Professional Documents
Culture Documents
Facebooking It To The Polls A Study in Online Social Networking and Political Behavior
Facebooking It To The Polls A Study in Online Social Networking and Political Behavior
Facebooking It To The Polls A Study in Online Social Networking and Political Behavior
Leticia Bode
To cite this article: Leticia Bode (2012) Facebooking It to the Polls: A Study in Online Social
Networking and Political Behavior, Journal of Information Technology & Politics, 9:4, 352-369,
DOI: 10.1080/19331681.2012.709045
Download by: [Biblioteca Universidad Complutense de Madrid] Date: 25 April 2016, At: 06:26
Journal of Information Technology & Politics, 9:352–369, 2012
Copyright © Taylor & Francis Group, LLC
ISSN: 1933-1681 print/1933-169X online
DOI: 10.1080/19331681.2012.709045
ABSTRACT. As Internet use continues to become integrated into the daily lives of average
Americans, it becomes necessary to understand the implications of each aspect of that use. As the
majority of adults now take advantage of social network sites, understanding how they make use of
those sites and what that means for American politics is increasingly important. This study considers
one such social network site, Facebook, and how engaging in specific behaviors that provide personal-
ized information, create community engagement, and generate social capital within that realm impacted
users’ decision to participate politically in the presidential primary of 2008. It concludes that certain
Facebook behaviors can be translated into participatory political behaviors, both online and offline.
Moreover, it is not whether or how often one uses Facebook, but rather the specific set of activities one
engages in during such time that drives the relationship between Facebook use and various types of
political participation.
KEYWORDS. 2008 election, Facebook, political participation, social media, social network
“The Internet may serve a similar function students have a Facebook profile. A majority
as offline interpersonal networks but on a of users (52%) log in to the site on a daily
much larger scale.” basis, and 56% update statuses at least weekly
—Hwang, Schmierbach, Paek, Gil de (Fox, 2011). All this activity makes Facebook
Zuniga, & Shah (2006, p. 467) the second most-trafficked Web site in the world
(“Facebook.com site info,” 2012).
The membership of Facebook is skyrocket- As a result of this activity, scholars are begin-
ing in parallel with its popularity. In 2012, the ning to consider what good or evil may come
site registered an amazing 845 million users, of Facebook use. There is a growing body
a number which is growing at the astounding of research, for example, focusing on privacy
rate of 30,000 per day, in countries all over the and self-presentation on Facebook. Authors are
world (Helvie-Mason, 2011; Protalinski, 2012). beginning to delve into the impact of low-
The numbers among undergraduates are noth- ered privacy expectations online, the likelihood
ing short of overwhelming—85% of college of users to release or post private informa-
Leticia Bode is an assistant professor in the Communication, Culture, and Technology department at
Georgetown University. She received her PhD in political science from the University of Wisconsin–Madison
in 2012. Her research and teaching interests include political communication, political behavior, gender and
politics, and constitutional law, with a focus on technology and politics. Her current primary research project
examines the implications of exposure to political information available via social media.
Address correspondence to: Leticia Bode, Communication, Culture, and Technology, Georgetown
University, 3520 Prospect St., NW, Suite 311, Washington, DC 20057 (E-mail: leticiabode@gmail.com).
352
Bode 353
tion on Facebook, and the potential implica- 2009), and the relationship between some types
tions of such displays (boyd & Hargittai, 2010; of social network use and civic engagement
Dwyer, Hiltz, & Passerini, 2007; Gross & (Pasek, more, & Romer, 2009). It remains to be
Acquisti, 2005). Additionally, researchers have seen, however, how the average, everyday use
begun to look into how users present them- of Facebook affects the political behavior of its
selves in an online setting, and whether and many users.
how they may share their identities with others Due to the extreme popularity and growing
(Stutzman, 2006). everyday use of Facebook within the adult pop-
Other research, more pertinent to this study, ulation of the United States, it is paramount that
has begun to examine the Facebook network, we begin to understand how members are using
both as a social tool and as a research venue Facebook, and what effects these uses have on
Downloaded by [Biblioteca Universidad Complutense de Madrid] at 06:26 25 April 2016
in and of itself (Lewis, Kaufman, Gonzalez, the rest of their lives, and particularly their polit-
Wimmer, & Christakis, 2008). Similar lines ical lives. Moreover, given that the majority of
of research are considering how the Facebook early Facebook adopters were newly suffraged
network may benefit its users by connecting and citizens (in that most of them were undergrad-
communicating with others (Donath & Boyd, uates and had limited experience voting and
2004; Subrahmanyam & Greenfield, 2008), or participating politically), an emphasis should be
harm them by creating a virtual echo chamber placed on discerning what effects Facebook use
(Sunstein, 2007). Perhaps most important to this may have on young people’s habits of political
study are a number of works that have demon- participation. Because Facebook is replicating
strated a positive relationship between online offline networks online, it is my belief that it
social network use and social capital (Ellison, will function similarly in its relationship to
Steinfield, & Lampe, 2007; Valenzuela, Park, & political participation as offline networks do.
