variable measured 11.28, df = 1, p < .01), education (c2 = 25.98,
by the two observed variables of online politi- df = 4, p < .001), and income (c2 = 12.90, df = 2,
cal information-seeking and online political p < .01).
discussion. To make a statistical inference to the larger
Some of the observed variables in the struc- American Internet population, a binary logistic
tural equation model were ordinal or categori- regression analysis was performed for political
cal in nature. Methodologists suggest correction Internet users in the GSS sample. In a series of
for estimation involving ordinal or categorical binary logistic regression analyses, the indepen-
variables in structural equation modeling. This dent variables were regressed upon the depen-
is because, for categorical and ordinal variables, dent variables, online political information
the assumptions of multivariate normality are seeking and online political discussion. Table 2
not met when using distribution-dependent summarizes the two logistic regression results.
methods, such as the maximum likelihood According to Table 2, the respondents’ gender,
(ML) estimator (Mueller, 1997). Therefore, income, political interest, and Internet skills
methodologists recommend using the weighted were statistically significant predictors of
least squares (WLS) method, which is based on online political information-seeking; that is,
the polychoric correlation matrix for estimation males with high income, high political interest,
involving ordinal and categorical variables and good skills were associated with greater
(Bollen, 1989). Following their suggestion, probability of using the Internet for political
polychoric correlations of the variables were information. For the predictors of online politi-
entered into LISREL 8.8, and the parameters cal discussion, only Internet skills and political
were estimated using the WLS method. interest were statistically significant; that is,
each one unit increase in Internet skills and/or
political interest was associated with greater
RESULTS probability of using the Internet for political
discussion. Hypotheses 2 and 3 were supported.
The first research question concerned prob- The structural equation model with path
ing the profiles of political and nonpolitical coefficients and standard errors is presented in
Internet users. Of the 684 Internet users in the Figure 1. This model also supported the hypoth-
GSS dataset, 295 (around 43 percent) were eses and corroborated the findings of the logis-
political Internet users. These were the people tic regression analysis. According to the model,
who said they had ever visited a political Web Internet skills (g = .63, p < .001, unstandardized
site or joined a political discussion group. The coefficient) and political interest (g = .42, p <
cross-tabulation results in the third and fourth .001, unstandardized coefficient) were strong
columns of Table 1 show that compared with predictors of Internet use for politics. Furthermore,
30 JOURNAL OF INFORMATION TECHNOLOGY & POLITICS

TABLE 1. Demographic and Socioeconomic Profiles of Internet Users, Political


Internet Users, and Nonpolitical Internet Users (%): General Social Survey, 2004

Total Internet users Political Internet users Nonpolitical Internet users


(n = 684) (n = 295) (n = 383)

Gender**
Male 42.3 49.3 36.6
Female 57.7 50.7 63.4
Race
White 81.5 82.5 80.7
Black 11.4 10.6 12.1
Asian 2.9 3.3 2.6
Hispanic 2.6 1.7 3.4
Age
18–29 21.4 21.9 21.1
30–49 46.7 47.0 46.4
50–64 24.2 24.5 24.0
Over 65 7.6 6.6 8.4
Education***
Less than high school 6.0 3.9 7.6
High school graduates 21.3 15.1 26.3
Some college 32.5 31.9 32.9
College graduates 21.8 28.3 16.6
Advanced degrees 18.4 20.7 16.6
Income**
Low (Under $40,000) 36.9 29.7 42.8
Middle ($40,000–74,999) 31.1 32.3 30.1
High (Over $75,000) 32.0 38.0 27.1

*p < .05.
**p < .01.
***p < .001.

as shown in Figure 1, the three skill dimensions and those who do not in terms of political inter-
of self-reported skill, knowledge, and practical est and Internet skill levels. Differences in
skill were highly correlated with the latent political Internet use in terms of gender,
Internet skills variable (all ps < .001), suggest- income, and education were observed in some
ing that the latent skills variable was a valid of the analyses as well. The gender difference
construct. The common fit indices suggest that was salient, in particular. These results resonate
the model fit the data very well: the chi-square with previous studies’ findings that citizens
index, which is used to test lack of fit of the who are active in digital civic engagements are
model, was not significant (c2 = 33.17, df = 22, usually highly sophisticated, male, middle-
p >.05); the root mean square error of approxi- class, and young professionals (Davis & Owen,
mation (RMSEA) was .03, which is considered 1998; Hill & Hughes, 1998; Norris, 1999).
an excellent fit; both the adjusted goodness of In the present study, however, Internet skills
fit index (AGFI) and the incremental fit index and political interest, relative to sociodemo-
(IFI) were .99, which indicate a very good fit. graphic factors, turned out to be particularly
strong predictors of political Internet use. They
were significant consistently in all models.
DISCUSSION Internet skills, in particular, had substantial
effects in all models, even after controlling for
The findings of this study support an Internet the level of education. This suggests that, in the
usage divide with regard to political affairs. democratic divide, which is one example of the
They show that there are clear differences second-level digital divide, what matters more
between people who visit political Web sites are individuals’ skills and political attitudes,
Min 31

FIGURE 1. Structural equation model of political Internet use.


