Journal of Broadcasting & Electronic Media/June 2006 Featured Topic: The Internet

Connections Between Internet Use and Political Efficacy, Knowledge, and Participation Kate Kenski and Natalie Jomini Stroud Using data from the 2000 National Annenberg Election Survey, this study looks at the relationships between Internet access and online exposure to information about the presidential campaign and political efficacy, knowledge, and participation. Results show that Internet access and online exposure to information about the presidential campaign are significantly associated with these important political variables. Several of the associations between Internet access and exposure with political efficacy, knowledge, and participation are detectable even when taking sociodemographic variables, party identification, partisan strength, political interest, and other media exposures variables into account. Although statistically significant, these associations are quite small. Over the past decades, the amount of available political information has expanded, thanks in part to the Internet. Political candidates have been using the Internet to update individuals through e-mail (Bimber, 1998), to provide information about their issue positions (Stromer-Galley, 2000), and to raise money (Dulio, Goff, & Thurber, 1999). Not only do the campaigns supply information, a host of other Web sites devoted to politics—ranging from news organizations to nonprofits to individual blogs—are available for interested Internet users. Information availability, however, does not guarantee that the Internet will change how users approach politics. Although some scholars fear that the Internet may cause people to abandon their social environment (Nie & Erbring, 2000), other researchers suggest that electronic media have the potential to strengthen social relationships (Hampton & Wellman, 2003; Wellman & Hampton, 1999) and to enhance democracy and increase participation (Anderson, 2003). Given the Internet’s relative infancy as a communication medium, many questions remain about how the Internet affects individuals and society. Using data from the Kate Kenski (Ph.D., University of Pennsylvania) is an Assistant Professor in the Department of Communication at the University of Arizona. Her research interests include political communication, public opinion, gender and politics, and research methods. Natalie (Talia) Jomini Stroud (Ph.D., University of Pennsylvania) is an Assistant Professor in the Department of Communication Studies at the University of Texas at Austin. Her research interests include political communication, media effects, and selective exposure to communication. © 2006 Broadcast Education Association

Journal of Broadcasting & Electronic Media 50(2), 2006, pp. 173–192

173

174

Journal of Broadcasting & Electronic Media/June 2006

2000 National Annenberg Election Survey (NAES), this study looks at the relationships between Internet access and online exposure to information about the presidential campaign and political efficacy, knowledge, and participation. It has been argued that these important political variables are indicators of a properly functioning democracy. A politically knowledgeable electorate is desirable in order for citizens to make informed decisions (Delli Carpini & Keeter, 1996). Political efficacy is a determinant of political behavior—without feelings of competency and beliefs that one’s actions are consequential, one has little incentive to participate in politics (Abramson & Aldrich, 1982). Political participation is important because levels of engagement in a democratic system have consequences on the system’s equity (Rosenstone & Hansen, 1993). Theoretically, an optimal democracy would contain citizens who possess high levels of political knowledge, efficacy, and participation. Results from the 2000 NAES show that Internet access and online exposure to information about the presidential campaign are significantly and positively associated with these important political variables. Several significant relationships are detectable even when taking sociodemographic variables, party identification, partisan strength, political interest, and several other media exposures variables into account. Although detectable, the magnitudes of the associations are small, suggesting that the Internet’s potential to improve these indicants of good citizenship is limited. Beginning in 1996, the Internet emerged as a major nontraditional medium used in political campaigns (Johnson, Braima, & Sothirajah, 1999). Not only were candidates maintaining campaign Web sites but nonprofit organizations were experimenting with informing the public about issues using the Internet. Today, an Internet presence is considered the norm for political campaigns, and the number of Web sites with political information has dramatically increased. As the amount of political content online has increased, so has the percentage of Americans accessing political information online. Looking at online political behavior, the Pew Internet and American Life Project found that 18% of Americans said that they went online for election news in 2000 (Kohut et al., 2000). In 2004, 29% of Americans reported using the Internet for political news (Rainie, Cornfield, & Horrigan, 2005).

Political Efficacy, Knowledge, and Participation Given the number of people who report accessing political information online, questions remain about the effects of Internet use on political variables such as efficacy, knowledge, and participation. In The Voter Decides, Campbell, Gurin, and Miller (1954) described the concept of political efficacy as “the feeling that political and social change is possible, and that the individual citizen can play a part in bringing about this change” (p. 187). Several scholars have noted that political efficacy appears to be made of two different constructs, a personal sense of efficacy, commonly known as internal efficacy, and a more system-oriented sense of efficacy, known as external efficacy (Balch, 1974; Converse, 1972; Niemi, Craig, & Mattei, 1991). Niemi

Kenski and Stroud/INTERNET USE AND POLITICS 175

et al. explain that internal political efficacy refers “to beliefs about one’s own competence to understand, and to participate effectively in, politics” (p. 1407), whereas external political efficacy refers “to beliefs about the responsiveness of government authorities and institutions to citizen demands” (p. 1408). Although political efficacy is frequently discussed as an overall concept, internal and external efficacy items typically fall into two unique dimensions, and these dimensions interact differently with other political variables (Balch, 1974). More widely studied than efficacy, political participation and knowledge need little description. Political participation involves taking part in activities related to politics such as donating to a campaign or attempting to convince others how to vote. Prior investigations of political participation have demonstrated its strong association with a number of demographic variables, including education, income, and age (Rosenstone & Hansen, 1993). Political knowledge, as defined by Delli Carpini and Keeter (1996), is “the range of factual information about politics that is stored in long-term memory” (p. 10). Political knowledge can be gained through formal education (Delli Carpini & Keeter, 1996; Nie, Junn, & Stehlik-Barry, 1996), interpersonal discussion about politics (Delli Carpini & Keeter, 1996), and traditional news media consumption, such as newspaper reading (Chaffee, Zhao, & Leshner, 1994; Scheufele & Nisbet, 2002). Research suggests that efficacy, knowledge, and participation are interrelated concepts (McLeod, Scheufele, & Moy, 1999). For example, internal efficacy is positively associated with education (and hence political knowledge) and participation (Morrell, 2003). Further, external and internal efficacy are related, although not identical, constructs (Niemi et al., 1991).