Kee, 2009). By providing personal information from trusted
While some have begun to ponder potential others, creating community norms, and gen-
outcomes related to Facebook use, few have erating social capital, I believe the committed
yet considered the major political implications use of Facebook will increase the likelihood of
of such use. In fact, the major reported uses political participation among its users.
and gratifications related to Facebook—social This study takes a first look at how a spe-
connection, shared identities, content, social cific population—college undergraduates—are
investigation, social network surfing, and sta- using Facebook, and how that use translates
tus updating—are clearly apolitical (Joinson, into important offline behavior, including vari-
2008). Moreover, most studies that have sought ous measures of political participation and elec-
to examine the political side of Facebook have toral turnout. For the purposes of this study,
concentrated on a top-down approach, examin- political participation will be congruent with the
ing the use of Facebook by politicians seeking definition proposed by Verba, Schlozman, and
office, as well as the aggregate potential of Brady (1995): “activity that has the intent or
their Facebook content to serve as an indica- effect of influencing government action—either
tor for electoral enthusiasm (Williams & Gulati, directly by affecting the making or implementa-
2008a, 2008b). Recent work has determined that tion of public policy or indirectly by influencing
Facebook use that is explicitly political (sup- the selection of people who make those poli-
porting a candidate, creating a political event, cies” (p. 40). I propose that Facebook and other
etc.) predicts political behaviors and political social networking sites function much in the
engagement among young people, but that kind same way as traditional, face-to-face social net-
of behavior is limited to a small sector of the works do, in that they tend to facilitate the
population (Bode, Vraga, Borah, & Shah, forth- transfer of information and norms, promote inti-
coming). We also have begun to learn how macy and trust, and generate social capital,
people create their online networks with respect thus resulting in positive effects on political
to their political ideologies (Gaines & Mondak, participation.
354 JOURNAL OF INFORMATION TECHNOLOGY & POLITICS
political interest and attention (Verba et al., is not necessarily a positive development for
1995). Additionally, the ability of nonpoliti- democracy, as it may be generated by or lead
cal activities to predict political participation to the development of anti-democratic organi-
depends on the degree of voluntarism of the zations and networks (Putnam, 1993; Putzel,
nonpolitical activities. For instance, “Evidence 1997).
demonstrates that those civic skills performed in Recently the possibility of online generation
more voluntary NPOs [nonprofit organizations] of social capital has been proposed and tested,
exert a much greater effect on a person’s level in a variety of manners and settings. Shah,
of political participation than do those skills per- Kwak, and Holbert (2001) found that overall
formed in the less voluntary setting of the work- Internet use is a positive, though weak, pre-
place” (Ayala, 2000, p. 109). Thus voluntary dictor of civic engagement. More specifically,
Downloaded by [Biblioteca Universidad Complutense de Madrid] at 06:26 25 April 2016
more importantly, it takes the current literature their ordinary lives call for particular study”
that confirms social capital generation on social (Norton, 2004, p. 11). As Internet access and use
networking sites and extends it to see whether skyrocket, they become inserted into our ordi-
that use further predicts political behaviors we nary lives as well: “Habits appear to be taking
have traditionally associated with the existence hold on the Internet as well, making the Web less
of social capital (Brehm & Rahn, 1997; La Due of a novel exploratory activity and more of a rou-
Lake & Huckfeldt, 1998). Whereas Ellison et tine instrumental activity” (Muhlberger, 2005a,
al. (2007) and Valenzuela et al. (2009) verify p. 167; see also Anderson & Tracey, 2001).
that Facebook use generates social capital, this Equally important, many studies have shown
study asks whether that mechanism (and oth- the possibility for certain online activities to
ers) further result in increasing the likelihood of result in positive effects on political behav-
Downloaded by [Biblioteca Universidad Complutense de Madrid] at 06:26 25 April 2016
has the potential to do this as well, as friends information on the Internet (Caputo & Trichter,
and classmates may discuss issues as they per- 2005, p. 4–5). Facebook also has structures
tain to the personal and the everyday. Greater that allow users to share political and election-
personalization of information may spur partici- related information, again lowering the cost of
pation at the margins. obtaining such material.