32 JOURNAL OF INFORMATION TECHNOLOGY & POLITICS

TABLE 2. Logistic Regression Results dimensions, and perhaps studying the “conse-
Predicting Online Political Information-Seeking quences” of Internet use for politics is as impor-
and Online Political Discussion tant as differential use itself. How does
differential Internet use for politics translate
Online political Online political into meaningful differences in real-world poli-
information–seeking discussion tics? This remains an important question to be
investigated. Third, the data used in this study
Slopeb SEb EXPb Slopeb SEb EXPb
were collected in 2004, and there may be
Age −.01 .01 .99 .00 .01 1.00 important differences between the Internet
Education .03 .04 1.03 .02 .04 1.03 landscape of today and then. Although the
Gender .40* .19 1.50 .31 .18 1.37
Income .05* .02 1.05 −.01 .02 .99
dataset is considered high-quality, it only
Race .07 .27 1.07 .19 .25 1.21 addressed two dimensions of digital politics:
Internet skill .23*** .05 1.26 .15*** .04 1.17 online political information-seeking and
Political .93*** .13 2.53 .70*** .12 2.01 online political discussion. Over the last few
interest
years, however, new political opportunities
N = 606. using the Internet such as blogging and online
*p < .05. citizen journalism became popular. Yet we
**p < .01.
***p < .001.
know very little about who is actively using
Model chi-square for online political info seeking = 128.35 these “Web 2.0” initiatives. Are these new
(df = 7). Nagelkerke’s R2 = .26. political initiatives dominated by politically
Model chi-square for online political discussion = 71.95 sophisticated people as well? Does participat-
(df = 7). Nagelkerke’s R2 = .15.
ing in the new initiatives require high techno-
logical skills? Further analysis is needed in
rather than socioeconomic or demographic fac- this matter.
tors. This resonates with recent second-level
digital divide studies that emphasize such fac-
tors as Internet efficacy and psychological atti- CONCLUSION
tudes (Adams et al., 2005; Eastin & LaRose,
2000). Indeed, for meaningful use of the Inter- This study attempted to portray individuals’
net for politics, Internet skills and political use of the Internet for politics. In so doing, it
motivations would be vital. found that Internet use for politics is not equal
The findings of this study, however, should depending on skills and motivational factors.
be interpreted with caution, as there are some This may serve as a warning against the techno-
limitations. First, due to the inherent weakness logical deterministic view that technologies
of the cross-sectional dataset, the findings could will bring a democratic utopia. Instead, what
not establish causality. For example, it is this study suggests is that the simple availabil-
unclear whether different Internet skills and ity of the new technology is not enough to
political interest caused differential use of the encourage the meaningful use of technology for
Internet for politics, or whether differential use politics. Human interest and capacity are
of the Internet affected skills and interest. equally important.
Nonetheless, the study shows that Internet use Overcoming the digital divide is not an easy
for politics is associated with social and indi- task, because in attempting to overcome one
vidual divisions, which is still a meaningful divide, we observe the emergence of a new
finding to understand the current political land- usage divide. If only a certain segment of the
scape on the Internet. Second, the present study population uses the Internet for politics, as sug-
is also limited in that it did not address some gested in this study, the democratic potential of
other dimensions of the democratic divide. In the Internet will be undermined. Therefore, the
the study, the democratic divide was operation- issue of the democratic divide warrants much
alized solely as differential Internet use for poli- attention. One solution to this problem is to
tics. However, the democratic divide has multiple build citizens’ digital literacy or capacity.
Min 33