Citizenship and the Internet With the onslaught of political information online, scholars have put forward several hypotheses about the relationship between the Internet and these important political variables. Theories of the relationship between the Internet and political efficacy, knowledge, and participation have developed along two main schools of thought— those contending that the Internet has a negative or null effect and those arguing that Internet exposure increases political efficacy, knowledge, and participation. Optimistically, the Internet has the potential to increase efficacy, knowledge, and participation. The Internet could enhance external political efficacy because it enables citizens to interact with public officials and to hold them accountable. The Internet also gives individuals easy access to information about politics as many Web sites are designed with the objective of informing citizens about candidates, issues, and politics in general. In this regard, the Internet also increases internal efficacy by providing accessible information to citizens. Cornfield (2003) suggests that the Internet may increase internal efficacy by making people less embarrassed about their political competence; he writes, “the anonymity [the Internet] offers may assuage the fear of public embarrassment” (p. 106). The Internet also provides new and less costly

176

Journal of Broadcasting & Electronic Media/June 2006

ways in which an individual can participate in politics, for example, through e-mailing a candidate or donating to a campaign. In terms of political knowledge, the Internet provides volumes of information to citizens with several benefits over other media. The Internet not only allows for information seeking at any time, it allows users to dig deeper into issues through the use of hyperlinks and search engines. More pessimistically, the Internet could bring about a decline in political efficacy, knowledge, and participation. Individuals attempting to contact their representative may be discouraged from receiving either a form letter response or no response at all, which may decrease external efficacy (Johnson & Kaye, 2003). Further, there are many complications to finding accurate and accessible political information online. The sheer volume of political information online (e.g., try typing “presidential candidates” into a Google search) may be overwhelming and thus lead to lower levels of personal confidence in one’s ability to understand the political world. These difficulties may be particularly pronounced for individuals already feeling less efficacious because these individuals tend to be less educated and less familiar with the Internet. Another important concern is that the Internet will only exacerbate the divide between the “tuned-in and the tuned-out, the activists and the disengaged” (Norris, 2001, p. 13). According to some accounts, individuals who access political information on the Internet are likely those who are already interested in politics; the Internet thus does not increase political involvement (Graber, 1996; Norris, 2001). Because the Internet is a medium in which individuals largely determine their exposure to content, individuals not interested in politics will not seek out politics online, and therefore, current levels of political knowledge and participation are likely to be maintained. In this sense, the Internet may replicate current knowledge gaps if those more able and likely to find political information are those who were already more knowledgeable (Delli Carpini & Keeter, 2003). Consistent with the variety of theoretical views, empirical findings about the influence of the Internet on political efficacy, knowledge, and participation have been mixed. Pinkleton, Austin, and Fortman (1998) found that active media use was a significant predictor of external self-efficacy when controlling for education and cynicism. Prior studies have also found that the Internet contributes to increases in political participation (Weber, Loumakis, & Bergman, 2003). In a survey of Web users, Quan-Haase, Wellman, Witte, and Hampton (2002) found that “the Internet supplements political activities but does not change people’s level of involvement” (p. 312). Further, studies have suggested that the Internet may expand the number of participating individuals and that it may erase well-known participation divides (e.g., those with lower incomes participate less) as access increases (Krueger, 2002). Horrigan, Garrett, and Resnick (2004) found that after controlling for other demographic and political variables, Internet use was positively related to the number of arguments that people reported having heard about the presidential candidates. This finding suggests that the Internet may be positively related to political knowledge. Other studies, however, have not found positive relationships between the Internet and efficacy, participation, and knowledge. When controlling for other variables, Lin

Kenski and Stroud/INTERNET USE AND POLITICS 177

and Lim (2002) did not find that the Internet was positively related to efficacy. Among their sample of politically interested Web users, Johnson and Kaye (2003) did not find a significant correlation between reliance on the Web or hours per week on Web seeking political information and external self-efficacy.1 Scheufele and Nisbet (2002) maintain, “The role of the Internet in promoting active and informed citizenship is minimal. In fact, respondents who used the Internet frequently for entertainment purposes were less likely to feel efficacious about their potential role in the democratic process” (p. 69). A recent study by Jennings and Zeitner (2003) did not find that Internet use for public affairs contributed to either political efficacy or knowledge. Several other studies have also found that Internet use does not contribute to candidate knowledge above other traditional media (DiMaggio, Hargittai, Neuman, & Robinson, 2001; Johnson et al., 1999). Further, research has suggested that the Internet does not reduce cynicism (Kaid, 2003). Because cynicism is known to be correlated with external efficacy (Niemi et al., 1991), this result is not suggestive of a positive relationship between efficacy and the Internet. Several methodological issues in these prior studies may explain the inconsistent findings and may suggest ways to reconcile these differences in future research. Looking at the relationship between the Internet and political efficacy, several prior studies have not distinguished between internal and external political efficacy and have instead combined these items into a scale (Lin & Lim, 2002; Scheufele & Nisbet, 2002).2 This makes it difficult to sort out the influence of the Internet on these two distinct notions of political efficacy. Further, attending to the Internet for entertainment purposes may negatively contribute to efficacy, knowledge, and participation (Scheufele & Nisbet, 2002), whereas attending to the Internet for information-seeking purposes may result in increases in these variables. Shah, Kwak, and Holbert (2001), for example, found that Internet use for information seeking was positively related to engagement, trust, and contentment, whereas use of the Internet for social recreation was not. Not every study, however, distinguishes between the different uses of the Internet. In addition, one of the inherent difficulties in studying the relationship between the Internet and political involvement is that Internet users searching for political information typically report high levels of internal efficacy (Muhlberger, 2002), external efficacy (Johnson & Kaye, 2003), and political knowledge (Delli Carpini & Keeter, 2003). It is difficult to know, therefore, whether the Internet caused these changes or if individuals already involved in politics were attracted to the Internet. To ward against spurious interpretations of the relationship between the Internet and political efficacy, knowledge, and participation, a diverse range of controls is necessary in survey studies of the Internet’s impact on political variables. External validity concerns also can be raised. In a theoretically thorough but geographically limited investigation of the effects of Internet use on efficacy, knowledge, and participation, Scheufele and Nisbet found political information seeking on the Web was not significantly associated with the dependent variables of interest but noted that the local nature of the sample may limit the study’s “geographical generalizability” (p. 68). Further, failure to detect associations with Internet use in prior studies may have been the result of insufficient sample sizes.

