Campaign-Related Social Networking and the Political Participation of College Students

Abstract

College students politically participate through traditional mechanisms at lower rates than their elders. Yet, members of this group may participate by other means, like friending candidates and joining political groups through social networking websites. We argue that these online activities serve as a meaningful form of civic engagement by broadening who participates and encouraging other forms of participation. Using a survey of randomly chosen undergraduates at a large Midwestern university, we discover that important distinctions exist between those who friend or join these online social networks and those who participate in more traditional off-line political activities. While interest in politics is a precursor to off-line engagement, it does not predict friending or joining an online social network that is political in nature. However, friending candidates or joining such networks appears to mobilize college students to engage in other forms of political participation.

Keywords

political participation political mobilization social networking online participation off-line participation civic engagement

Younger voters have frequently been at the center of movements for political and social change. Yet, their energy rarely makes its way to the voting booth at the levels that many anticipate. In fact, some evidence suggests that young adults’ level of participation in the political process has worsened over time. The percentage of young adults who exercise their rights to vote in the first election that they are able to do so has declined fairly substantially since 1972 (Levine & Lopez, 2002; Miller & Shanks, 1996). In a similar vein, many accounts of civic engagement frequently bemoan low levels of political participation among young adults. Studies show that they are much less likely to engage in traditional political activities like voting, contributing money to candidates or political parties, or contacting elected officials than their elders (see e.g., Rosenstone & Hansen, 1993; Zukin, Keeter, Andolina, Jenkins, & Delli Carpini, 2006). Based on these findings, one could easily conclude that young adults are politically disengaged and likely to remain so.

Perhaps, scholars have been looking in the wrong places. When Zukin, Keeter, Andolina, Jenkins, and Delli Carpini (2006) broadened their study of civic engagement to include activities such as signing e-mail petitions, the generation they christened the “dot-nets” did not seem so disengaged. In fact, the youngest generation missed being the most likely to report signing an e-mail petition by only one percentage point. This generation, like Zukin et al.’s (2006) label of it implies, should favor online forms of political engagement. While engagement in traditional forms of political activity may be low among the young, Internet use is not; 89% of adults 18 to 24 are online, a higher percentage than any other age group old enough to vote (Jones & Fox, 2009). It is often the first to take advantage of advancements on the Internet. This has generated optimism among some scholars that the Internet could help equalize political participation rates across generations (Delli Carpini, 2000; Shah, Cho, Eveland, & Kwak, 2005).

While the Internet allows for multiple forms of participation, we investigate one in particular: “friending” or joining an online social network that is political in nature. Friending a candidate is a relatively low-cost form of political participation that takes little more than a click of a mouse. While some scholars have expressed skepticism that such a seemingly inconsequential form of participation constitutes meaningful civic engagement, we argue that this form of participation is important. It serves as an easy entry point for young adults into the political process and is likely to result in participating in other, more traditional ways. In addition, friending embodies no less a “political act” than putting a sign on one’s lawn or wearing a campaign button—two acts that are often measured in studies of campaign activity.

We begin our analysis by identifying the varying factors associated with engaging in campaign-related friending and engaging in more traditional off-line forms of participation. While some of these mirror pathways to traditional forms of participation, we identify some key differences between the precursors of online and off-line participation among college students using a random sample survey of full-time undergraduate students at a large master’s level university in the Midwest. We find substantial evidence that joining online political social networks is positively related with greater levels of off-line participation.

Low-Cost Online Political Engagement: The Precursors to Friending

Online social networking has changed and grown dramatically since the 2000 election. None of the prominent social networking sites in 2008 existed when George W. Bush first ran for president. Facebook did not launch until 2004, and competitors MySpace and LinkedIn emerged on the scene only 1 year prior. By Fall of 2008, Nielsen reported over 59 million MySpace users, 39 million Facebook users, and nearly 12 million LinkedIn users. While the growth of MySpace had slowed to 1%, users of both Facebook and LinkedIn had more than doubled since 2007 (Nielsen, 2008).

Campaigns, too, have increasingly used the Internet such that both major presidential candidates had a heavy online presence by 2008. During this election year, the campaigns blogged, produced Internet-only advertisements that were posted on YouTube, and solicited campaign contributions online. Barack Obama quickly saw the use of social networking as a potential campaign tool. In February 2007, he met with one of Facebook’s board members about how he might leverage social networking tools to spread his message and build his support base as part of his campaign for president (Carr, 2008). Other candidates soon followed suit but were unable to fully match his online efforts. One pundit stated that “Were it not for the Internet, Barack Obama would not be president.”¹

Due to patterns in Internet use, these efforts disproportionately reached young adults, a group historically difficult to mobilize. Those aged 32 and under are far more likely to read and write blogs, watch videos on the web, and use social networking sites than older generations (Jones & Fox, 2009). Conventional wisdom might suggest these efforts are wasted. Instead, we investigate how social networking broadened the political participation of college students during this election. To accomplish this task, we examine how political characteristics, differences between online and off-line forms of political participation, as well as Internet usage and respondent characteristics predict participation patterns.

Political Characteristics

Interest in Politics

Interest in politics strongly predicts most standard forms of political participation (e.g., Campbell, Converse, Miller, & Stokes, 1960; Verba & Nie, 1972; Verba, Schlozman, & Brady, 1995). However, studies of Internet use and political engagement have been divided over the role of interest in generating online engagement.² Some emphasize that even if the Internet lowers the costs of participating, this cost will only be borne by those who are already interested, reinforcing the participatory divide (see e.g., Bimber, 1999; Krueger, 2006; Xenos & Moy, 2007). Others identify the potential of the Internet to make political information more easily accessible (see e.g., Tolbert & McNeal, 2003). When this material becomes easier to access, this narrows the existing participatory divides that knowledge gaps create (see e.g., Tolbert & McNeal, 2003). Delli Carpini (2000, p. 348) effectively captures elements of both perspectives as they apply to young adults:

“What these technologies seem clearly to provide is access to young adults, an increased ability for organized interests to more effectively reach young adults, and new or easier opportunities for already engaged (and perhaps interested but not yet engaged) young adults to participate and do so more effectively. Less clear (but still possible) is that the new technology can also serve as a way to increase the motivation of currently disinterested and disengaged young adults.”