Facebook also might be able to lower the The potential of Facebook to have positive
costs of online information in a specific way. effects on political participation is not a given,
Although the Internet itself lowers the cost of however, and this literature review would be
finding information in general, “ . . . as the incomplete without making note of certain pos-
amount of information, particularly faulty infor- sible limitations on its ability to have such
mation, proliferates on the Internet, the attention effects. Most importantly, “ . . . having access
Downloaded by [Biblioteca Universidad Complutense de Madrid] at 06:26 25 April 2016
costs of identifying useful and trustworthy infor- to the wealth of political information and com-
mation grow, as does the expertise needed to munication available through the Internet is not
evaluate it” (Muhlberger, 2005a, p. 167).1 by itself connected with participation” (Bimber,
Social network sites, and Facebook specifi- 2001, p. 61). The simple act of lowering the cost
cally, allow for a natural filter for information. of information is not responsible for increas-
Remembering that intimacy promotes trust, and ing political participation. Information must be
trust is needed for one to believe that informa- coupled with interest to result in positive par-
tion is reliable, Facebook combines these two ticipation effects. Additionally, online talk has
elements (intimacy/trust and information) to not been shown to result in opinion change
effectively reduce the cost of obtaining valuable (Muhlberger, 2005b). However, since the scope
and trustworthy information online (Huckfeldt of this study is limited to measures of partici-
et al., 1995). Moreover, the “Newsfeed” func- pation, rather than opinion, this does not pose a
tion of Facebook provides constant updates of direct problem for this particular work.
one’s friends’ activities, which further reduces By examining to what extent and under what
the cost of information. Having said that, some circumstances the use of Facebook can spur
research indicates that information flow is con- political participation, this study contributes a
strained in online environments by pre-existing new and important piece to the question of why
social networks and boundaries, suggesting that people participate in the electoral process and in
information may not be as readily shared in politics in general. More specifically, it updates
these venues as we might have expected (Cho the literature on social networks to include
& Lee, 2008). However, greater access to infor- newer manifestations of networks facilitated by
mation potentially lowers the cost of political innovation in communication technology.
participation, thus making it a more likely action
(Riker & Ordeshook, 1968).
Finally, there are clear positive correlations RESEARCH QUESTIONS AND
between various online activities and voting HYPOTHESES
behavior. Klotz (2005) has concluded that the
“Web sphere” in general has a tendency to pro- The specific purpose of this study is to deter-
mote voter mobilization. Tolbert and McNeal mine whether and how certain behaviors on
(2003) find the effect of mere access to the Facebook predict various measures of political
Internet to be quite substantial on voting turnout: participation, including voting, online political
“Internet access increased the probability of vot- participation, and offline political participation.
ing by [about] 7% in the 1996 presidential elec- In order to examine this research question
tion and the 1998 midterm election and 12.5% in in more depth, it is useful to first generate
the 2000 presidential election” (p. 183). Seeking some testable hypotheses. Broadly, I expect that
information online also seems to correlate pos- behavior on Facebook that represents engage-
itively with turning out to the polls, especially ment with a community will increase the
for new voters. In 2004, 44% of new voters likelihood of all types of political participa-
reported finding political and election-related tion. Engagement in a social network increases
358 JOURNAL OF INFORMATION TECHNOLOGY & POLITICS
Facebook use, which is a measure reflect- is doing so. This population therefore allows
ing to what extent a user is engaging with for the testing of various relationships between
the Facebook community. Because engagement specific Facebook behaviors and political out-
should increase social capital and information comes. Moreover, while this cohort is not rep-
flows (Muhlberger, 2005a; Shah et al., 2001), resentative of the entire U.S. population, it is
I hypothesize that as this measure increases, likely to represent future directions of social
political behaviors will also increase (H1). media use by the broader population, and is thus
Beyond mere engagement with the Facebook particularly advantageous to consider.
community, one might imagine that the degree Respondents were e-mailed by their respec-
to which one feels close to that community tive professors and invited to participate in a
might also encourage the creation of social cap- survey having to do with online social network-
ital and facilitate the transmission of trusted ing practices. Most students were offered a small
information and participatory norms, thus con- amount of extra credit in return for their partic-
tributing to increased political outcomes, includ- ipation in the online survey. Participation was
ing voting, online, and offline political behavior completely voluntary, and one e-mail reminder
(H2). This is supported by works in the lit- was sent to the students prior to the end of
erature that suggest that networks build trust the survey period. These e-mail prompts yielded
conducive to the transmission and acceptance of 542 separate responses.3 The sample was 90%
information and norms (Huckfeldt et al., 1995; white, 58% female, 27% Republican, and had
Klofstadt, 2005). an average age of about 20 years.
A final measure of interest is one that asks The survey consisted of 133 questions, rang-
respondents what their primary motivation for ing in scope from behaviors online, to behaviors
using Facebook might be. Answers include on Facebook, to general inclinations offline,
meeting new friends, keeping in touch with peo- as well as basic controls and demographics.
ple I otherwise wouldn’t, interacting with my Respondents reported that the survey took about
current friends, playing games, gathering infor- 20 minutes to complete. Attrition was not at
mation, and finding out about what’s going on all an issue (not a single respondent failed to
in my community. I expect that the motivations complete the survey).
of keeping in touch with people and interact-
ing with current friends would be most likely
to result in social capital creation, thus increas- Variables of Interest
ing the likelihood of the same political outcomes
listed above (H3). Before beginning the data analysis, a more
thorough description of the variables in play is
useful. Of greatest importance are those vari-
METHODS AND DATA ables of substantive interest to this study—those
having to do with Facebook behaviors hypothe-
In order to begin to assess the nature of the sized to be involved in spurring political partici-
many relationships just discussed, a survey was pation.