178

Journal of Broadcasting & Electronic Media/June 2006

With the previously reviewed findings and limitations in mind, this study uses the data from the 2000 NAES to test the effects of Internet access and online exposure to information about the presidential campaign on the two distinct dimensions of political efficacy, knowledge, and participation with a sample generalizable to the U.S. adult population. This study hypothesized that both access and exposure would be positively associated with efficacy, knowledge, and participation. More formally, the hypotheses are as follows: H1a: Internet access is positively associated with internal political efficacy. H1b: Online exposure to information about the presidential campaign is positively associated with internal political efficacy. H2a: Internet access is positively associated with external political efficacy. H2b: Online exposure to information about the presidential campaign is positively associated with external political efficacy. H3a: Internet access is positively associated with political knowledge. H3b: Online exposure to information about the presidential campaign is positively associated with political knowledge. H4a: Internet access is positively associated with political participation. H4b: Online exposure to information about the presidential campaign is positively associated with political participation.

The advantages of using the NAES data are fourfold. First, the data were collected from adults in the U.S. population through random-digit dialing, thus avoiding the generalizability problems that arise from studies based on college students, small regional samples, or convenience samples. Second, the survey asked specifically about going to the Internet for political information about the presidential campaign, rather than using more general measures that conflate using the Internet for entertainment purposes with political information seeking. Third, the survey’s comprehensive battery of media exposure questions allowed control for exposure to network news, cable news, local news, newspaper, and political talk radio. Fourth, the large sample size allowed finding small and moderate effects of Internet access and online exposure to information about the presidential campaign.

Methods The analyses in this study use data from the 2000 NAES (for a copy of the data, see Romer, Kenski, Waldman, Adasiewicz, & Jamieson, 2004). The Annenberg Public Policy Center at the University of Pennsylvania conducted a year-long rolling cross-sectional (RCS) survey, the largest academic survey on political attitudes and behavior ever conducted of the U.S. population (Kenski, 2003a). The NAES began November 8, 1999, and ended January 19, 2001. Telephone interviews were conducted with adults in the United States each day. Random subsamples of randomly sampled telephone numbers were released for interviewing each night to maximize

Kenski and Stroud/INTERNET USE AND POLITICS 179

the representativeness of each daily cross-section.3 Once a telephone number was released into the field, up to 18 callbacks were made over a 2-week period. The response rate (RR1) for the national sample was 25.3%. The period of interest for the current study was the general election campaign defined in the analyses as July 15 through November 6, 2000.4 This time period was selected for two reasons. Initially, prior research has suggested that people are most engaged in the political process, as exemplified by Internet exposure to presidential campaign information and talking about politics with family and coworkers, during the general election (Jamieson & Kenski, 2004). Therefore, if the Internet was to have a significant impact on political efficacy, knowledge, and participation, the general election is the most likely time for such an impact to occur. The second reason is practical, as the sample size stabilized at approximately 300 respondents per day beginning on July 15.

Dependent Variables Political Efficacy. External efficacy was measured with the statement, “People like me have no say over who gets to be president,” whereas internal efficacy was measured with the statement, “Sometimes presidential elections seem so complicated that a person like me cannot really understand what is going on.” Respondents were asked if they 1 (strongly agreed), 2 (somewhat agreed), 3 (somewhat disagreed), or 4 (strongly disagreed) with each of these statements.5 Disagreement with the statements indicated that the respondent was efficacious. Political Knowledge. Survey participants were asked questions about the major party presidential candidates and their issue positions. Each question named one of the candidates, George W. Bush or Al Gore, and asked the respondent whether that candidate favored or opposed a given issue position. Five questions were asked about George W. Bush, and the same five questions were asked about Al Gore. The issue position questions pertained to abortion, handgun license requirements, patients’ right to sue their health maintenance organizations, government funds to give all children health insurance, and Social Security contributions in the stock market. Individuals could either give a substantive response or say that they did not know the answer to a question. When a respondent said that they did not know the answer, they were encouraged to give a substantive response. If the respondents did not give a substantive response, their responses were recoded so as to give them a response that they would have obtained had they randomly guessed, in accordance with recommendations from Mondak (2001) and Mondak and Anderson (2004).6 Then, a 10-point political knowledge scale was created by summing the number of correct responses. Political Participation. Respondents were asked whether they had tried to convince others how to vote, had attended a meeting for a candidate, had done work for a candidate, had given money to a candidate, or had worn a button or shown a sign

180

Journal of Broadcasting & Electronic Media/June 2006

supporting a candidate. “Yes” responses to these items were added together to create a 5-point participation index.

Independent Variables Eighteen independent variables were included in this project. As noted in the review of the literature, these variables have known relationships to the dependent variables under consideration. Although Internet access and online campaign exposure were the focus of this study, the other independent variables were used as controls in the multivariate analyses.

Sociodemographics. The sociodemographic variables included gender (female = 1, male = 0), age (years), Black (Black = 1, else = 0), Hispanic (Hispanic = 1, else = 0), education (years of schooling), and household income (in thousands). Political Identification and Attitudes. A party identification question was used to create two dummy variables for Democrats (Democrat = 1, else = 0) and Republicans (Republican = 1, else = 0). Strength of partisanship (strong = 1, not strong = 0) was also incorporated into the models. Although it is common in the political science literature to combine responses to the party identification question with partisanship to create a 7-point party identification scale, the authors have opted to treat these variables as separate constructs. The justifications are twofold. First, it was suspected that identification with a major party makes one more attentive and receptive to communication, thus increasing one’s relationship with the dependent variables under consideration. Consequently, respondents at both ends of the traditional 7-point should report higher levels of efficacy, knowledge, and participation than those in the middle, making the traditional scale unsuitable for this study. Second, prior research has suggested that “independent leaners are almost as highly sophisticated and active as the strong party identifiers” (Neuman, 1986, p. 98). Unlike many studies that simply assume that Independents are either suppressed partisans or political apathetics and thus do not ask them whether they consider themselves to be “strong” Independents, the strength of partisanship question was asked of Independents as well as major party identifiers.7 Political interest was assessed by responses to the following question: Some people seem to follow what is going on in government and public affairs most of the time, whether there is an election or not. Others are not that interested. Would you say you follow what is going on in government and public affairs most of the time, some of the time, only now and then, or hardly at all?