Along with universal access to the Internet, Hargittai, E. (2002a). Beyond logs and surveys: In-depth
ongoing civic education on ICTs and their ben- measures of people’s Web use skills. Journal of the
eficial use is essential in the current informa- American Society for Information Science and Tech-
nology Perspectives, 53(14), 1239–1244.
tion-based society. Implementing such a Hargittai, E. (2002b). Second-level digital divide: Differ-
program will require the full cooperation and ences in people’s online skills. First Monday, 7(4),
commitments of all parties involved—the civil Retrieved Sep. 7, 2008, from http://firstmonday.org/
society, government, and business. issues/issue7_4/hargittai/index.html.
Hargittai, E. (2005). Survey measures of Web-oriented
digital literacy. Social Science Computer Review,
NOTES 23(3), 371–379.
Hill, D., & Sen, K. (2005). The Internet in Indonesia’s
1. The General Social Survey dataset used in this new democracy. New York: Routledge.
study and relevant documentations are publicly available Hill, K., & Hughes, J. (1998). Cyberpolitics: Citizen activ-
from the National Opinion Research Center (NORC, http:// ism in the age of the Internet. London: Rowman &
www.norc.org) or the Inter-University Consortium for Littlefield.
Political and Social Research (ICPSR, http://www.icpsr. Jung, J. Y., Qiu, J. L., & Kim, Y. C. (2001). Internet
umich.edu). connectedness and inequality: Beyond the divide.
Communication Research, 28(4), 507–535.
Katz, J., & Rice, R. (2002). Social consequences of Inter-
net use: Access, involvement, and interaction. London:
REFERENCES MIT Press.
Lin, C. A. (1998). Exploring the personal computer adop-
Adams, N., Stubbs, V., & Woods, V. (2005). Psychologi- tion dynamics. Journal of Broadcasting & Electronic
cal barriers to Internet usage among older adults in the Media, 41(1), 95–112.
UK. Medical Informatics and the Internet in Medicine, Margolis, M., & Resnick, D. (2000). Politics as usual: The
30(1), 3–17. cyber space ‘revolution.’ London: Sage.
Atkin, D., & Jeffres, L. (1998). Understanding Internet Matei, S., & Ball-Rokeach, S. (2003). The Internet in the
adoption as telecommunications behavior. Journal of communication infrastructure of urban residential
Broadcasting & Electronic Media, 42(4), 475–490. communities: Macro or meso-linkage? Journal of
Attewell, P., & Battle, J. (1999). Home computers and Communication, 53(4), 642–657.
school performance. The Information Society, 15(1), Mossberger, K., Tolbert, C., & Gilbert, M. (2006). Race,
1–10. place, and information technology. Urban Affairs
Bollen, K. (1989). Structural equations with latent vari- Review, 41(5), 583–620.
ables. New York: John Wiley & Sons. Mueller, R. (1997). Structural equation modeling: Back to
Compaine, B. (1988). Information gaps: Myth or reality? basics. Structural Equation Modeling 4(4), 353–369.
In B. Compaine (Ed.), Issues in new information tech- National Telecommunications and Information Adminis-
nology (pp. 179–191). Norwood, NJ: Ablex Publishing tration. (1995). Falling through the net: A survey of the
Company. “have-nots” in rural and urban America. Retrieved
Compaine, B. (2001). Epilogue. In B. Compaine (Ed.), Aug. 7, 2008, from http://www.ntia.doc.gov/ntiahome/
The digital divide: Facing a crisis or creating a myth? fallingthru.html.
(pp. 337–339). Cambridge, MA: MIT Press. National Telecommunications and Information Adminis-
Cook, T., Appleton, H., Conner, R., Shaffer, A., Tamkin, tration. (1999). Falling through the net: Defining the
G., & Weber, S. (1975). Sesame Street revisited. New digital divide. Retrieved Aug. 10, 2008, from http://
York: Russell Sage Foundation. www.ntia.doc.gov/ntiahome/fttn99.
Davis, R., & Owen, D. (1998). New media and American National Telecommunications and Information Adminis-
politics. New York: Oxford University Press. tration. (2000). Falling through the net: Toward digital
DiMaggio, P., & Hargittai, E. (2001). From the digital inclusion. Retrieved Aug. 10, 2008, from http://
divide to digital inequality: Studying Internet use as www.ntia.doc.gov/ntiahome/fttn00/contents00.html.
penetration increases. Center for Arts and Cultural National Telecommunications and Information Adminis-
Policy Studies Working Paper. Princeton, NJ: Prince- tration. (2004). A nation online: Entering the broad-
ton University. band age. Retrieved Aug. 10, 2008, from http://www.
Eastin, M. S., & LaRose, R. (2000). Internet self-efficacy and ntia.doc.gov/reports/anol2004/ Nation OnlineBroadb
the psychology of the digital divide. Journal of Computer- and04.htm.
Mediated Communication, 6. Retrieved Nov. 22, 2008, Norris, P. (1999). Who surfs? New technology, old voters,
from http://jcmc.indiana.edu/vol6/issue1/eastin.html. and virtual democracy. In K. Elaine & J. Nye, Jr. (Eds.),
34 JOURNAL OF INFORMATION TECHNOLOGY & POLITICS