Bode 359
often he or she updates it, how many peo- Facebook intensity 0.495 3.52 1.12
ple the respondent has “friended” on Facebook, Facebook closeness 0.534 24.98 19.75
and how many groups he or she is a part of Time on Facebook 44.30 38.69
Motivation—new friends 0.01 0.09
were each divided into six categories reflect- Motivation—keep in touch 0.51 0.50
ing frequency or number. These measures reflect Motivation—interact 0.42 0.49
a respondent’s general Facebook use—greater Motivation—play games 0.01 0.08
scores on each suggest greater connection to Motivation—information 0.01 0.12
Downloaded by [Biblioteca Universidad Complutense de Madrid] at 06:26 25 April 2016
Political interest is ascertained using a scale behaviors. First, engaging in Facebook might
that combines three measures. Respondents just be a proxy for engaging in one’s community
were asked to what extent they agreed or dis- in general. Such behavior would also produce
agreed (on a scale of 1 to 5) with each of the social capital, facilitate information flows, and
following statements: “I closely follow politi- promote trust, though not necessarily through
cal issues,” “I am interested in political strat- the medium of Facebook, as is the interest of this
egy,” and “I consider politics to be among my study. Because of this, I included a scale measur-
hobbies.” These three measures were averaged ing a respondent’s Community Ties, composed
into a scale ranging from 1 to 5 with a mean of two measures with which respondents agreed
of 3.10 called Political Interest (S.D. = 1.08, or disagreed: “I feel ‘at home’ in the community
Chronbach’s α = 0.903). I live in,” and “I feel connected to those who
Downloaded by [Biblioteca Universidad Complutense de Madrid] at 06:26 25 April 2016
Political knowledge was determined in a sim- live around me” (1–5 agree/disagree, for a scale
ilar way. Respondents were asked four mul- ranging from 1 to 5, mean = 3.81, S.D. = 0.92,
tiple choice questions about modern politics. Chronbach’s α = 0.837).
For each question they answered correctly, they In a similar vein, the variables of interest
received one point. These points were then in this study may, in fact, only be capturing
divided by four to produce a scale ranging from the respondents’ tendency to affiliate with other
0 to 1 with a mean of 0.75, called Political people, or to be sociable. This tendency is cap-
Knowledge (S.D. = 0.24).5 tured in a scale composed of 15 items, reflecting
The habit or comfort level of talking with oth- respondents’ Sociability (1 to 5, mean = 2.55,
ers about politics might also be predictive of S.D. = 0.70, Chronbach’s α = 0.886).6
various political behaviors, and so some mea-
sure of this tendency should also be included Demographics
in the model. This variable is also the result
of a scale (ranging 1 to 6, with a mean of Finally, due to recorded differences in social
3.44) that combined the respondents’ answers media use for each, I include a measure of
to three questions asking about their comfort Gender (coded female = 1, 58.5% female)
talking about politics with friends and fam- and Party Identification, coded from 1 to 7,
ily, and with strangers, as well as the fre- Republican to Democrat (mean = 4.77, S.D.
quency with which they talk to friends and = 2.26) (Lenhart, Madden, Smith, & Macgill,
family about politics (S.D. = 0.90, Chronbach’s 2007).
α = 0.74).
In addition, I chose to include a measure of Outcome Variables
news use, in order to ensure that the relation-
ship I have hypothesized is not being distorted The outcome variables of interest consist of
by a simple tendency to keep up with cur- one’s reported voting in the 2008 presidential
rent events, some of which are certainly polit- primaries (for either party), political behavior
ical in nature. Thus, I created a News Use online, and political behavior offline. For the
scale, which is the average reported frequency voting measure, respondents were simply asked
of use of seven different news sources (rang- whether or not they voted in the recent presiden-
ing from “multiple times a day” to “virtually tial primaries (75% reported doing so).