These categories were coded as 1 (hardly at all), 2 (only now and then), 3 (some of the time), or 4 (most of the time). Individuals were also asked how many days in the past week they had talked with family and friends about politics (0 to 7 days).

Kenski and Stroud/INTERNET USE AND POLITICS 181

Media Access and Exposure. Respondents were asked whether or not they had cable or satellite television at home. They were asked how many days in the past week they had used five different types of traditional media: network news, cable news, local news, newspaper, and talk radio. Answers ranged from 0 to 7 days per week.8 The key Internet variables were access and exposure. Survey participants were asked: “Do you have access to the Internet or the World Wide Web at home, at work or someplace else?” If survey participants said that they had Internet access, they were asked: “How many days in the past week did you see information about the campaign for president online?”9

Analytical Procedures Because political efficacy, knowledge, and participation have been shown to be correlated with several demographic, political orientation, and media variables, this study tested the hypotheses using hierarchical multiple regression in which the independent variables were entered into the model in blocks. The authors ran two models for each of the four dependent variables. The first was for the total sample and used Internet access as the independent variable of interest. The second was only for those with Internet access and used online exposure to information about the presidential campaign as the independent variable of interest. The final block entered into each model was either the Internet access variable or the online exposure variable, allowing determination of how much the Internet variables contributed to the prediction of the dependent variables after the other independent variables had been taken into consideration.

Results An overview of the sample composition is provided in Table 1. Well over a majority of respondents reported that they had Internet access (67.1%). Of those who had Internet access, survey participants on average had seen political information about the presidential campaign online 1.61 days in the past week (Mdn = 0, SD = 2.49). Many Internet access respondents (62.2%) had not been exposed to presidential campaign information on the Internet in the past week. Turning attention to the dependent variables, when asked about their agreement with the statement “Sometimes presidential elections seem so complicated that a person like me cannot really understand what is going on,” 19.6% strongly agreed, 31.4% somewhat agreed, 23.0% somewhat disagreed, and 26.0% strongly disagreed. When asked about the statement “People like me have no say over who gets to be president,” 17.5% of respondents strongly agreed with this statement, 25.4% somewhat agreed, 26.3% somewhat disagreed, and 30.8% strongly disagreed. On the 10-point political knowledge scale, respondents answered correctly 6.42 items on

182

Journal of Broadcasting & Electronic Media/June 2006

Table 1 Univariate Descriptions of Sample

Female Age Hispanic Black Education (in years) Household income (in thousands) Party identification Republican Democrat Independent Other Strength of partisanship Strong Not strong Political interest Most of the time Some of the time Now and then Hardly at all Cable or satellite television access Internet access Information exposure in days Talk politics with friends/family Watching network news Watching cable news Watching local news Reading a newspaper Listening to political talk radio Seen political information about the presidential campaign online (among those with access)

N

%

35,316 34,997 35,065 34,804 35,118 31,578 33,908

55.6

M

SD

Mdn

45.88

16.54

44

13.97 54.42

2.49 43.84

14 42.5

2.16 3.14 2.57 4.17 3.79 1.51 1.61

2.27 2.66 2.72 2.67 2.87 2.31 2.49

2 3 2 5 4 0 0

7.8 8.7

28.5 32.0 30.4 9.2 30,142 54.8 45.2 35,068

35,176 35,273 29,627 35,153 35,169 35,139 35,252 335,227 23,562

35.1 35.0 18.5 11.3 80.3 67.1

average (Mdn = 6, SD = 1.91). And finally, on the 5-point political participation scale, the mean score was a low 0.36 (Mdn = 0, SD = 0.72). Before progressing to multivariate analysis, bivariate analysis confirmed that Internet access and online exposure to campaign information were related to the dependent variables. Relying solely on the bivariate relationships, however, was unsatisfactory because there were several potential alternative explanations for the observed associations. For example, one could argue that a causal interpretation of an observed relationship between Internet access and internal efficacy is spurious

Kenski and Stroud/INTERNET USE AND POLITICS 183

because education leads not only to higher levels of Internet access but also to higher levels of internal efficacy. Similar explanations can be advanced for a variety of different third variables known to be related to efficacy, knowledge, and/or participation. To analyze these alternative explanations, t tests and proportion tests were conducted between those with and without Internet access on the other independent variables and correlations between online campaign exposure and the rival explanation variables. These tests revealed that Internet access was significantly associated with all independent variables under consideration. Females, Blacks, and Hispanics were less likely to have Internet access than their counterparts. Those with Internet access were younger, had more education, had higher household incomes, were more likely to be Republicans, and expressed more interest in government and public affairs than those without access. Respondents without Internet access were more likely to be Democrats and watched more network news and local news. Further, those with access were more likely to have access to cable or satellite television, watch cable news, read newspapers, listen to political talk radio, and talk about politics with friends and family. Among those with Internet access, the authors examined the correlations between the days in the past week that respondents were exposed online to information about the presidential campaign and the other independent variables. Results indicated that online campaign exposure was not associated with being Hispanic or a Republican. It was negatively associated with age, being female, and being a Democrat. Positive associations were found with education, income, strength of partisanship, political interest, cable/satellite access, watching network, cable, and local news, newspaper reading, listening to political talk radio, and talking about politics with friends and family. To test whether there was a relationship between Internet access and online campaign exposure and the dependent variables, multivariate regressions were run to account for potential rival explanations.

