The relative costs and expertise associated with using the Internet, labeled technological and social access, have led to a concern about the rise of a "digital divide" between information haves and have-nots. To address whether and to what extent the Internet has become a medium of the masses and to identify the factors associated with social access to the Internet, I examine Internet use data from two statewide surveys, the Carolina Poll and the Indiana Poll, conducted during spring 1998. Multivariate analysis reveals that income, education, age, and family structure are important social determinants of on-line access and that Internet use is lowest among single mothers, members of lower socioeconomic groups, and older respondents. Although the online population is beginning to diversify, the Internet cannot yet claim a committed, nonelite mass audience. It is argued that the disparities in Internet use portend a looming information gap between those with access and those without.
Advocates of cyberpolitics, with growing frequency and force, have applauded the democratizing potential of the Internet, noting its capacity to facilitate political discussion and on-line activism as well as its ability to distribute news and information to a rapidly growing mass audience (Grossman 1995; Hill and Hughes 1998; Rash 1997; Schwartz 1996). With each passing election cycle, more political campaigns, media organizations, incumbent officeholders, party organizations, nonprofit civic groups, and governmental agencies go on-line, allowing increased access to the political system (Demchak et al. 1997; Faucheux 1998). The relative expertise and costs associated with using the Internet have led to a concern, however, about the rise of a "digital divide" between information haves and have-nots based on education, income, and other demographic characteristics (National Telecommunications and Information Administration [NTIA] 1998). Rather than engendering broad democratic participation and equalizing information differences between citizens at opposite ends of the socioeconomic spectrum, the Internet may instead exacerbate existing knowledge disparities and serve to further exclude disadvantaged groups from the political process.
Though concern about a widening knowledge gap is not unique to on-line media and has surfaced when other communication technologies and program genres were new (Tichenor et al. 1970), the Internet raises questions not only about information benefits, but also about social and technological access to the technology (Kling 1999). Technological access refers to the physical availability of computer hardware and software, and social access refers to the mix of professional knowledge, economic resources, and technical skills required for effectual use of information and communication technologies (Kling 1999). As more governmental functions become available on-line (Demchak et al. 1997) and as more political campaigns are waged in cyberspace (Margolis et al. 1997), issues relating to both forms of access take on increasing urgency. Indeed, the civic opportunities and technical challenges posed by digital convergence demand that citizens learn new skills and adapt to the changing information environment or risk being left behind.
Fed by the amazing growth of the World Wide Web since its introduction in the early 1990s, the transition to digital information technology has been rapid: The Internet is now growing at a faster rate than any previous new electronic medium (Berthon et al. 1996). In the United States, approximately 41 percent of adults aged eighteen and older were on-line as of the third quarter of 1998--almost twice as many as in 1996 (Pew Research Center 1999). Although the number of Internet users continues to grow, a majority of the public still does not have technological access, especially minorities, the poor, and single-mother households (NTIA 1998). Computer owners are much more likely to be on-line than the public at large, as are younger, more educated, upper-income professionals (NTIA 1998). If current trends continue and the online audience "goes ordinary," as recent surveys by the Pew Research Center for the People and the Press suggest, this may change. Increasingly, women, nondegree holders, and modest income earners are joining or planning to join the ranks of Internet users (IntelliQuest 1999; Pew Research Center 1999).
Whether the Internet becomes a true medium of the masses and follows a normal diffusion curve, with later adopters consisting of less sophisticated users (Rogers 1995), remains to be seen. Because on-line information sources require a certain level of cognitive ability or "Internet literacy" to navigate successfully (Hofstetter 1998), there is reason to believe that the digital divide will not be completely remedied through universal physical access to computer technology alone. As longitudinal studies have shown, the Internet is a demanding medium even for users with substantial on-line experience; when less sophisticated users encounter buggy software and unfamiliar terminology, procedures, and features, or when they experience technical difficulties that require help desk support, they may back away from on-line media altogether (Kiesler et al. 1997). Hence over time, social access to the Internet may be a far better predictor of adoption than the availability of suitable equipment.
Research Questions
To address whether and to what extent the Internet has become a medium of the masses and to identify the factors associated with social access to the Internet, this study asks the following research questions:
RQ1: How widespread is Internet use, and how does it compare with the consumption of other public affairs media, including television news, newspapers, and talk radio?