Compared to many other forms of Internet engagement that have received prior scholarly attention,³ friending is a particularly low-cost activity that does not require that the impetus to participate comes from the individual participating. Since friending requires very little investment of time, effort, or resources, many of the standard precursors to political participation might not explain engagement in this activity. Friend requests can be made or granted with the click of a mouse and posting a profile on a social networking site can rapidly expand one’s social network. Although interest in politics still might encourage friending, no prior political knowledge or involvement is required.

Campaign-related friend requests differ from campaign-related e-mails in several ways. While campaign spam can generate such outrage that campaign consultants typically confine sending e-mails to those who request them (Krueger, 2006), unsolicited friend requests are not met with the same ire as unsolicited e-mails. Becoming a part of an online social network fosters a sense of social connection and lacks the impersonality of an e-mail listserv. Further, fewer taboos exist with friending strangers than e-mailing them, especially when these “friends” are celebrities of some kind. Thus, although unsolicited political e-mail fails to exert a mobilizing influence (Krueger, 2010), friending, regardless of who first initiates it, has greater potential to encourage political participation.

Online social networks can rapidly expand outward as those in one individual’s social network become connected with each other and their social networks. Joining an online political social network can just as easily come at the request of campaign supporters or those affiliated with the campaign as it can from the individual who joins. When someone otherwise uninterested and uninvolved in politics has politically engaged “friends” who “friend” candidates, that person may then decide to expand their online network to also include political candidates. Since friending carries such little cost in terms of time and effort and social networks can spread so easily, interest in politics should have less of an impact on this particular form of participation than it does on traditional forms of participation. While those who initially actively seek out opportunities to join political social networks can still be expected to have greater interest in politics, as these networks spread, people may respond favorably to requests to join regardless of their level of interest in politics.

Major Political Issues

Two issues dominated the 2008 campaign: the economy and the war in Iraq. Of the two, the war in Iraq seemed to be the more immediate issue among college students and the one on which the candidates offered the clearest policy differences. Students who opposed the Bush administration’s handling of the war should have been more energized to participate politically because the election offered the potential for bringing policy change. Moffett, Rice, and Madupalli (in press) discovered that college students who opposed Bush’s handling of the Iraq war are more civically engaged, and Best and Krueger (2005) found that those who participate online are more likely to oppose the war in Iraq. Consequently, we expect that students who oppose the war will be more civically engaged than those who support it. We expect them to turn in particular to social networking to participate, since this method of participation is more familiar to the generation than off-line forms of engagement.

Opinions about Bush’s handling of the economy, on the other hand, should not affect friending campaigns. College students were more isolated from its effects, the economy was a less polarizing issue, and both sides agreed that economic policies needed changed to promote recovery. We include it primarily as a control to help ensure that opposition to the war captures views on Iraq and not simply opposition to the Bush administration.⁴

Candidate Preference and Ideology

We also investigate the link between candidate preference amd friending of candidates. Since the Obama campaign placed greater emphasis on social networking as part of their overall strategy, opportunities to join his campaign’s social network should have been more visible, attracting individuals to join at their own initiative, and more frequently, offering unsolicited invitations to join. Thus, we expect those who supported Obama to be more likely to report engaging in campaign-related friending than those who supported other candidates.

Ideology can also shape the likelihood of participation, especially when there is a candidate that energizes a particular ideological group (Beck & Jennings, 1979). In the case of 2008, the impetus appears to have been on the left. However, the presence of a strong liberal candidate may in turn mobilize conservative detractors. There is also some prior evidence that participation online may be ideologically skewed. Best and Krueger (2005) found that those who participate online are more likely to be liberal. Accordingly, we control for respondents’ ideology and expect that in the 2008 election liberals and conservatives may have greater incentives for participation than moderates.

Online and Off-Line Participation

If the same people who already participate more off-line also form a disproportionate share of those who join online political social networks, then social networking may do little to expand the participatory base.⁵ Several studies of other forms of Internet participation suggest that those who participate off-line are also more likely to participate online (see e.g., Wellman, Hasse, White, & Hampton, 2001).⁶ However, online and off-line participation have different precursors. For example, Best and Krueger (2005) found that civic skills do not predict online activity. In addition, they found that exposure to political mobilization online led to further online participation but it did not translate to greater off-line activity (Best & Krueger, 2005). This suggests that those who spend time online, even if they are not otherwise prone to participate politically, may be mobilized to do so but the mobilization effects only hold true for activities that can be done online. In contrast, Shah, Kwak, and Holbert (2001) found that informational use of the Internet, particularly among the youngest generation they examine—Generation X—is a strong predictor of more traditional forms of civic engagement.

We believe there is good reason to expect students’ online political activity in political social networks to translate into higher levels of off-line engagement. One mechanism through which this could occur is through facilitating political discussion. McClurg (2003) finds that political discussion in off-line social networks leads to greater exposure to political information as well as greater levels of participation. Shah, Cho, Eveland, and Kwak (2005) demonstrate how informational uses of the Internet can lead to interpersonal political discussion which in turn can stimulate greater civic participation. Min’s (2007) experiment among college students concludes that face-to-face and online deliberation both, “have positive impacts on the participants' issue knowledge, political efficacy, and willingness to participate in politics, although the effect of online deliberation on the participation scale turned out to be somewhat smaller.”