Political Efficacy As shown in Table 2, the block of demographic variables explained 14.0% of the variance in internal political efficacy for the total sample and 12.2% of the variance when the sample was restricted to those with Internet access. The partisanship and political interest block explained 5.2% of the variance in internal efficacy for the total sample and 6.5% of the variance for those with Internet access. Inside this block, strength of partisanship (β = .05 for total sample; β = .06 for those with Internet access) and political interest (β = .17 for total sample; β = .19 for those with Internet access) variables were significant (p < .001), whereas identification with either of the major parties per se was not. In comparison to these first two blocks, the contribution made by the media exposure and political talk variables was considerably smaller but nevertheless statistically

184

Journal of Broadcasting & Electronic Media/June 2006

Table 2 Regressions Predicting Internal and External Political Efficacy Internal Political Efficacy

Predictor Variables Demographics Female Age Hispanic Black Education Income R2 (%) Partisanship/interest Republican Democrat Strength of partisanship Political interest Incremental R2 (%) Discussion/traditional media Political talk with friends/family Network news Cable news Local news Newspaper Political talk radio Cable access Incremental R2 (%) New media Internet access Online exposure to campaign information Incremental R2 (%) Total R2 (%) n

External Political Efficacy

Total Sample

Those With Internet Access

Total Sample

Those With Internet Access

–0.6*** –.06*** .01 .00 .19*** .09*** 14.0***

–.05*** –.06*** –.02 .00 .20*** .08*** 12.2***

.03** –.10*** .00 .02 .09*** .04*** 3.3***

.03** –.10*** .00 .02 .08*** .05*** 2.1***

.01 .01 .05*** .17*** 5.2***

.01 .00 .06*** .19*** 6.5***

.07*** .08*** .04*** .07*** 1.6***

.09*** .07*** .04** .07*** 1.7***

.15*** –.01 .01 –.06*** .02** .02* –.04*** 2.4***

.16*** –.03* .00 –.06*** .02 .01 –.04*** 2.7***

.06*** –.02 .01 .01 –.01 –.01 .00 0.3***

.06*** –.03 .01 .01 –.01 –.01 .00 0.3***

.05*** 0.2*** 21.9 10,816

.04*** .02* .04* 21.5 7,542

0.1*** 5.4 10,793

–.01 0.006 4.2 7,537

Note: Standardized regression coefficients. *p ≤ .05. **p ≤ .01. ***p ≤ .001.

significant. Political talk with family and/or friends had the largest standardized coefficient of the group (β = .15, p < .001, for total sample; β = .16, p < .001, for those with Internet access). Both local news exposure and cable access were negatively associated with internal efficacy (p < .001) in both equations. The cable access findings are

Kenski and Stroud/INTERNET USE AND POLITICS 185

consistent with the contention that entertainment may have a negative effect on these political variables (Scheufele & Nisbet, 2002) because, for many, cable access is used for obtaining entertainment content. Prior (2002) contended that cable television gives people who want to watch entertainment more opportunity to do so. Internet access and online exposure to campaign information were added last to the models. Both were positive and significant, providing support for the first hypotheses (H1a and H1b). Although the incremental R-square made by these additions was statistically significant in both instances, these changes were incredibly small, with Internet access making a contribution of .2% for the total sample and online exposure to campaign information among those with Internet access making an even smaller contribution at .04%.10 As with the models predicting internal efficacy, the models predicting external efficacy suggested that the demographic variables made the largest contribution, the partisanship and interest block the second largest, the media exposure and talk variables the third largest, with the Internet variables explaining the smallest amount of the variance overall. The results of the partisanship and interest block merited examination. Unlike the models predicting internal efficacy, in the external efficacy models, major party identification with either the Democrats or the Republicans resulted in significant and positive associations (p < .001), meaning that those who were major party identifiers were more inclined to report that they felt that they had an impact on the political system. Strength of partisanship and political interest were also positively associated with external efficacy. Similar to the internal efficacy model, the contribution made by Internet access was statistically significant (p < .001), but the R-square change was tiny at .1%. Internet access was positively related to external political efficacy. Online exposure to information about the presidential campaign made no statistically significant contribution to explaining external efficacy. As demonstrated by the overall R-squares, the amount of variance in external efficacy responses explained by the models was unimpressive to say the least. H2a was supported, whereas H2b was not.

Political Knowledge For both models shown in Table 3, demographic variables explained the most variation, with R-squares explaining 11.6% for the total sample and 11.8% for the Internet access sample. The contribution of the political interest variable was notable (β = .16, p < .001, for total sample; β = .17, p < .001, for those with Internet access). An interesting relationship appeared in the multivariate models. Being a Democratic was positively associated with political knowledge (β = .09, p < .001, for both samples), whereas being a Republican was negatively associated with knowledge (β = –.03, p < .05, for both samples). Consistent with prior research (Delli Carpini & Keeter, 1996), political talk with family and/or friends was positively associated with knowledge (β = .15, p < .001, for

186

Journal of Broadcasting & Electronic Media/June 2006

Table 3 Regressions Predicting Political Knowledge and Participation Political Knowledge

Predictor Variables Demographics Female Age Hispanic Black Education Income R2 (%) Partisanship/interest Republican Democrat Strength of partisanship Political interest Incremental R2 (%) Discussion/traditional media Political talk with friends/family Network news Cable news Local news Newspaper Political talk radio Cable access Incremental R2 (%) New media Internet access Online exposure to campaign information Incremental R2 (%) Total R2 (%) n

Political Participation

Total Sample

Those With Internet Access

Total Sample

Those With Internet Access

–.06*** .01 –.06*** –.03** .18*** .06* 11.6***

–.06*** .01 –.06*** –.01 .20*** .06*** 11.8***

–.04*** –.02 .02** .01 .01 .03* 3.0***

–.04*** –.01 .03** .01 .00 .02* 2.5***

–.03** .09*** .03*** .16*** 5.5***

–.03* .09*** .03** .17*** 5.7***

.09*** .06*** .10*** .09*** 7.7***

.09*** .06*** .10*** .10*** 8.4***

.15*** –.02 .01 –.05*** .01 .01 –.03*** 2.1***

.15*** –.03* .00 –.06*** .01 .00 –.01 2.3***

.29*** .00 .02* –.02* .01 .08*** –.01 7.9***

.27*** .00 .01 –.02 .01 .08*** –.02 7.3***

.03** 0.1** 19.3 10,935

.01 .02* 0.05* 19.9 7,576

0.01 18.6 10,898

.04*** 0.2*** 18.4 7,650

Note: Standardized regression coefficients. *p ≤ .05. **p ≤ .01. ***p ≤ .001.

both samples) and produced the largest standardized coefficient out of the traditional media exposure and political talk block. Also noteworthy were the significant, negative effects of watching local news on political knowledge in both samples. These findings are consistent with those of other researchers who have found that local tele-

Kenski and Stroud/INTERNET USE AND POLITICS 187

vision news is not an effective information resource when it comes to national politics (Romer & Jamieson, 2000). Contrary to Johnson et al. (1999) and DiMaggio et al. (2001), these results suggested that the Internet contributes to political knowledge above and beyond other news media variables. By conventional measures of statistical significance (p < .05), both H3a and H3b obtained support. Yet, as shown in Table 3, the contributions were small.