RQ2: Do members of higher socioeconomic groups have greater social access to the Internet than members of lower socioeconomic groups?
RQ3: How do other demographic, political, and media-use factors influence Internet use?
Method
Data for this study were collected through two statewide telephone surveys: the spring 1998 Carolina Poll and the spring 1998 Indiana Poll. The first, a random survey of 696 North Carolina residents, was conducted by the School of Journalism and Mass Communication and the Institute for Research in Social Science at the University of North Carolina, Chapel Hill, between March 21 and 29, 1998. The second survey, which randomly interviewed 521 Indiana residents, was conducted by the Indiana University Center for Survey Research between April 9 and June 7, 1998. Both surveys used a scientifically selected sample of adult residents obtained through a random-digit-dialing procedure. The response rate for the Indiana Poll was approximately 47 percent, while the response rate for the Carolina Poll was 48 percent.
Results
Both samples consisted of more women than men, a ratio of three to two in each state. Respondents ranged in age from eighteen to ninety-one, with an average age of forty-six. The vast majority of survey respondents were white: 79.4 percent in the Carolina Poll and 89.2 percent in the Indiana Poll. The next largest racial or ethnic group was African-Americans, comprising 14.9 percent and 6.2 percent of respondents in the respective surveys. The average education level for both samples was some college, but the mean annual income level was slightly higher in the North Carolina sample ($30,000 to $40,000) than in the Indiana sample ($25,000 to $35,000). Overall, respondents' political orientations grouped just to the right of the ideological center (2.25 among North Carolina respondents and 2.2 among Indiana respondents on a three-point scale where 1 is liberal and 3 is conservative).
Internet Use
Just over a third of North Carolina residents (34 percent) and over a quarter of Indiana residents (26 percent) surveyed reported logging on to the Internet or the World Wide Web one or more days per week.(n1) Among those with access, 53 percent of the North Carolina sample reported heavy Internet use, logging on four to seven days a week, and 47 percent reported light to medium use, using the Internet one to three days a week. In the Indiana sample, the figures were reversed: 44 percent reported heavy use of the Internet for news and information, and 56 percent reported light to medium use.
Other Public Affairs Media Use
Although hailed for its democratizing potential, the Internet is still establishing its mass appeal. Compared to use of other public affairs media, including newspapers and local and national television news, use of the Internet is relatively low. In the Indiana sample, 87.1 percent reported watching local television news, 77.2 percent reported watching national television news, and 78.7 percent reported reading a daily newspaper one to seven days in the last week. Talk radio use was much lower, with only 29.8 percent of respondents reporting any talk-radio listening. In the North Carolina sample, 98.9 percent reported television viewing, and 98.7 percent said they read a newspaper one or more days in the last week. (No talk-radio question was asked in the Carolina Poll.) As Figure 1 shows, respondents in both samples devoted more days per week to television and newspapers than to the Internet.
Social Access to the Internet
The social access argument holds that know-how, arising from a combination of education, training, and economic resources, is vital to utilizing information technologies in ways that enhance professional practices and social life (Kling 1999). Though it might appear that physical access is the main impediment to expanded Internet use, other factors, particularly computer and print media literacy, could be equally influential (see Corrado and Firestone 1996). In its current form, the Internet and networked computing are simply too difficult to benefit many people (Kiesler et al. 1997). This has important implications for citizenship because as the knowledge-gap literature has shown, when the amount of news about public affairs increases (as it is now with the Internet), people with higher socioeconomic status tend to acquire this information at a faster rate than those with lower status (see Viswanath and Finnegan 1996). To assess the relationships between social factors and Internet use, a variety of inferential statistics were employed. Table 1 summarizes the differences between light and heavy Internet use in the two samples based on various social and demographic characteristics.
In the North Carolina sample, analysis of variance revealed that Internet use varied by income level. Similar differences were found in the Indiana sample between income groups with respect to use of the Internet for news and information. Respondents with an annual income of $50,000 or higher used the Internet substantially more in both North Carolina and Indiana than respondents in either the middle ($25,000 to $49,999) or lower ($24,999 and under) income category. Figure 2 shows average weekly Internet use by income group.