never”), including local newspapers, national For the two political behavior measures,
newspapers, news Web sites, political blogs, respondents were asked how often they engaged
national evening news, local evening news, and in a series of political behaviors online (used e-
news/talk radio. This scale ranges from 1 to 6 mail to contact a politician; contributed money
(mean = 2.59, S.D. = 0.79, Chronbach’s α = online to a politician, social group, or cause;
0.701). forwarded a political e-mail or news story
Finally, two further possible spurious expla- to friends; signed an online petition; visited
nations exist for why Facebook behavior the Web site of a political campaign or can-
might come across as correlating with political didate; or posted comments or questions on
Bode 361
political news stories or political blogs) and Voting in the 2008 Presidential Primaries
offline [volunteered in the community; attended
a political meeting, rally, or speech; displayed To predict a respondent’s tendency to report
a campaign button, sticker, or sign; worked voting in the 2008 presidential primaries (a
for a political party or candidate (phonebank- dichotomous measure), I used a logistic estima-
ing, blockwalking, etc.); contacted a politician tion. All the variables described in detail above
by phone or mail; donated money in per- were included in this model. As can be seen
son to a politician, social group, or cause; or in Table 2, two variables of interest deserve
signed a petition (on paper)]. For each ques- particular attention. Both intensity of Facebook
tion, respondents could report frequencies rang- use and time spent on Facebook have significant
ing from “never” to “more than once a week.” effects on one’s likelihood of voting, but in
Downloaded by [Biblioteca Universidad Complutense de Madrid] at 06:26 25 April 2016
The online political behavior scale ranged from different directions. Intensity of Facebook use
1 to 5 with a mean of 1.6 (S.D. = 0.55, is positively related to voting, whereas time
Chronbach’s α = 0.765), whereas the offline spent using Facebook is negatively related
political behavior scale ranged from 1 to 5 (see Figure 1 for an illustration). This suggests
with a mean of 1.7 (S.D. = 0.52, Chronbach’s support for hypothesis H1, which posited that
α = 0.722).7 intensity of Facebook use would encourage
transmission of information and electoral norms
across the network as well as the generation of
social capital, thus leading to desirable political
ANALYSIS AND DISCUSSION behaviors.
Moreover the disparate relationships demon-
I estimated three sets of models, reflecting strated indicate that it is truly intensity of
each of the outcome variables in which I am use, and not merely use or amount of use of
interested. I then re-estimated each model with- Facebook, that is having the effect. That is, in
out the variables of interest to indicate the order to be motivated to participate electorally,
increase in predictive power achieved by includ- engagement with one’s Facebook network is
ing the variables reflecting various aspects of required, above and beyond simply spending
Facebook use. time on the site in a more superficial manner.
FIGURE 1. The effects of time spent on Facebook and intensity of Facebook use on respondent’s
likelihood of voting.
Downloaded by [Biblioteca Universidad Complutense de Madrid] at 06:26 25 April 2016
Note also that the motivation of making new removing the variables of interest, we can then
friends as the primary motivation for using see if we truly gain predictive power by includ-
Facebook has a negative effect on one’s like- ing the Facebook-oriented variables. As noted in
lihood of voting. I would suggest that this is Table 1, the Akaike Information Criterion (AIC)
because if you are primarily using Facebook for the first model was 525.95. Table 3, which
to meet people, you are less actively engag- reflects the reduced model of voting turnout, has
ing within the Facebook community, which is an AIC of 553.47.8 This suggests that the model
the action I have hypothesized does the bulk of including the Facebook variables does a better
the work of motivating participation by trans- job of estimating the true relationship than does
ferring information and norms throughout the the model lacking them.
network and creating social capital. This relates However, it is still possible that we are captur-
to H3, which predicted that motivations related ing a spurious relationship, in that the measure
to engagement would positively predict politi- of intensity of Facebook use is acting as a proxy
cal behavior. While we do not find clear support for some other activity. Perhaps, it could be
for that hypothesis, we do see the reverse—that argued, the more intense Facebook users are
motivations that detract from engaging with the being exposed to more politics on Facebook as
Facebook community have a negative impact on a result of their increased engagement, which
the likelihood of participating electorally. The would then be responsible for their increased
closeness one feels to one’s Facebook commu- likelihood of voting. In order to ascertain if this
nity does not seem to be a relevant motivator was, in fact, occurring, I included an additional
for electoral participation, and thus H2 is not variable in the original model. This variable
supported. asked respondents how often a political story
It is also reassuring that various control vari- showed up on their Newsfeed. (This is the user’s
ables we would expect to predict a respondent’s home page that features stories of what their
likelihood of voting do, in fact, appear to be sig- Facebook friends have been doing on Facebook.
nificant and in the expected direction. Political It can include news stories, notes, pictures, and
interest, community ties, and political talk all actions of their friends.) This measure, Political
increase the likelihood of turning out to vote. Facebook Exposure, ranged from 1 to 6 (“vir-
To further isolate the impact of Facebook, tually never” to “multiple times a day”) with a
I re-estimated the same model, but without mean of 1.97 (S.D. = 0.94). The re-estimated
any variables reflecting Facebook behavior. By model can be seen in Table 4. The measure
Bode 363
does not appear significant, and adds nothing of ordinary least squares (OLS) is an appropriate
substance to the predictive power of the model model to estimate. Table 5 reflects this OLS
(AIC = 525.5). This reassures us that it is estimation, with the same independent vari-
truly engagement with the community, and not ables as were included in the model predicting
merely exposure to politics, that is producing the voting.