Political Participation Comparing the R-square contributions of the various blocks revealed that the partisanship and interest block and the media exposure and political talk block explained political participation more than did the demographic variables. For the total sample, the partisanship and political interest variables explained 7.7% of the variance in political participation compared to 3.0% explained by the demographic variables. The traditional media exposure and political talk variables explained an additional 7.9%. As shown in Table 3, this pattern also appeared when the sample was restricted to those respondents with Internet access. The most notable association with participation was political talk with family and/or friends (β = .29, p < .001, for total sample; β = .27, p < .001, for those with Internet access). Internet access, in comparison, did not make a significant contribution to the participation model. H4a was thus not supported. Among those with Internet access, online campaign exposure was positively and significantly associated with participation (β = .04, p < .001, for those with Internet access), a finding that is consistent with Weber et al. (2003), who found that Internet participation was a significant predictor of political participation. H4b was supported. Online exposure’s addition, however, only increased the R-square by .2%.

Discussion The impact of the Internet on the political sphere has been the topic of much debate. Although one school of thought contends that the Internet has the potential to revitalize democracy, others believe that the Internet will not change patterns of political interest, efficacy, participation, and knowledge. The empirical implications of this study lean toward the former interpretation but only to a limited extent. This research found significant bivariate associations between Internet access and online exposure to campaign information and internal efficacy, external efficacy, political knowledge, and participation. The associations were not large, but they were positive and significant. After controlling for a host of other variables, furthermore, the findings of the positive and significant impact of the Internet on political knowledge, participation, and efficacy persisted. Yet, the incremental R-squares revealed that after all of the other variables were taken into consideration, the Internet variables explained an additional .2% of the variance at most.

188

Journal of Broadcasting & Electronic Media/June 2006

A study by Jennings and Zeitner (2003) looked at the relationship between Internet access and use for following public affairs with political efficacy and knowledge controlling for several demographic variables. That study, however, did not control for both political interest and media use variables in a single model. To remedy concerns about spurious associations that appear when comprehensive controls are not in place, this study did use an extensive news exposure and political involvement battery to uncover the effects of Internet access and online exposure to campaign information on political efficacy, knowledge, and participation. Despite the extensive controls, Internet access and online exposure to campaign information still affected internal efficacy and political knowledge. Taking into account the concerns of Graber (1996) and Norris (2001) that Internet use may not increase political involvement because those who have access to the Internet may be already more interested in politics, the authors controlled for interest and still found that online campaign exposure contributed to participation. But given the rather limited contribution made by the Internet variables in this study, enthusiasm over the Internet’s potential to revitalize democracy is bridled. To be sure, this test for Internet access and exposure making a significant contribution to explaining political efficacy, knowledge, and participation was a tough one. Given the number of controls implemented, uncovering any significant Internet relationship is notable, although the magnitude is small. The results of this inquiry reaffirm the large role played by demographics, as the demographic variables tended to explain more of the variance in political efficacy and knowledge than any of the partisanship or media variables. The only instances in which demographic variables accounted for less variance than the partisanship and media variables were in the regressions for political participation. Further, the block of partisanship and political interest variables was the second largest contributor to explaining the variance in political efficacy and knowledge. Media use was a distant third in explaining the variance associated with political knowledge and efficacy. This underscores the comparatively minor role of media after taking other variables into account. One limitation of these analyses is that the data consist of a single cross section. Although, like other researchers, the authors have assumed that it is Internet access and exposure that have causal impacts on important political variables, it is possible that the causal directions are reversed. So, although this study shows that Internet access and exposure are positively associated with political efficacy, knowledge, and participation, it does not show that Internet access and exposure are causally related. Further, this analysis evaluates responses to a question asking respondents how often they used the Internet to access information about the campaign for president. The wording of this question does not allow evaluations of particular Web sites but instead assesses the Internet as a whole. The Internet is not monolithic, however, and various Web sites may have different impacts on political knowledge, efficacy, and participation (Lupia & Baird, 2003; Lupia & Philpot, 2005). Avoidance of an ecological fallacy means that these findings do not permit inferences about the functioning of specific sites.

Kenski and Stroud/INTERNET USE AND POLITICS 189

A third observation about the study is warranted. The large sample of the NAES found many significant associations. Kenski (2003b) argues that “the statistical significance reported should not be confused with substantive significance, which is often a matter of interpretation. … The presence of statistical significance does not establish substantive significance” (p. 195). Although positive associations with the Internet variables were found, it should be noted that they were not large. Overall, the findings of the study should ease the concerns of cyber-pessimists who feared that Internet would have a negative effect on the efficacy, knowledge, and participation. Nevertheless, the positive and significant associations between the Internet variables and the dependent variables were small, suggesting that the Internet is not going to be a panacea for democracy.