We suspect similar but larger effects should hold true for participation in politically relevant online social networks but via a slightly different mechanism. Writing before the development of online social networking, Wellman, Hasse, White, and Hampton (2001) noted that the Internet facilitates organizational involvement by making it easier to connect to and communicate with individuals with similar interests. They predicted that “People already participating off-line will use the Internet to augment and extend their participation. People already participating online will get more involved in person with organizations and politics” (Wellman et al., 2001, p. 450). Online social networking makes the latter far easier.

When students become a “friend” of a candidate or political party, they not only are exposed to political information but also receive requests for further participation. Thus, friending a candidate can potentially result in further mobilization. Members of the online social network can easily discover how they can get further involved and may often be invited to do so. These mobilization attempts may help translate into off-line political participation. The simple decision to join a political social network may open the door to receiving requests to attend campaign events, volunteer for the campaign, or donate money that the person might not have otherwise received. In this way, friending can potentially broaden who participates and serves as an easy entry point to engaging in other forms of political participation.

Internet Usage and Respondent Characteristics

Beyond the role that political characteristics play in fostering online engagement, Internet access and online skills are associated with greater levels of online participation among the population at large (Best & Krueger, 2005). When examining a population of undergraduates, online skills should be less of an issue. Nearly 90% of all 18- to 24-year-olds are Internet users (Jones & Fox, 2009). On college campuses, the percentage approaches 100%. At the university in our study, all students have, at minimum, on-campus Internet access and are expected to use university e-mail to communicate about coursework. However, not all students use the Internet equally. While differences exist in patterns of Internet use by gender for other generations, among this age group, men and women are equally likely to be a part of an online social network (Hoffman, 2008). As a result, we do not expect differences in online participation on the basis of gender.^7,8

Instead, we expect that those students who spend more time online are more likely to friend political candidates and campaigns. A long line of research links informational uses of the Internet with greater participation (Kenski & Stroud, 2006; Norris, 1998; Shah et al., 2005; Shah, Kwak, & Holbert, 2001; Tolbert & McNeal, 2003). Thus, we control for two such forms of online activity: frequency of reading news online and frequency of reading blogs, as Kenski and Stroud (2006) discover that these should yield higher levels of participation. We also control for the posting of political views on blogs or other websites. To the extent posting is part of online deliberation, the findings of Min (2007) suggest it should encourage greater participation.

Other factors can also affect the propensity of individuals to “friend” candidates or join online groups through social networking websites. Because of the way by which online social networks expand, the level at which an individual’s friends are politically engaged may be at least as important, if not more important, in determining whether one joins an online political social network than one’s own interest in politics. Political science majors and strong partisans typically have more interest in politics than the average person and thus can be expected to participate more. They also are more likely to have politically engaged friends. Thus, we expect those with more involved friends, strong partisans, and political science majors should all be more likely to engage in friending political candidates and groups. Further, we expect that the impact of each of these variables on friending should eclipse that of interest in politics.

However, when it comes to more costly and traditional forms of political involvement, the converse should hold true—interest in politics should play a greater role in predicting activity than having engaged friends. Finally, we expect that members of the military might be less likely to be civically engaged because of this institution’s long-standing norms against political activity by its personnel (Sarkesian, 1981). This should hold especially true for campaign-related friending since it can reveal one’s political preference in a relatively public way.

Data and Methods

Between October 20, 2008, and November 4, 2008, we performed an Internet-based survey of randomly selected, full-time, undergraduate students who are between 18 and 25 years of age at a 4-year, master’s-level University located about 20 miles from St. Louis, Missouri, with approximately 14,000 students. We used a web-based instrument to study online civic engagement because one cannot engage in such activity when she or he lacks the ability to use a computer to complete basic tasks, like participating in a survey of this nature. We used the complete list of university assigned student e-mail addresses to select appropriately aged undergraduate students into our sample. In our initial contact to these students, we described the survey, our aims for the project, and gave them the opportunity to enter into a drawing for ten fifty dollar gift cards for participating in the survey. During this 3-week period, we sent eight follow-up e-mails approximately 2 days apart to those students who were in our sample but had not completed the survey. These e-mails reminded the students about our survey, along with the opportunity to enter the drawing. Of the 2,035 students who were randomly selected to participate in the survey, 351 completed it.⁹

Our sample of students was fairly diverse and resembled the demographic makeup of the university. Eighteen (5%) of the students in our sample have some form of military experience, and the average age of students in our sample is 20.8. In addition, our sample includes 205 (59%) females, and 142 (41%) males. Further, our sample includes 302 (85%) Whites, 29 African Americans (8%), 7 (2%) Hispanics, and 4 (1%) Asian Americans. Moreover, 298 (86%) of the students in our sample come from the state in which this university is located. While these characteristics may suggest that our sample is predominantly female, White, and from the state in which our university is located, it is fairly consistent with the population of undergraduate students at this institution.¹⁰ Female undergraduates comprised 54% of the population; Whites comprised 84% of the student body, and 90% of our undergraduates come from the state in which this university is located.

Our dependent variable asked about the frequency with which each respondent has friended or joined a group related to a presidential candidate or political party on a social networking website. Respondents were asked to rate the frequency with which they have engaged in this activity on a 5-point scale that ranged from “never” to “very often.” A more detailed discussion of this and the other variables used in this analysis is available in Appendix B. Since our dependent variable is ordered, we employ an ordered logistic regression to test our expectations.¹¹