A similar pattern emerges between college degree holders and nondegreed respondents. College-educated North Carolinians reported significantly more days of Internet use per week (M = 2.2) than their nondegreed counterparts (M = 0.96), as indicated by an independent samples t-test. Among Indiana residents, the differences between graduates (M = 2.2) and nongraduates (M = .59) were even more pronounced. Correlations also showed consistently positive (though modest) associations between Internet use and income (Tau bs =. 20 and. 23, ps < .01), as well as Internet use and education (Tau bs = .28 and .30, ps < .01). These positive associations are consistent with the social-access argument.
Age was negatively correlated with Internet use in both samples (Tau bs = -.18 and -.19, ps < .01), suggesting that older respondents use the Internet less. Analysis of variance of Internet use by age group indeed showed this to be the case in both North Carolina and Indiana, with mean weekly access rates revealing a downward trend from the under thirties to the over sixties. The finding that Internet use is higher among younger groups does not indicate that on-line access will decline as people age. Rather, this result indicates a possible cohort effect; as younger users grow older, their on-line activity may remain comparatively high. Figure 3 illustrates average weekly Internet use by age group.
Gender Differences
Although both samples consisted of fewer men than women, male respondents reported significantly more Internet use than female respondents. The differences between genders were most pronounced in the Indiana sample, where men (M = 1.3) reported almost twice as much daily Internet use per week as women (M = .70). In the North Carolina sample, the differences were not significant. A crosstabulation of Internet access by gender similarly showed that significantly fewer women (31 percent in North Carolina and 22 percent in Indiana) reported using the Internet than men (38 percent in North Carolina and 32 percent in Indiana).(n2) Among heavy users--those on the Internet four to seven days a week--men outnumbered women by a margin of more than two to one in Indiana (17 percent to 8 percent), slightly less so in North Carolina (22 percent to 15 percent).
The gender disparities among heavy Internet use widened when controlling for income, with significant differences emerging in the Indiana sample between men and women at the lower and higher income levels. Only 2 percent of lower-income women reported heavy Internet use, compared to 15 percent of lower-income men. At the higher income level, just 8 percent of women reported heavy Internet use, compared to 23 percent of men. There were no significant gender differences for Internet use at the middle income level ($25,000 to $49,999).
Other Demographic, Political, and Media-Use Factors
The data also provided insight into other demographic and media-consumption factors affecting Internet use. Perhaps the most interesting demographic factor was family structure, including the number of people in the respondent's household, whether the respondent is a parent, and whether the respondent has a live-in partner. Analysis of variance showed that Internet use increases with household size in both samples (from. 84 and 1.13 days per week for two-person households to 1.5 and 1.84 days per week for six-person households), and these differences significantly varied in both samples. As might be expected, the number of people in the respondent's household was positively correlated with the number of children (rs = .66 and .88, ps < .01). In the Indiana data, parents reported more Internet use than nonparents. Among light users (one to three days a week), 20 percent of parents reported Internet use, compared to just 11 percent of nonparents.
Similar to the national NTIA findings, single women in the Indiana sample (those without a live-in partner), especially single mothers, were much less likely to use the Internet, averaging a half a day or less per week, than single men or single fathers, who averaged 1.6 and 2.5 days of use per week. This pattern was similar in the North Carolina data but not as marked.
Race was another salient influence. In the North Carolina sample, white respondents (M = 1.46) reported significantly more Internet use than African-Americans (M = .75). When analyzed categorically, more white respondents reported being on-line, and more reported heavy Internet use (20 percent) than African-Americans (11 percent). Among light Internet users, however, the same percentage of African-Americans and whites (16 percent) reported using the Internet one to three days a week. In the Indiana sample, there were no significant differences for Internet use and race, perhaps because the number of minorities sampled was too small (10 percent).
Politically, there was a main effect for party identification on Internet use in both North Carolina and Indiana. As Figure 4 shows, declared Republicans in both samples reported the most on-line use of any party-affiliated group (an average of 1.5 days per week), followed by Independents/Other (an average of 1.2 days per week), and Democrats (an average of .92 days per week).
Among traditional media-use items, local television news in the Indiana sample revealed a weak negative correlation with Internet use (r = -.13, p < .01), suggesting that as television news viewing increases, on-line activity decreases. By contrast, talk radio was positively associated with on-line activity (r = .08, p < . 10), though the relationship only approached significance. Further analysis showed that local news viewing was significantly higher among older, lower-income respondents without a college degree--the opposite demographic traits of active Internet users. In the North Carolina survey, newspaper reading was positively, but again weakly, associated with Internet use (r =. 07, p <. 10).