increased likelihood of voting. Intensity of Facebook use is still a signifi-
cant, positive predictor of the outcome variable
Online Political Behavior (again supporting H1), and having one’s pri-
mary motivation to use Facebook to be meet-
The second outcome variable of interest ing new people is still a negative predictor.
reflects one’s political behavior online. Because However, it is important to note that one’s
this variable was scaled from six others, the time spent on Facebook switches signs in this
index is roughly continuous. As a result of this, model. This is not surprising, as more time
364 JOURNAL OF INFORMATION TECHNOLOGY & POLITICS
spent on Facebook means more time spent Again, it is useful to estimate the model with-
online in general, thus giving more opportunity out the Facebook variables to see what effect
and inclination to engage in political behav- they are having on the predictive power in
ior online, rather than offline (remembering that general. A reduced model reflecting this change
those two outcome variables are not highly can be seen in Table 6. As indicated in the final
correlated). line of each of the tables, the reduced model has
Apart from that, much of the model stays a lower adjusted R2 value, indicating that it has
very similar to the model estimating turnout, in less predictive power than the full model includ-
Table 2. This indicates that the important predic- ing the Facebook variables. The full model does
tors hold up across various behaviors related to five percentage points better than the reduced
political participation. model, or 14% better overall.
Bode 365
transmitting norms and information in this new introductory statistics class will tell you, cor-
type of network. The null finding could also be relation does not equal causation. We cannot
an artifact of the way it was measured. While decisively determine the direction of causality
the proportion of close friends and family in in any of these models—it is just as possible
one’s Facebook network reflects the proportion that political behavior leads to more intense use
of the network to which one feels close, it does of Facebook as it is that Facebook use leads to
not necessarily reflect how close one actually political behavior. However, the mechanism by
feels to the community. Future research might which political behavior might lead to intense
re-operationalize this variable with a more direct Facebook use is unclear, whereas the relation-
measure of how close one feels to those in the ship between Facebook network engagement
network and the network as a whole. and political behavior is both intuitive and based
The third hypothesis predicted that motiva- in decades of theory relating to offline social net-
tions related to engagement with the Facebook works, making it the more plausible explanation
community would be most likely to spur politi- of the relationship we see.
cal participation. While there was no clear evi- Of course, survey data themselves are some-
dence of this (neither keeping in touch with peo- what suspect, as we can never be certain whether
ple one otherwise wouldn’t nor interacting with recollections are an accurate portrayal of the
my current friends were significant predictors of reported behaviors (see for instance, Norton,
political participation), there is some evidence 2004, p. 117). Additionally, these data capture
that other motivations might decrease the likeli- only one point in time, one social network site
hood of political participation. Specifically, the (though by far the most used), and a single
motivation to meet new people on Facebook was university. Again, this sample was deliberately
negatively related to voting and online political chosen due to the level of Facebook use in the
participation. This suggests that intimacy may undergraduate population, and is likely unbi-
still play a role. Meeting new people does not ased for that population. However we cannot
seem to result in the type of deep engagement generalize to all Facebook users due to the
necessary to facilitate catalysts of political par- nature of the sample. Because of this, future
ticipation. This also may be a reflection of those research might include both longitudinal studies
who use Facebook primarily for business net- using a sample outside the undergraduate popu-
working purposes, thus decreasing the intimacy lation, and track users before and after they join
they feel and they level of engagement in which Facebook, monitor their intensity of use over
they participate in their Facebook community. time, and engage in qualitative research, con-
As with any study, this one has its share sisting of interviews or participant observation,
of limitations. As any undergraduate in an that might give a more well-rounded view of the
Bode 367
relationship between Facebook use and political Scale (Cheek & Buss, 1981; Leary, 1983). For a complete
behavior. list of these items, please contact the author.
While this study is limited in some ways, 7. Despite this similarity of mean, the correlation
the general conclusion is important. Intensity between the two variables is only 0.08, suggesting that
more of one does not indicate more of the other.
of Facebook engagement has clear, significant, 8. AIC is a measure of goodness of fit that both
and robust effects on several types of political rewards fit and penalizes the number of estimated parame-
participation, even above and beyond the normal ters, thus discouraging against model overfitting. A smaller
predictors. While more work certainly needs to AIC reflects a better fit (Burnham & Anderson, 2002). It is
be pursued in this area, this study represents thus preferable to other criteria for model comparison that
an important first step in understanding how a do not account for overfitting.
staple of online life may affect the political par-
Downloaded by [Biblioteca Universidad Complutense de Madrid] at 06:26 25 April 2016
presidential election. Paper presented at the annual Teaching arts and science with the new social
meeting of the American Political Science Association, media (Cutting-edge technologies in higher education,
Washington, DC. Vol. 3) (pp. 61–87). Bingley, England: Emerald Group
Cheek, J. M., & Buss, A. H. (1981). Shyness and sociabil- Publishing Limited.
ity. Journal of Personality and Social Psychology, 41, Huckfeldt, R. (2001). The social communication of polit-
330–339. ical expertise. American Journal of Political Science,
Cho, H., & Lee, J. (2008). Collaborative information seek- 45(2), 425–438.
ing in intercultural computer-mediated communication Huckfeldt, R., Beck, P., Dalton, R., & Levine, J. (1995).
groups: Testing the influence of social context using Political environments, cohesive social groups, and the
social network analysis. Communication Research, communication of public opinion. American Journal of
35(4), 548–573. Political Science, 39(4), 1025–1054.