We also constructed an index of off-line civic engagement (α = .73). This scale was constructed based on answers to each appendage to this phrase: “During 2008 how often have you . . .” (1) “worn a campaign button or shirt, put a campaign sticker on your car, or placed a sign in your window or in front of your residence,” (2) “tried to talk to people and explain why they should vote for or against one of the parties or candidates,” (3) “contacted a newspaper, radio, or TV talk show to express your opinion on an issue,” (4) “attended any political meetings, rallies, speeches, dinners, or things like that in support of a particular candidate,” (5) “participated in political activities such as protests, marches, or demonstrations,” (6) “contributed money to a Republican candidate or political party,” (7) “contributed money to a Democratic candidate or political party,” (8) “worked or volunteered on a political campaign for a candidate or party,” (9) “contacted or visited someone in government who represents your community,” (10) “worked with a group to solve a problem in a community,” (11) “made a purchasing decision based on the conduct or values of a company,” and (12) “participated in any community service or volunteer activity? By volunteer activity, we mean actually working in some way to help others for no pay.”¹² For each activity, the same response choices were offered as for friending. Since this variable is relatively continuous, we employ ordinary least squares (OLS) regression to test our expectations.¹³

Each respondent was asked about the degree with which she or he is interested in politics. The possible responses that we used to measure this concept were answered on a 4-point scale for which three points were allocated to “very interested,” two points for “somewhat interested,” one point for “not very interested,” and zero points for “not at all Interested.”¹⁴

We employed two sets of issue-specific variables to investigate the effects of President Bush’s handling of the Iraq war and the economy on online engagement. The first set of independent variables asked whether respondents disapproved of how former President George W. Bush handled the war in Iraq. We coded this variable one for those respondents who disapprove. Our secondary independent variable asked whether respondents who disapproved of how Bush handled the war in Iraq did so strongly, or not strongly. We coded this variable one for those who strongly disapprove. The comparison category for both variables consists of those who approve of the way in which Bush handled the war in Iraq, or responded that they did not know whether they approve or disapprove. We also asked comparable questions regarding respondents’ approval or disapproval of George W. Bush’s handling of the economy and coded them in the same manner as the procedure described for the Iraq war variables.

To investigate the effects of ideology on civic engagement, we asked each respondent whether she or he identifies herself or himself as a conservative, moderate, liberal, other, or did not know. If the respondent replied, “don’t know” or “unknown,” then this was followed up with a question about whether that respondent identifies him or herself as a liberal or conservative. Based on replies to both questions, we created dummy variables for liberals and conservatives.¹⁵ For instance, the binary variable for liberals is coded one when the respondent replied that she or he was a liberal or self-identified as a liberal in the follow-up question that was directed to those who did not know their ideology.¹⁶ To measure voter intent, each respondent was asked, “Who do you think you will vote for in the election for President?” From here, we constructed a dummy variable for those who intended to vote for Barack Obama.¹⁷

Also, we employed three questions that asked respondents about the frequency with which they engage in a variety of online activities. First, we asked each respondent how frequently she or he read news on the Internet about politics and current events. Second, we asked each student how frequently she or he read Internet blogs about politics and current events. Finally, we asked each student how frequently she or he expressed her or his views about politics on a website, blog, or chat room. The items that we used to measure each construct were answered on a 5-point response scale for which zero point was allocated to “not at all” and four points for “very often.”

Finally, we utilized four sets of variables to investigate the effects of respondent characteristics on online engagement activity. First, we constructed an index based on the replies that respondents gave to three questions that investigate the degree to which the friends of each student engage in a series of activities to investigate the effects of peer civic experiences on online engagement activities. The items that we used to measure this construct—voting in elections, engaging in volunteer work, and encouraging expression about politics and current events—were answered on a 5-point response scale for which zero point was allocated to “strongly disagree,” one point for “disagree,” two points for “neutral,” three points for “agree,” and four points for “strongly agree” (α = .65).¹⁸

Second, to measure the degree of partisanship among respondents, each person was asked whether she or he identifies herself or himself or as a Republican, Democrat, independent, or something else. Using two questions with slightly different wording, each respondent who self-identified as a Democrat or Republican was asked whether she or he strongly or not strongly identified as a Democrat or Republican, respectively. From this data, we constructed a variable that is coded one for strong partisans.

Third, we asked whether each student is a political science major through a dummy variable constructed based on a question about student major/majors in the survey. Finally, we asked each respondent whether she or he has or is currently serving in the U.S. military, the National Guard, military reserves, or in a Reserve Officers’ Training Corps (ROTC) program. We used this to create a binary variable for those who are currently serving or have formerly served in the military.

Results

Table 1 provides the results of two separate models of civic engagement. Model 1 uses campaign-related friending or joining activity as the response variable. We used odds ratios to interpret many of the coefficients in this model. These ratios allow us to assess the impact that each coefficient has on our dependent variable in a more direct manner than predicted probabilities can when using ordered logistic regression.^{19 20} Model 2 employs off-line civic engagement as the dependent variable.

Table 1.

Online and Off-Line Civic Engagement Activities Among College Voters and the 2008 Election.

Independent variable	Campaign-related friending or joining activity	Odds ratio (friending or joining)	Off-line civic engagement
Off-line civic engagement	.128*** (.026)	1.137	−
Campaign-related friending activity	−		1.577*** (.268)
Political posting activity	.232* (.129)	1.261	1.522*** (.313)
Political characteristics
Interest in politics	−.239 (.213)	.787	1.655** (.519)
Disapprove of Bush’s handling of the war in Iraq	1.462** (.520)	4.313	−1.079 (1.332)
Strongly disapprove of Bush’s handling of the war in Iraq	.790* (.429)	2.202	.565 (1.047)
Disapprove of Bush’s handling of the economy	.018 (.477)	1.018	.447 (1.123)
Strongly disapprove of Bush’s handling of the economy	.306 (.457)	1.357	.045 (1.117)
Liberal	.187 (.347)	1.205	−.312 (.880)
Conservative	.818* (.447)	2.266	−1.439 (1.089)
Intend to vote for Obama	.540 (.402)	1.715	−.785 (.974)
Internet use and respondent characteristics
Frequency of online news reading	.278* (.147)	1.320	.597* (.360)
Frequency of political blog reading	−.015 (.142)	.985	−.002 (.364)
Peer civic experiences	.059 (.068)	1.061	.342* (.166)
Strong partisan	.733** (.304)	2.081	.797 (.781)
Political science major	1.154* (.695)	3.170	3.512* (1.703)
Military	−.154 (.564)	.857	2.530* (1.477)
Cut Point 1	3.244*** (.851)		−
Cut Point 2	3.932*** (.864)		−
Cut Point 3	4.978*** (.884)		−
Cut Point 4	6.475*** (.926)		−
Constant	−		−2.822 (1.994)
N	240		240
Log likelihood	−286.007		−
Pseudo R²	.186		−
R ²	−		.539
Adjusted R²	−		.505
Chi-square	130.26		−
Prob>chi-square	<.0001		−
F statistic	−		16.26
Prob>F statistic	−		<.0001

Note. The first model uses ordered logistic regression. The model for off-line civic engagement employs OLS. The values in parenthesis are standard errors.