Finally, multiple linear regression was used to determine if the effects of the demographic and media-use variables would persist while each was held constant. Table 2 shows that in the Indiana data, the negative relationship between local news and Internet use not only persisted, but also actually strengthened (Beta = -.15, p < .01), as did the positive association with talk-radio listening (Beta = .11, p < .05). The most influential demographic items in both samples were education, age, income, and number of people in the household.
Conclusions
Consistent with the NTIA national survey of forty-eight thousand U.S. residents, this state-level analysis found that socioeconomic status, demographic characteristics, and family structure are important social determinants of online access. This finding suggests two things. First, although the on-line population is beginning to diversify, the Internet is not yet a medium of the masses. Second, given the disparities in on-line access between people of higher and lower socioeconomic groups, there may be a looming information gap between the best-positioned members of society and the rest. Beyond the physical hardware needed to go on-line, social access to the Internet requires that citizens have the cognitive ability and technical skills necessary to profit from a complex media environment. As an information resource, the Internet places considerably different demands on the user than the reigning medium of television. Before the new technology can claim a committed, nonelite mass audience, questions of social access will have to be addressed.
Notes
(n1.) The Internet question in the Indiana Poll asked about news and informational use of the Internet or World Wide Web, whereas the Carolina Poll asked about Internet use generally.
(n2.) The Internet-use variable was transformed from a continuous scale (zero to seven days) to a three-level categorical variable (none, one to three days, four to seven days).
Legend for Chart: B - Indiana None C - Indiana 1-3 Days D - Indiana 4-7 Days E - North Carolina None F - North Carolina 1-3 Days G - North Carolina 4-7 Days A B C D E F G GENDER Men 68.1 15.0 16.9 61.9 16.0 22.0 Women 78.6 13.7 7.7 69.2 15.7 15.0 RACE White 74.4 14.3 11.3 64.5 15.7 19.8 Minority 74.1 13.0 13.0 73.6 15.7 10.7 AGE 18-29 63.0 19.0 18.0 52.2 24.6 23.2 30-39 70.3 15.8 13.9 61.1 19.1 19.8 40-49 66.1 21.1 12.8 59.0 20.5 20.5 50-59 78.2 12.6 9.2 70.3 9.3 20.3 60+ 91.8 4.1 4.1 89.3 3.6 7.1 INCOME[a] To $24,999 88.6 4.5 6.8 76.7 12.1 11.2 $25,000-$49,999 77.0 13.7 9.3 74.6 11.6 13.9 Over $50,000 60.0 24.1 15.9 47.5 26.5 26.0 EDUCATION College Degree 47.1 25.2 27.7 44.3 26.0 29.8 No Degree 82.5 11.0 6.5 77.3 10.8 11.9 FAMILY Children 67.4 20.5 12.1 64.2 18.3 17.5 No Children 77.7 11.2 11.2 67.4 14.3 18.3 PARTY IDENTIFICATION Democrat 77.4 14.2 8.4 70.1 18.3 11.6 Republican 65.7 17.9 16.4 62.5 16.5 21.0 Independents/Other 76.8 12.1 11.1 65.5 13.6 20.9 [a] In the North Carolina data, the lowest group included incomes up to $29,999; the middle group ranged from $30,000 to $49,999.
Legend for Chart: A - Variable B - Indiana (n = 334) B(SE) C - Indiana (n = 334) Beta D - North Carolina (n = 543) B(SE) E - North Carolina (n = 543) Beta A B C D E Age -.03 (.01) -.21[***] -.02 (.01) -.13[*] Education .21 (.04) .25[***] .42 (.08) .29[***] Income .09 (.05) .07[A] .18 (.08) .12[*] Number of children -.22 (.16) -.08 -.08 (.17) -.05 Number in household .23 (.12) .12[*] .07 (.16) .04 TV news viewing .02 (.05) .02 -- Newspaper reading .07 (.04) .08[A] -.04 (.04) -.06 Local news viewing -- -.12 (.05) -.15[**] National news viewing -- .09 (.05) .12[A] Talk radio listening -- .12 (.06) .11[*] R[sup 2] .14 .18 Adjusted R[sup 2] .13 .15 B = regression coefficient, Beta-standardized regression coefficient. [A] p <. 10. [*] p < .05. [**] p < .01. [***] p < .001.