Democracy.com? Governance in a networked world Smith, T., Kim, J., Koch, A., & Park, A. (2006). Social
(pp. 71–94). Hollis, NH: Hollis Publishing. science research and General Social Surveys. Compar-
Norris, P. (2001). Digital divide: Civic engagement, infor- ative Sociology, 5(1), 33–43.
mation poverty, and the Internet worldwide. New York: Stanley, L. D. (2003). Beyond access: Psychological barriers
Cambridge University Press. to computer literacy. The Information Society, 19(5),
Ott, D., & Rosser, M. (2000). The electronic republic? The 407–416.
role of the Internet in promoting democracy in Africa. Thierer, A. (2000, April 20). How free computers are filling
Democratization, 7(1), 138–156. the digital divide. The Heritage Foundation Backgrounder,
Pew Internet and American Life Project (2008, Dec). no. 1361. Washington, DC: Heritage Foundation.
Trend data. Retrieved Mar 11, 2009, from http:// Tichenor, P., Donohue, G., & Olien, C. (1970). Mass
pewinternet.org/trends.asp. media flow and differential growth in knowledge.
Rheingold, H. (1993). The virtual community: Homestead- Public Opinion Quarterly, 34(2), 159–170.
ing on the electronic frontier. New York: Addison- Van Dijk, J. (1999). The network society: Social aspects of
Wesley. new media. London: Sage.
Rogers, E. (1995). Diffusion of innovations. 4th ed. New Van Dijk, J. (2005). The deepening divide: Inequality in
York: Free Press. the information society. London: Sage.
Ryan, B., & Gross, N. (1943). The diffusion of hybrid Van Dijk, J., & Hacker, K. (2003). The digital divide as a
seed corn in two Iowa communities. Rural Sociology, complex and dynamic phenomenon. The Information
8(1), 15–24. Society, 19(4), 315–326.
Schwartz, E. (1996). Netactivism: How citizens use the Verba, S., Schlozman, K., & Brady, H. (1995). Voice and
Internet. Sebastopol, CA: Songline Studios. equality: Civic voluntarism in American politics.
Selwyn, N. (2004). Reconsidering political and popular Cambridge, MA: Harvard University Press.
understandings of the digital divide. New Media & Warschauer, M. (2003). Technology and social inclusion:
Society, 6(3), 341–362. Rethinking the digital divide. Cambridge, MA: MIT
Shelley, M. C., Thrane, L. E., & Shulman, S. W. (2006). Press.
Generational differences in information technology use Wilhelm, A. (2000). Democracy in the digital age: Chal-
and political involvement. International Journal of lenges to political life in cyberspace. New York: Rout-
Electronic Government Research, 2(1), 36–53. ledge.

APPENDIX

Major Questions Used from the 2004 General Social Survey (Variable names are in
brackets)

[POL30]
772–G. In the past 30 days, how often have you visited a web site for
G) Political information
Never 1–2 times 3–5times More than 5 times
[WEBABLE]
812. Would rate your ability to use the World Wide Web as excellent, good, fair, poor or very
poor?
Excellent Good Fair Poor Very Poor
[DOWNLOAD]
791-C. Do you know how to download a file from the World Wide Web to your computer?
Yes No
[UPLOAD]
791-D. Do you know how to send a file that is on your computer’s hard drive to someone using
another computer?
Yes No
Min 35

[ADVSRCH/PREFSETS/NEWSGRPS]
815-E, H, I. Are you very familiar, somewhat familiar, or not familiar with the following Inter-
net Terms:
E) Advanced Search
H) Preference settings
I) Newsgroups
Very familiar Somewhat familiar Not Familiar
[INTERPOL]
1468-H. Here are some different forms of political and social action that people can take. Please
indicate, for each one, whether you have done any of these things in the past year, whether
you have done it in the more distant past, whether you have not done it but might do it, or
have not done it and would never, under any circumstances, do it.
“Joined an Internet political forum or discussion group.”
[POLINT1]
1474. How interested would you say you personally are in politics?
Very interested Fairly interested Not very intereste Not at all Interested

You might also like