*p < .05. **p < .01. ***p < .001, all one-tailed tests.

We uncovered mixed evidence that links the political characteristics of individual respondents to their propensity to engage in differing levels of online and off-line forms of engagement. On one hand, the expected value of the off-line engagement index increased by roughly 1.7 points for each increase in a respondent’s level of interest in politics. However, interest in politics was not a significant predictor of engaging in friending.

Respondents’ views on issues appear to shape online political behavior largely in the ways we anticipated. Views on George W. Bush’s handling of the Iraq War were positively related to the propensity to which respondents engaged in political friending or joining activity. More specifically, the odds that favored higher levels of this activity increased by 331.3% among those who disapproved of Bush’s handling of the Iraq War, and by 120.2% among those who strongly disapproved of his handling of the war in Iraq. In contrast, war views had no direct impact on off-line civic engagement. This suggests that students’ opinions are more likely to translate into action when those actions require less time and effort and can be done at any time that the student finds convenient. Meanwhile, as we expected, views on Bush’s handling of the economy had no significant impact on either form of activity.

We found very little evidence to suggest that ideology and vote choice drove online or off-line participation. Interestingly, political conservatives were more likely to exhibit higher levels of campaign-related friending or joining activity, since the odds that favored higher levels of such activity increased by about 122.7%.

Yet, as expected, we unearthed mixed evidence that Internet usage predicts online and off-line forms of political participation. On one hand, we found that each unit increase in the frequency of online news reading increased the odds that favored higher levels of friending or joining activity by about 3%. Moreover, each additional unit in the frequency of online news reading increased the expected value of off-line civic engagement by roughly 0.6 points. This is consistent with Shah et al. (2001) and Kenski and Stroud’s (2006) findings that informational uses of the Internet are a strong predictor of civic engagement. That said, we uncovered no evidence that frequency of blog reading affected campaign-related levels of friending or joining activity or off-line civic engagement. However, each unit increase in posting one’s political views on a blog or other website increased the odds that favored higher levels of political friending or joining activity by 13.7% and increased the expected value of off-line civic engagement by approximately 1.5 points.

While the precursors to friending campaigns and more traditional forms of off-line engagement share some similarities, the differences prove consequential. Most notably, interest in politics is not a prerequisite to political friending. This suggests that online political social networks draw students into the political process who might not otherwise participate. As a result, online social networking may help close the participatory divide in ways other, older forms of Internet activity failed to achieve. We return to these implications in the concluding section.

As anticipated, the findings for respondent characteristics suggest that a different process is at work producing off-line civic activity than online civic activity. The experiences of one’s peers and interest in politics matter, as the expected value of the off-line engagement index increased by approximately 0.3 points for each unit increase in the civic experiences of his or her peers. Yet, the civic engagement of one’s friends was not a significant predictor of friending activity.

In contrast, students who self-identified as strong partisans were more likely to engage in political friending and joining activity but were no more likely to have engaged in off-line forms of civic engagement.²¹ More specifically, the odds that favored higher levels of campaign-related friending or joining activity increased by 108.1% when one was a strong partisan.

Political science majors were more likely to be engaged in friending political campaigns as well as in off-line forms of civic engagement. The odds that favored higher levels of campaign-related friending or joining activities increased by 217%, when she or he was a political science major. Further, the expected value of the off-line engagement index increased by approximately 3.5 points if one was a political science major.

Whereas those with some level of military service seem less likely to engage in political friending, they were significantly more likely to engage in off-line forms of civic activities. The expected value of this index increased by approximately 2.5 points among those who are or were formerly in the military. These diverging patterns may reflect conflicting values, as these students must balance a strong sense of civic duty common among members of this group alongside norms against political activity by active military.

Comparing Online and Off-Line Engagement

Finally, and perhaps most importantly, the findings suggest that online engagement is associated with greater off-line engagement. The off-line civic engagement index increases by approximately 1.6 and 1.5 points, respectively, for each unit increase in political friending or joining activity and posting activity. This means that a person who engaged in friending political candidates or groups very often would increase their level of civic engagement by about 6.3 on a scale that has a maximum possible point value of 60, and a person who posts very frequently would increase their level of civic engagement by roughly 6.1. This is equivalent to participating in more than one additional off-line activity. Although the differences are relatively small, as expected, friending candidates is associated with greater off-line activity than is posting. This lends support to the idea that friending can be a mobilizing activity.²²

However, off-line civic engagement also has a substantial positive relationship with campaign-related friending, as the odds that favor a higher level of political friending or joining activity increase by 13.7% for each unit increase in off-line civic engagement. Those who engage in greater levels of off-line civic activities are more likely to engage in friending candidates or political parties.

To perform a more nuanced analysis of the results, we used CLARIFY to compute change in the predicted probability of each level of friending when engaging in posting or participating off-line (King, Tomz, & Wittenberg, 2000).²³ In each figure, the vertical axis denotes the change in the predicted probability, while the horizontal axis is the frequency of campaign-related friending.