Notes 1Although the authors refer to “political efficacy” in their study, the questions that they used to measure this construct have traditionally been considered measures of “external” efficacy. 2Niemi et al. (1991) only found a moderate correlation of .26 between external and internal efficacy items in their analysis of the 1988 NES, suggesting that these items are not one and the same. Balch (1974) argued that it was necessary to test the individual efficacy items against varied multiple external criteria. To that end, he correlated four individual efficacy items with the propensity to engage in protest behavior and political trust. His results showed that the internal efficacy items correlated with protest behavior and trust in different ways than the external efficacy items. In fact, the signs of the correlations were in the opposite directions. Consequently, by combining all political efficacy items into a single scale, the effects of the individual items are diluted and obscured. “In sum, the efficacy scale has consistently demonstrated a case for its decomposition” (Balch, 1974, p. 23). 3“Area code, exchange, and bank, representing the first eight digits of a ten-digit phone number, were randomly generated proportional to telephone company estimates of the count of residential numbers in each combination of area code, exchange, and bank. The last two digits of each phone number were generated entirely at random” (Waldman, 2004, pp. 13–14). Upon speaking with an adult in the household and determining how many adults resided there, an adult was randomly selected based on age. See Waldman (2004) and Kenski (2004) for details about the NAES survey protocols. 4Some items were issued to the entire sample during this period, whereas other items were issued to a random subsample. Consequently, the sample sizes vary depending on the number of respondents who were asked each question. 5Tests of these single efficacy items revealed that they operated similarly to entire efficacy indexes (Kenski & Jomini, 2004). Consistent with prior research (Morrell, 2003; Niemi et al., 1991), the internal efficacy item was more strongly correlated with education, media exposure, talking about politics with others, political interest, political participation, and political knowledge than the external efficacy item. Further, when it came to political trust and cynicism, the external efficacy item was more correlated with these items than the internal efficacy item. 6Randomization is used to eliminate guessing response-sets, which occur when some individuals are more likely to guess than others. Knowledge measures that do not randomize the “don’t know” responses and treat them as incorrect answers conflate knowledge with the propensity to guess (for a detailed explanation, see Mondak, 2001). 7Of Independents, 57.3% report that they are “strong” identifiers. This percentage is comparable to those obtained from Republicans (53.7%) and Democrats (53.4%). 8There is some imprecision in using “days per week” as a measure of media exposure. This wording equates those who are exposed to a medium for a few moments each day with those

190

Journal of Broadcasting & Electronic Media/June 2006

who are exposed to a medium for many hours each day. Due to this imprecision, the relationship between the media and the dependent variables may be understated. 9The Internet variable is constrained by the words “about the campaign for president,” making it more specific than studies that merely use a global “Internet use” variable. The wording of the Internet exposure item is sufficiently ambiguous to allow respondents to include a wide variety of information sources (e.g., Internet discussion, e-mails, and Web sites). 10Before examining analyses, one should decide whether the conventional p < .05 level of statistical significance (as found in the online exposure model) is notable when one is dealing with a sample size of over 7,500 cases. Just as researchers sometimes lower their threshold for acceptable significance levels when dealing with small samples (as is common practice in many experimental studies), one could argue a corollary: One’s expectations should be higher when dealing with a very large sample, such as the one used here. The authors, however, maintain that one should not adjust one’s standard for the threshold of statistical significance based on sample size and therefore use the conventional p < .05 as the standard of statistical significance. Simply put, statistical significance is not synonymous with substantive significance, and it should not be used as a gauge for the strength of the relationship between variables.

References Abramson, P. R., & Aldrich, J. H. (1982). The decline of electoral participation in America. American Political Science Review, 76, 502–521. Anderson, D. M. (2003). Cautious optimism about online politics and citizenship. In D. M. Anderson & M. Cornfield (Eds.), The civic web: Online politics and democratic values (pp. 19–34). Oxford, England: Rowman & Littlefield. Balch, G. I. (1974). Multiple indicators in survey research: The concept “sense of political efficacy.” Political Methodology, 1, 1–43. Bimber, B. (1998). The Internet and political mobilization: Research note on the 1996 election season. Social Science Computer Review, 16, 391–401. Campbell, A., Gurin, G., & Miller, W. E. (1954). The voter decides. Evanston, IL: Row, Peterson. Chaffee, S. H., Zhao, X., & Leshner, G. (1994). Political knowledge and the campaign media of 1992. Communication Research, 21, 305–324. Converse, P. E. (1972). Change in the American electorate. In A. Campbell & P. E. Converse (Eds.), The human meaning of social change (pp. 263–337). New York: Russell Sage Foundation. Cornfield, M. (2003). Adding in the Net: Making citizenship count in the digital age. In D. M. Anderson & M. Cornfield (Eds.), The civic web: Online politics and democratic values (pp. 97–112). Lanham, MD: Rowman & Littlefield. Delli Carpini, M. X., & Keeter, S. (1996). What Americans know about politics and why it matters. New Haven, CT: Yale University Press. Delli Carpini, M. X., & Keeter, S. (2003). The Internet and an informed citizenry. In D. M. Anderson & M. Cornfield (Eds.), The civic web: Online politics and democratic values (pp. 129–153). Lanham, MD: Rowman & Littlefield. DiMaggio, P., Hargittai, E., Neuman, W. R., & Robinson, J. P. (2001). Social implications of the Internet. Annual Review of Sociology, 27, 307–336. Dulio, D. A., Goff, D. L., & Thurber, J. A. (1999). Untangled web: Internet use during the 1998 election. PS: Political Science and Politics, 32, 53–59. Graber, D. (1996). The “new” media and politics: What does the future hold? PS: Political Science and Politics, 29, 33–36. Hampton, K., & Wellman, B. (2003). Neighboring in Netville: How the Internet supports community and social capital in a wired suburb. City & Community, 2, 277–311. Horrigan, J., Garrett, K., & Resnick, P. (2004). The Internet and democratic debate. Pew Internet and American Life Project and the University of Michigan School of Information. Retrieved May 6, 2005, from http://www.pewinternet.org/pdfs/PIP_Political_Info_Report.pdf