Dahlgren, P. (2005). The Internet, public spheres, and Hwang, H., Schmierbach, M., Paek, H., Gil de Zuniga,
Downloaded by [Biblioteca Universidad Complutense de Madrid] at 06:26 25 April 2016
political communication: Dispersion and deliberation. H., & Shah, D. (2006). Media dissociation, Internet
Political Communication, 22, 147–162. use, and antiwar political participation: A case study of
Donath, J., & Boyd, D. (2004). Public displays of connec- political dissent and action against the war in Iraq. Mass
tion. BT Technology Journal, 22(4), 71–82. Communication and Society, 9(4), 461–483.
Dwyer, C., Hiltz, S., & Passerini, K. (2007, August). Trust Joinson, A. (2008). Looking at, looking up or keep-
and privacy concerns within social networking sites: ing up with people? Motives and use of Facebook.
A comparison of Facebook and MySpace. Proceedings In Conference on Human Factors in Computing
of the Thirteenth Americas Conference on Information Systems—Proceedings of the Twenty-Sixth Annual
Systems, Keystone, Colorado. Retrieved from http:// SIGCHI Conference on Human Factors in Computing
csis.pace.edu/~dwyer/research/DwyerAMCIS2007.pdf Systems (pp. 1027–1036). New York: Association for
Ellison, N. B., Steinfield, C., & Lampe, C. (2007). The ben- Computing Machinery.
efits of Facebook “friends:” Social capital and college Kenny, C. (1992). Political participation and effects from
students’ use of online social network sites. Journal of the social environment. American Journal of Political
Computer-Mediated Communication, 12(4), 1–34. Science, 36(1), 259–267.
Eulau, H., & Rothenberg, L. (1986). Life space and social Klofstad, C. A. (2005). “The art of associating:” The cen-
networks as political contexts. Political Behavior, 8(2), tral role of peers in civic life. Unpublished dissertation,
130–157. Harvard University, Cambridge, MA.
Facebook.com site info. (2012). Alexa. Retrieved from Klotz, R. (2005, September). Internet campaigning and
http://www.alexa.com/siteinfo/facebook.com participation. Paper presented at the annual meet-
Flanagin, A. J., & Metzger, M. J. (2001). Internet ing of the American Political Science Association,
use in the contemporary media environment. Human Washington DC.
Communication Research, 27, 153–181. Kwak, N., Shah, D., & Holbert, R. (2004). Connecting,
Fox, Z. (2011, November 18). Who is an aver- trusting, and participating: The direct and interac-
age Facebook user? Mashable. Retreived from tive effects of social associations. Political Research
http://mashable.com/2011/11/18/facebook-stats/ Quarterly, 57(4), 643–652.
Gaines, B. J., & Mondak, J. J. (2009). Typing together? La Due Lake, R., & Huckfeldt, R. (1998). Social capi-
Clustering of ideological types in online social net- tal, social networks, and political participation. Political
works. Journal of Information Technology & Politics, Psychology, 19(3), 567–584.
6(3–4), 216–231. Leary, M. R. (1983). Social anxiousness: The construct and
Green, M., & Brock, T. (2005). Organizational mem- its measurement. Journal of Personality Assessment,
bership versus informal interaction: Contributions to 47, 66–75.
skills and perceptions that build social capital. Political Lenhart, A., Madden, M., Smith, A., & Macgill, A. (2007,
Psychology, 26(1), 1–25. December 19). Teens and social media. Washington,
Gross, R., & Acquisti, A. (2005, November). Information DC: Pew Internet & American Life Project. Retrieved
revelation and privacy in online social networks. Paper from http://www.pewinternet.org/Reports/2007/Teens-
presented at the Workshop on Privacy in the Electronic and-Social-Media.aspx
Society 2005, Alexandria, Virginia. Lewis, D., Kaufman, J., Gonzalez, M., Wimmer, A., &
Harvard University Institute of Politics. (2012). Survey of Christakis, N. (2008). Tastes, ties, and time: A new
young Americans’ attitudes toward politics and public social network dataset using Facebook.com. Social
service: 21st edition. Retrieved from http://www.iop. Networks, 30(4), 330–342.
harvard.edu/sites/default/files_new/spring_poll_12_ MacDougall, R. (2005). Identity, electronic ethos, and
exec_summ.pdf blogs: A technologic analysis of symbolic exchange on
Helvie-Mason, L. (2011). Facebook, “friending,” and the new news medium. American Behavioral Scientist,
faculty-student communication. In C. Wankel (Ed.), 49(4), 575–5995.