As Figure 1 depicts, an increase in the level of posting from rarely to regularly results in a decrease in the predicted probability that one never befriends candidates or joins online groups that support candidates by 8.9%. Further, this change in amount of posting results in an increase in the predicted probabilities that one rarely engages in friending or joining activity by 2.6%, sometimes engages in this activity by 3.3%, regularly engages in this activity by 2.2%, and engages in this activity very often by 0.8%.

Figure 1.

Effects of online posting on campaign-related friending and joining.

Figure 2 demonstrates that a 1 standard deviation (SD) increase in off-line engagement results in a 15.1% decrease in the probability that one never engages in political friending or joining activity. Yet, this change in off-line engagement results in an increase in the predicted probabilities that one rarely friends or joins by 4.3%, sometimes engages in this activity by 5.7%, regularly engages in this activity by 3.7%, and engages in this activity very often by 1.3%.

Figure 2.

Effects of offline civic engagement on campaign-related friending and joining.

If we compare the variables that have the biggest impact in each model, both complementary and slightly different stories emerge. To tell these stories, we computed the predicted discrete changes in the dependent variable as one changes the value of an independent variable. To compute these changes, we change each independent variable from its minimum to its maximum value while holding all other independent variables at the mean.²⁴ High levels of off-line civic engagement make participation in friending or joining activities extremely likely. The greatest potential impact we see on friending candidates and political groups is when off-line civic engagement is at its highest. If an individual moved from absolutely no off-line civic engagement to our highest observed value on the scale, his or her predicted probability of friending very often increases by approximately 72.1%. In comparison, friending very often increases the expected off-line civic engagement score by about 7.8, more than any other variable.

Causality and Endogenity Concerns

Together, the findings suggest that friending candidates can lead to off-line engagement and off-line engagement can lead to friending candidates. This indicates that the presence of both direct and indirect effects and endogeneity issues are likely, and we expect the causal arrows do in fact point both ways. Unfortunately, the currently available methods that would allow us to establish causality are inappropriate, given the structure of our dependent variable and endogenous regressors. We are not aware of any approach that would allow us to have an ordered dependent variable and binary endogenous regressors in the same model.²⁵

That said, some respondents also completed a brief postelection survey that enables us to partially address endogenity concerns and provide further evidence that joining political online social networks sparks further off-line engagement. To do this, we compared civic engagement levels reported in the postelection survey (T2) across the different levels of campaign-related friending reported in the preelection survey (T1). If friending candidates sparks greater off-line engagement, such friending at T1 should lead to greater civic engagement levels at T2. We performed a series of difference in means tests to investigate this. The results of this test also inform us whether the results that we discover are statistically significant or are completely due to random sampling error (Kmenta, 1997).

To begin, we generated an index of off-line civic engagement that was constructed based on our postelection survey (α = .77).²⁶ We then computed the mean level of postelection survey civic engagement for those students who had never friended a candidate or joined a political group through a social networking website, and compared it with those who rarely did so, sometimes did so, regularly engaged in this activity, and did so very often. Table 2 provides the results of this comparison. As it shows, those who never friended had a mean civic engagement score of about 2.8, those who rarely engaged in this activity averaged 4.00 activities, sometimes friended averaged slightly more than four activities, regularly friended averaged around four activities, and friended very often averaged about six activities. The difference in mean civic engagement activities at T2 between those who did not friend at T1 and those who engaged in varying levels of friending activity at T1 are statistically significant.

Table 2.

Differences in Postelection Civic Engagement by Levels of Preelection Campaign-Related Friending.

Comparison	Never friend mean civic engagement	Comparison group mean civic engagement	T test	Degrees of freedom	P value
Never friend vs. rarely friend	2.776	4.000	2.944	113	.002
Never friend vs. sometimes friend	2.776	4.400	4.154	121	<.0001
Never friend vs. regularly friend	2.776	4.056	3.057	114	.0014
Never friend vs. friend very often	2.776	5.864	7.060	118	<.0001

Note. The number of respondents who never friended candidates or political groups is 98, rarely friended, 17; sometimes friended, 25; regularly friended, 18; and friend very often, 22.

While these results furnish some evidence that friending candidates during the preelection survey period led to greater reported levels of civic engagement in the postelection survey, they do not rule out the possibility that greater off-line activity also led to increased friending. If campaign-related friending leads to greater off-line civic engagement, then those who friended for the first time after the preelection survey should engage in more off-line activity than those who never joined an online political social network. To begin considering this possibility, we asked each respondent whether she or he has, “Friended or joined a group related to a presidential candidate or political party on a social networking site such as MySpace or Facebook” since October 1, 2008. If a respondent had engaged in such friending since October 1, 2008, but reported never having friended in the preelection administration of this instrument, we generated a binary variable, late friender, that is coded one for those in this situation, and zero otherwise. However, if a respondent had not engaged in friending since October 1, 2008, and reported never having friended in the preelection administration of this instrument, we generated a binary variable, never friender, that is coded one for those in this situation and zero otherwise.

To compare levels of civic engagement between late and never frienders, we computed the mean levels of preelection and postelection survey civic engagement for those students who never friended a candidate in either administration of this survey instrument. Also, we computed the mean levels of preelection and postelection survey civic engagement for those students who were late frienders. As Table 3 shows, those who never friended a candidate had an average preelection civic engagement score of approximately 6.1 out of a 60-point scale, while late frienders averaged about 7.2 on this measure. However, this difference was not statistically significant, suggesting that prior to friending, the two groups did not significantly vary in their levels of off-line civic engagement. Table 3 also shows that those who never friended had an average postelection civic engagement score of around 2.8 out of a 14-point scale, while late frienders averaged about 3.7 on this measure. This difference, in contrast, is statistically significant and provides further, albeit, suggestive evidence that friending candidates or campaign-related groups leads to greater off-line engagement.

Table 3.