Kenski and Stroud/INTERNET USE AND POLITICS 191

Jamieson, K. H., & Kenski, K. (2004). Why the National Annenberg Election Survey? In D. Romer, K. Kenski, P. Waldman, C. Adasiewicz, & K. H. Jamieson (Eds.), Capturing campaign dynamics: The National Annenberg Election Survey (pp. 1–11). New York: Oxford University Press. Jennings, M. K., & Zeitner, B. (2003). Internet use and civic engagement: A longitudinal analysis. Public Opinion Quarterly, 67, 311–334. Johnson, T. J., Braima, M. A. M., & Sothirajah, J. (1999). Doing the traditional media sidestep: Comparing the effects of the Internet and other nontraditional media with traditional media in the 1996 presidential campaign. Journalism & Mass Communication Quarterly, 76, 99–123. Johnson, T. J., & Kaye, B. K. (2003). A boost or a bust for democracy? How the Web influenced political attitudes and behaviors in the 1996 and 2000 presidential elections. Harvard International Journal of Press/Politics, 8(3), 9–34. Kaid, L. L. (2003). Effects of political information in the 2000 presidential campaign. American Behavioral Scientist, 46, 677–691. Kenski, K. (2003a). The National Annenberg Election Survey 2000. The Polling Report, 19(15), 1, 7–8. Kenski, K. (2003b). Testing political knowledge: Should knowledge questions use two response categories or four? International Journal of Public Opinion Research, 15, 192–200. Kenski, K. (2004). The rolling cross-section design. In D. Romer, K. Kenski, P. Waldman, C. Adasiewicz, & K. H. Jamieson (Eds.), Capturing campaign dynamics: The National Annenberg Election Survey (pp. 56–65). New York: Oxford University Press. Kenski, K., & Jomini, N. (2004, May). The reciprocal effects of external and internal political efficacy: Results from the 2000 U.S. presidential election. Paper presented at the conference of the World Association for Public Opinion Research, Phoenix, AZ. Kohut, A., Rainie, L., Doherty, C., Parker, K., Flemming, G., Dimock, M., et al. (2000). Youth vote influenced by online information: Internet election news audience seeks convenience, familiar names. Pew Internet and American Life Project. Retrieved May 2, 2005, from http://www.pewinternet.org/pdfs/PRC_Politics_Report.pdf Krueger, B. S. (2002). Assessing the potential of Internet political participation in the United States: A resource approach. American Politics Research, 30, 476–498. Lin, Y., & Lim, S. (2002). Relationships of media use to political cynicism and efficacy: A preliminary study of young South Korean voters. Asian Journal of Communication, 12, 25–39. Lupia, A., & Baird, Z. (2003). Can Web sites change citizens? Implications of Web white and blue 2000. PS: Political Science & Politics, 36, 77–82. Lupia, A., & Philpot, T. S. (2005). Views from inside the Net: How Websites affect young adults’ political interest. Journal of Politics, 67, 1122–1142. McLeod, J. M., Scheufele, D. A., & Moy, P. (1999). Community, communication, and participation: The role of mass media and interpersonal discussion in local political participation. Political Communication, 16, 315–336. Mondak, J. J. (2001). Developing valid knowledge scales. American Journal of Political Science, 45, 224–238. Mondak, J. J., & Anderson, M. R. (2004). The knowledge gap: A reexamination of gender-based differences in political knowledge. Journal of Politics, 66, 492–512. Morrell, M. E. (2003). Survey and experimental evidence for a reliable and valid measure of internal political efficacy. Public Opinion Quarterly, 67, 589–602. Muhlberger, P. (2002, October). Political values and attitudes in Internet political discussion: Political transformation or politics as usual? Paper presented at Euricom Colloquium: Electronic Networks & Democracy, Nijmegen, The Netherlands. Retrieved December 8, 2004, from http://oase.uci.kun.nl/~jankow/Euricom/papers/Muhlberger.pdf Neuman, W. R. (1986). The paradox of mass politics: Knowledge and opinion in the American electorate. Cambridge, MA: Harvard University Press. Nie, N. H., & Erbring, L. (2000). Internet and society: A preliminary report. Stanford Institute for the Quantitative Study of Society. Retrieved December 20, 2004, from http://www.stanford .edu/group/siqss/Press_Release/Preliminary_Report.pdf

192

Journal of Broadcasting & Electronic Media/June 2006

Nie, N. H., Junn, J., & Stehlik-Barry, K. (1996). Education and democratic citizenship in America. Chicago: University of Chicago Press. Niemi, R. G., Craig, S. C., & Mattei, F. (1991). Measuring internal political efficacy in the 1988 National Election Study. American Political Science Review, 85, 1407–1413. Norris, P. (2001). Digital divide: Civic engagement, information poverty, and the Internet worldwide. New York: Cambridge University Press. Pinkleton, B. E., Austin, E. W., & Fortman, K. K. J. (1998). Relationships of media use and political disaffection to political efficacy and voting behavior. Journal of Broadcasting & Electronic Media, 42, 34–49. Prior, M. (2002). Liberated viewers, polarized voters—The implications of increased media choice for democratic politics. The Good Society, 11(3), 10–16. Quan-Haase, A., Wellman, B., Witte, J. C., & Hampton, K. N. (2002). Capitalizing on the Net: Social contact, civic engagement, and sense of community. In B. Wellman & C. Haythornthwaite (Eds.), The Internet in everyday life (pp. 291–324). Malden, MA: Blackwell. Rainie, L., Cornfield, M., & Horrigan, J. (2005). The Internet and campaign 2004. Pew Internet and American Life Project. Retrieved May 2, 2005, from http://www.pewinternet.org/pdfs/ PIP_2004_Campaign.pdf Romer, D., & Jamieson, K. H. (2000). Does local television news inform as well as local newspapers? In K. H. Jamieson (Ed.), Everything you think you know about politics … and why you’re wrong (pp. 149–153). New York: Basic Books. Romer, D., Kenski, K., Waldman, P., Adasiewicz, C., & Jamieson, K. H. (Eds.). (2004). Capturing campaign dynamics: The National Annenberg Election Survey. New York: Oxford University Press. Rosenstone, S. J., & Hansen, J. M. (1993). Mobilization, participation, and democracy in America. New York: Macmillan. Scheufele, D. A., & Nisbet, M. C. (2002). Being a citizen online: New opportunities and dead ends. Harvard International Journal of Press/Politics, 7(3), 55–75. Shah, D. V., Kwak, N., & Holbert, R. L. (2001). “Connecting” and “disconnecting” with civic life: Patterns of Internet use and the production of social capital. Political Communication, 18, 141–162. Stromer-Galley, J. (2000). On-line interaction and why candidates avoid it. Journal of Communication, 50(4), 111–132. Waldman, P. (2004). Survey procedures, content, and dataset overview. In D. Romer, K. Kenski, P. Waldman, C. Adasiewicz, & K. H. Jamieson (Eds.), Capturing campaign dynamics: The National Annenberg Election Survey (pp. 12–33). New York: Oxford University Press. Weber, L. M., Loumakis, A., & Bergman, J. (2003). Who participates and why? An analysis of citizens on the Internet and the mass public. Social Science Computer Review, 21, 26–42. Wellman, B., & Hampton, K. (1999). Living networked on and offline. Contemporary Sociology, 28, 648–654.