GRAPH: Figure I; Public Affairs Media Use Note: Local news and national television news viewing were averaged in the Indiana data. Talk radio listening was not measured in the North Carolina survey.
GRAPH: Figure 2; Internet Use by Income Category
GRAPH: Figure 3; Internet Use by Age Group
GRAPH: Figure 4; Internet Use by Political Party Affiliation
References
Berthon, Pierre, Leyland Pitt, and Richard T. Watson. 1996. "The World Wide Web as an Advertising Medium: Toward an Understanding of Conversion Efficiency." Journal of Advertising Research 36(1):43-54.
Corrado, Anthony, and Charles M. Firestone, eds. 1996. Elections in Cyberspace: Toward a New Era in American Politics. Washington, D. C.: Aspen Institute.
Demchak, Chris C., Christian Friis, and Todd M. LaPorte. 1997. "Governance in an Information Age: Early Patterns of Global Diffusion of the Web and Openness across Public Agencies." Nov. http://www.cyprg.arizona.edu/CyGovN97.htm.
Faucheux, Ron. 1998. "How Campaigns Are Using the Internet: An Exclusive Nationwide Survey." Campaigns and Elections 19(8): 22-25.
Grossman, Lawrence K. 1995. The Electronic Republic. New York: Penguin.
Hill, Kevin A., and John E. Hughes. 1998. Cyberpolitics: Citizen Activism in the Age of the Internet. Lanham, MD: Rowman and Littlefield.
Hofstetter, Fred T. 1998. Internet Literacy. Boston: Irwin/McGraw-Hill.
IntelliQuest. 1999. "IntelliQuest Internet Study Shows 100 Million Adults Online in 2000." Mar. 3. http://www.Intelliquest.com/press/release72.asp.
Kiesler, Sara, Robert Kraut, Tridas Mukhopadhyay, and William Scherlis. 1997. June. "HomeNet Overview: Recent Results from a Field Trial of Residential Internet Use." June. http://homenet.andrew.cmu.edu/progress/ovrview8697.html.
Kling, Rob. 1999. "What Is Social Informatics and Why Does It Matter?" D-Lib Magazine 5 (1). http://www.dlib.org/dlib/january99/kling/01kling.html.
Margolis, Michael, David Resnick, and Chin-chang Tu. 1997. "Campaigning on the Internet: Parties and Candidates on the World Wide Web in the 1996 Primary Season." Press/Politics 2(1):59-78.
National Telecommunications and Information Administration. 1998. "Falling through the Net II: New Data on the Digital Divide? July 28. http://www.ntia.doc.gov/ntiahome/net2/ falling.html.
Pew Research Center for the People and the Press. 1999. "The Internet News Audience Goes Ordinary: Online Newcomers More Middle-Brow, Less Work-Oriented." Jan. 16. http://www.people-press.org/tech98sum.htm.
Rash, Wayne Jr. 1997. Politics on the Nets: Wiring the Political Process. New York: W.H. Freeman.
Rogers, Everett M. 1995. Diffusion of Innovations. 4th Edition. New York: Free Press. Schwartz, Ed. 1996. NetActivism: How Citizens Use the Internet. Sebastopol, CA: Songline Studios.
Tichenor, Phillip J., George A. Donohue, and Clarice N. Olien. 1970. "Mass Media Flow and Differential Growth in Knowledge." Public Opinion Quarterly 34(2):159-70.
Viswanath, K., and John R. Finnegan, Jr. 1996. "The Knowledge Gap Hypothesis: Twenty-five Years Later." Communication Yearbook 19:187-227.
Paper submitted January 19, 1999; accepted for publication March 17, 1999.
~~~~~~~~
By Erik P. Bucy
=Erik P. Bucy is an assistant professor in the Department of Telecommunications at Indiana University, Bloomington. He served as deputy press secretary and national scheduler for Jerry Brown's 1992 presidential campaign and previously worked as a staff writer for the Los Angeles Herald Examiner. Address: Department of Telecommunications, 327 Radio-TV Center, Indiana University, Bloomington, IN 47405-6901; phone: 812-856-5207; fax: 812-855-7955; e-mail: ebucy@indiana.edu.