Differences in Civic Engagement Levels Between Late and Never Frienders by Pre- and Postelection Levels of Civic Engagement.

Comparison	Never friend mean civic engagement	Late friending mean civic engagement	T test	Degrees of freedom	P value
Never frienders vs. late frienders (precivic engagement levels)	6.068	7.231	1.077	99	.142
Never frienders vs. late frienders (postcivic engagement levels)	2.776	3.667	1.872	108	.032

Note. The number of respondents who never friended candidates or political groups (preelection civic engagement) is 88, late friended (preelection civic engagement), 13; never friended (postelection civic engagement), 98; late friended (postelection civic engagement), 12.

Conclusion

In Rosenstone and Hansen’s (1993) landmark study of political participation, one of the chief inhibitors of political participation was the fact that “nobody asks.” The political use of social networking sites has the potential to counteract this problem, particularly among younger voters. The simple decision to friend a candidate, political party, or other political group, can open the door to a number of invitations to participate in the electoral and political process via other means. Measuring these invitations was outside the scope of our study, but we presented evidence that those who engaged in friending candidates and political groups were more politically mobilized or engaged. In fact, friending was associated with more off-line civic engagement than any other variable we examined. Its impact eclipsed even that of interest in politics.²⁷

Unlike more traditional forms of participation, interest in politics had no significant impact on who friends candidates and political groups. In this respect, campaign-related friending appears to engage students who otherwise might not be civically engaged. Scholars had long pointed to the potential of the Internet to expand the participatory base (Delli Carpini, 2000; Shah et al., 2005; Tolbert & McNeal, 2003), but many studies of specific forms of Internet use instead confirmed that they reinforced the participatory divide (see e.g., Krueger, 2006, 2010; Wellman et al., 2001). We identify one mechanism, campaign-related social networking, which appears to ameliorate part of this divide.

However, it by no means fully erases it. Not everyone is equally likely to engage in campaign-related friending. Strong partisans and political science majors as well as frequent online newsreaders were still more likely to engage in this form of political participation. Those who posted or engaged in political activities off-line were also more likely to friend. Perhaps they, too, result in receiving more friend requests from campaigns and political groups. Issue positions also play a role. In 2008, both those who disapproved of Bush’s handling of the war and those who labeled themselves conservatives were more likely to engage in campaign-related friending. Finally, we uncovered weak evidence in support of the impact of the Obama campaign’s social networking efforts over those of McCain’s. While the coefficient for Obama supporters was positive, it did not achieve standard levels of significance once other political views were accounted for.

Finally, our findings also suggest that among college students, participation, regardless of type, leads to more participation. Just as civic engagement could start by attending a political meeting, it can also begin with the click of a button to grant a friend request. To the extent social networking efforts continue to both disproportionately reach younger adults and lead to greater participation in more traditional forms, they have the potential to help level the political playing field. Regrettably, we must leave the further disentangling of these relationships as well as their potential differential impact across generations to future studies.

Additional research into the mobilizing power of campaign-related social networking is clearly needed. First, studies need to examine a larger population. Future studies should also examine both the general political information and the content and quantity of mobilization messages that individuals receive through friending a campaign or political group. It may also prove instructive to study the impact of whether friend requests come at the request of the campaign or the individual. Unsolicited political e-mails also have the potential to mobilize the yet to participate, but Krueger (2010) finds they fail to do so. However, responding favorably to a friend request requires far less initiative than signing up to be on an e-mail list, and once someone clicks yes, the messages received are no longer unsolicited. Thus, we expect that messages sent through social networking will increase participation, regardless of who first initiated the “friendship.” While there is ample room for additional survey research on this subject, experimental research that varies mobilization messages sent via social networking across treatment and control groups may best provide leverage over endogeneity concerns.

Much work on the topic has yet to be done. Still, the results presented here demonstrate that although joining an online political social network may require little more effort than clicking a button, for college students that click is anything but trivial. Instead, it serves as a pathway to greater participation.

Footnotes

Appendix A

Appendix B

Appendix C

Table C1. Campaign-Related Friending and Political Posting Among College Students in the 2008 Election.

Frequency of campaign-related friending activity	Frequency of political posting activity
Frequency of campaign-related friending activity	Never	Rarely	Sometimes	Regularly	Very Often	Total
Never	139 (67.48%)	16 (30.19%)	16 (34.78%)	2 (12.50%)	7 (20.00%)	180 (50.56%)
Rarely	18 (8.74%)	11 \|(20.75%)	3 (6.52%)	2 (12.50%)	2 (5.71%)	36 (10.11%)
Sometimes	22 (10.68%)	12 (22.64%)	13 (28.26%)	5 (31.25%)	3 (8.57%)	55 (15.45%)
Regularly	19 (9.22%)	8 (15.09%)	9 (19.57%)	3 (18.75%)	7 (20.00%)	46 (12.92%)
Very often	8 (3.88%)	6 (11.32%)	5 (10.87%)	4 (25.00%)	16 (45.71%)	39 (10.96%)
Total	206 (100.00%)	53 (100.00%)	46 (100.00%)	16 (100.00%)	35 (100.00%)	356 (100.00%)

Note. The values in parentheses are column percentages. Chi-square = 106.96. p < .0001; Tau-b = .388.

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The College of Arts and Sciences Development Fund and The Department of Political Science at Southern Illinois University Edwardsville provided financial assistance to help carry out the survey described in this article.

Notes

References

Beck

P. A.

Jennings

M. K.

(1979). Political periods & political participation. American Political Science Review, 73, 737–750.

Best

S. J.

Krueger

B. S.

(2005). Analyzing the representativeness of internet political participation. Political Behavior, 27, 183–216.

Bimber

(1999). Information & political engagement in America: The search for effects of information technology at the individual level. Political Research Quarterly, 54, 53–67.

Bimber

(2000). The study of information technology & civic engagement. Political Communication, 17, 329–333.