Bode 369
McClurg, S. (2003). Social networks and political partici- Meetup.com. Paper presented at the annual meet-
pation: The role of social interaction in explaining polit- ing of the American Political Science Association,
ical participation. Political Research Quarterly, 56(4), Washington, DC.
449–464. Scheufele, D., Hardy, B., Brossard, D., Waismel-Manor, I.,
Muhlberger, P. (2005a). Human agency and the revitaliza- & Nisbet, E. (2006). Democracy based on difference:
tion of the public sphere. Political Communication, 22, Examining the links between structural heterogene-
163–178. ity, heterogeneity of discussion networks, and demo-
Muhlberger, P. (2005b, September). Attitude change cratic citizenship. Journal of Communication, 56(4),
in face-to-face and online political deliberation: 728–753.
Conformity, information, or perspective taking? Paper Shah, D., Cho, J., Eveland, W., & Kwak, N. (2005).
presented at the annual meeting of the American Information and expression in a digital age: Modeling
Political Science Association, Washington, DC. Internet effects on civic participation. Communication
Downloaded by [Biblioteca Universidad Complutense de Madrid] at 06:26 25 April 2016
Mutz, D. C., & Martin, P. S. (2001). Facilitating commu- Research, 32(5), 531–565.
nication across lines of political difference: The role of Shah, D., Kwak, N., & Holbert, R. (2001). “Connecting”
mass media. American Political Science Review, 95(1), and “disconnecting” with civic life: Patterns of Internet
97–114. use and the production of social capital. Political
Mutz, D. C., & Martin, P. S. (2002a). Cross-cutting Communication, 18, 141–162.
social networks: Testing democratic theory in practice. Stutzman, F. (2006, April). An evaluation of identity-
American Political Science Review, 96(2), 111–126. sharing behavior in social network communities. Paper
Mutz, D. C., & Martin, P. S. (2002b). The consequences presented at the iDMAa and IMS Code Conference,
of cross-cutting networks for political participation. Oxford, Ohio.
American Journal of Political Science, 46(4), 838–855. Subrahmanyam, K., & Greenfield, P. (2008). Online com-
Norton, A. (2004). 95 theses on politics, culture, and munication and adolescent relationships. Children and
method. New Haven, CT: Yale University Press. Electronic Media, 18(1), 119–146.
Pasek, J., more, e., & Romer, D. (2009). Realizing the Sunstein, C. (2007). Republic.com 2.0. Princeton, NJ:
social Internet? Online social networking meets offline Princeton University Press.
civic engagement. Journal of Information Technology & Tolbert, C., & McNeal, R. (2003). Unraveling the effects
Politics, 6(3–4), 197–215. of the Internet on political participation? Political
Protalinski, E. (2012, February 1). Facebook has Research Quarterly, 56(2), 175–185.
over 845 million users. ZDNet. Retrieved from Valenzuela, S., Park, N., & Kee, K. (2009). Is there social
http://www.zdnet.com/blog/facebook/facebook-has- capital in a social network site? Facebook use and
over-845-million-users/8332. college students’ life satisfaction, trust, and participa-
Putnam, R. (1993). Making democracy work: Civic tra- tion. Journal of Computer-Mediated Communication,
ditions in modern Italy. Princeton, NJ: Princeton 14, 875–901.
University Press. Verba, S., Schlozman, K., & Brady, H. (1995). Voice and
Putnam, R. (1995). Tuning in, tuning out: The strange dis- equality. Cambridge, MA: Harvard University Press.
appearance of social capital in America. PS: Political Weatherford, M. (1982). Interpersonal networks and polit-
Science and Politics, 28(4), 664–683. ical behavior. American Journal of Political Science,
Putzel, J. (1997). Accounting for the “dark side” of social 26(1), 117–143.
capital: Reading Robert Putnam on democracy. Journal Williams, C., & Gulati, G. (2008a, March). The political
of International Development, 9, 939–949. impact of Facebook: Evidence from the 2006 midterm
Riker, W. H., & Ordeshook, P. C. (1968). A theory of elections and 2008 nomination contest. Politics and
the calculus of voting. The American Political Science Technology Review, 11–21.
Review, 62(1), 25–42. Williams, C., & Gulati G. (2008b, August). What is a social
Rogers, R. (2005). Poignancy in the US political blog- network worth? Facebook and vote share in the 2008
sphere. New Information Perspectives, 57, 356–368. presidential primaries. Paper presented at the annual
Sander, T. (2005, September). E-associations? Using meeting of the APSA 2008 Annual Meeting, Boston,
technology to connect citizens: The case of Massachusetts.