Boulianne

(2009). Does internet use affect engagement? A meta-analysis of research. Political Communication, 26, 193–211.

Campbell

Converse

P. E.

Miller

W. E.

Stokes

D. E.

(1960). The American voter. Chicago, IL:University of Chicago Press.

Carr

(2008, November 9). How Obama tapped into social networks' power. The New York Times, pp. B1.

Delli Carpini

M. X.

(2000). Gen.com: Youth, civic engagement, & the new information environment. Political Communication, 17, 341–349.

Hoffman

2008. The Social Media Gender Gap. Bloomberg Businessweek. May 19. Retrieved from http://www.businessweek.com/stories/2008-05-19/the-social-media-gender-gapbusinessweek-business-news-stock-market-and-financial-advice

10.

Jones

Fox

(2009). Generations Online in 2009. Pew Internet Project Data Memo. January 28. Retrieved from http://www.pewinternet.org/Reports/2009/Generations-Online-in-2009.aspx

11.

Katz

J. E.

Rice

R. E.

Aspden

(2001). The internet, 1995-2000: Access, civic involvement, & social interaction. American Behavioral Scientist, 45, 405–419.

12.

Kenski

Stroud

N. J.

(2006). Connections between internet use & political efficacy, knowledge, & participation. Journal of Broadcasting & Electronic Media, 50, 173–192.

13.

King

Tomz

Wittenberg

(2000). Making the most of statistical analyses: Improving interpretation & presentation. The American Journal of Political Science, 44, 341–355.

14.

Kmenta

(1997). Elements of Econometrics 2nd ed. Ann Arbor: University of Michigan Press.

15.

Krueger

B. S.

(2002). Assessing the potential of internet political participation in the United States: A resource approach. American Politics Research 30, 476–498.

16.

Krueger

B. S.

(2006). A Comparison of Conventional & Internet Political Mobilization. American Politics Research, 34, 759–776.

17.

Krueger

B. S.

(2010). Opt in or tune out: Email mobilization & political participation. International Journal of E-Politics, 1, 55–76.

18.

Levine

Lopez

(2002). Youth voter turnout has declined by any measure. Center for Information & Research on Civic Learning & Engagement. Retrieved from http://www.civicyouth.org/research/products/fact_sheets_outside.htm

19.

Long

J. S.

(1997). Regression models for categorical & limited dependent variables. Thousand Oaks, CA: Sage.

20.

Long

J. S.

Freese

(2006). Regression models for categorical dependent variable using stata. College Station, TX: Stata Press.

21.

McClurg

S. D.

(2003). Social networks & political participation: The role of social interaction in explaining political participation. Political Research Quarterly, 56, 449–464.

22.

Miller

C. C.

(2008). How Obama’s Internet Campaign Changed Politics. The New York Times. November 7. Retrieved from http://bits.blogs.nytimes.com/2008/11/07/how-obamas-internet-campaign-changed-politics/

23.

Miller

W. E.

Shanks

J. M.

(1996). The new American voter. Cambridge, MA: Harvard University Press.

24.

Min

(2007). Online vs. face-to-face deliberation: Effects on civic engagement. Journal of Computer-Mediated Communication, 12. Reterivede from http://jcmc.indiana.edu/vol12/issue4/min.html

25.

Moffett

K.W.

Rice

Madupalli

(in press). Youth voters & war: The Iraq war as a catalyst for political participation. Social Science Quarterly.

26.

Nielsen. (2008). Twitter grows fastest, myspace still the social king. Nielsenwire. October 23. Retrieved from http://blog.nielsen.com/nielsenwire/online_mobile/leading-social-networking-sites-still-growing/

27.

Norris

(1998). Virtual democracy. Harvard International Journal of Press/Politics, 3, 1–4.

28.

Pasek

Romer

(2009). Realizing the social internet? Online social networking meets offline social capital. Journal of Information Technology & Politics, 6, 197–215.

29.

Rosenstone

S. J.

Hansen

J. M.

(1993). Mobilization, participation, & democracy in America. NewYork, NY: Longman.

30.

Sarkesian

S. C.

(1981). Military professionalism & civil-military relations in the West. International Political Science Review, 2, 283–297.

31.

Shah

D. V.

Cho

Eveland

W. P.

Kwak

(2005). Information & expression in a digital age: Modeling internet effects on civic participation. Communication Research, 32, 531–565.

32.

Shah

D. V.

Kwak

Holbert

R. L.

(2001). “Connecting” & “Disconnecting” with civic life: Patterns of internet use & the production of social capital. Political Communication, 18, 141–162.

33.

Stromer-Galley

(2002). New voices in the public sphere: A comparative analysis of interpersonal and online political talk. The Public, 9, 23–42.

34.

Thies

C. G.

Hogan

R. E.

(2005). The state of undergraduate training in political science. PS:Political Science & Politics, 38, 755–763.

35.

Tolbert

C. J.

McNeal

R. S.

(2003). Unraveling the effects of the internet on political participation? Political Research Quarterly, 56, 175–185.

36.

Verba

Nie

N. H.

(1972). Participation in America: Political democracy & social equality. NewYork, NY: Harper & Row.

37.

Verba

Schlozman

K. L.

Brady

H. E.

(1995). Voice & equality: Civic voluntarism in American politics. Cambridge, MA: Harvard University Press.

38.

Wellman

Hasse

A. Q.

White

Hampton

(2001). Does the Internet increase, decrease, or supplement social capital? American Behavioral Scientist, 45, 436–455.

39.

Xenos

Moy

(2007). Direct & differential effects of the internet on political & civic engagement. Journal of Communication, 57, 704–718.

40.

Zukin

Keeter

Andolina

Jenkins

Delli Carpini

M. X.

(2006). A new engagement? Political participation, civic life, & the changing American citizen. New York, NY: Oxford University Press.