Professional Documents
Culture Documents
To cite this article: Leticia Bode & Kajsa E. Dalrymple (2014): Politics in 140 Characters or Less: Campaign
Communication, Network Interaction, and Political Participation on Twitter, Journal of Political Marketing, DOI:
10.1080/15377857.2014.959686
Disclaimer: This is a version of an unedited manuscript that has been accepted for publication. As a service
to authors and researchers we are providing this version of the accepted manuscript (AM). Copyediting,
typesetting, and review of the resulting proof will be undertaken on this manuscript before final publication of
the Version of Record (VoR). During production and pre-press, errors may be discovered which could affect the
content, and all legal disclaimers that apply to the journal relate to this version also.
Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained
in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no
representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the
Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and
are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and
should be independently verified with primary sources of information. Taylor and Francis shall not be liable for
any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever
or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of
the Content.
This article may be used for research, teaching, and private study purposes. Any substantial or systematic
reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any
form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http://
www.tandfonline.com/page/terms-and-conditions
Politics in 140 Characters or Less: Campaign Communication, Network Interaction,
and Political Participation on Twitter
Abstract
The methods by which politicians and policy makers communicate with the public are
landscape, this study considers the case of Twitter. Specifically, the authors conduct a
survey of political Twitter users, in order to understand their use of the medium and their
political behaviors within it. Results indicate that political Twitter users are more
interested in and engaged in politics in general, and less trusting of the mainstream
media. Moreover, the study investigates the extent to which followers of a campaign may
each other, the methods through which politicians and policy makers engage with the
public must also adapt. Historically, this process has entailed strategic communication
campaign tactics such as creating candidate websites, facilitating and engaging in online
1
fundraising, and employing email lists to reach potential supporters. More recently, the
emergence of social media has prompted further adaptation on the part of political
candidates. For example, in 2008, then-candidate Obama was praised for his campaign’s
ability to embrace social media in order to reach parts of the electorate that might
otherwise have been missed: “By bolting together social networking applications under
the banner of a movement, they created an unforeseen force to raise money, organize
locally, fight smear campaigns and get out the vote that helped them topple the Clinton
Downloaded by [North Dakota State University] at 23:54 28 February 2015
machine and then John McCain and the Republicans” (Carr, 2008).
This devotion to social media in the campaign context is a trend that promises only to
increase in upcoming elections. In the 2010 midterm elections, for instance, nearly every
candidate running for office at the national level had a Twitter account focused on
reaching out to current supporters and recruiting new followers. This trend is not only
perceived as a way of gathering supporters’ opinions, but also as a new method for
reaching out to swing voters. In fact, the Congressional Management Foundation recently
reported that 72 percent of congressional staffers “believe that social media allows their
Members to reach people that they had previously not communicated with” (2011).
Although some would dispute this interpretation (Morozov 2013), adoption of social
media has the potential to change the playing field for candidates who were previously
restricted to target audiences and voting demographics, and assist in swaying formerly
2
Citizens are also taking notice of the role that social media can play in political
campaigns, and are engaging with the new media available to them. This is highlighted
by research indicating that during the 2010 campaign, 22% of adult Internet users
reported using Twitter or social networking sites for political purposes (Pew Internet &
promises to grow with increased access to the Internet and to social media, as well as
with generational replacement. As a recent article puts it, “in purely quantitative terms,
Downloaded by [North Dakota State University] at 23:54 28 February 2015
mediated public communication, and the sheer number of tweets…testifies to its use”
(Larsson & Moe, 2012). Although some research suggests interactivity and content
production is beyond the scope of most media consumers (Larsson 2013), trends point to
the importance of social media platforms, like Twitter, in mobilizing public groups and
importance of Twitter is not only in its saturation rate in the American citizenry, but also
in the attention paid to it by elites, which is extensive and growing (Marwick & Boyd,
explicitly for political purposes, and politicians are flocking to these media to interact
with such social groups, we know very little about the political dynamics of this new
media genre. In order to address this lack of understanding, this study focuses on
increasing our knowledge of political social media users and the politicians they choose
to follow. Using a novel sampling technique, we consider only those Twitter users who
3
have chosen to follow a candidate for national political office on Twitter, and employ
survey data to examine how they compare to the electorate at large and other highly
how they use Twitter for political purposes, how those social media behaviors vary
within the population of users, and how our methodology can help identify influential
subpopulations within American politics and savvy new media users amongst political
candidates.
Downloaded by [North Dakota State University] at 23:54 28 February 2015
Current literature has limited its consideration of Twitter to a broad, aggregate view of
the network itself. The majority of extant research has two major concerns: first, which
users have the most influence, and second, how information spreads throughout the
network. Specifically, much work has focused on the way political information is
disseminated via Twitter, both independently and compared to other media (Yang and
Counts, 2010a). Research has demonstrated that network structure can affect the spread
of information (Lerman and Ghosh, 2010), and that speed, scale, and range, are important
factors to consider when examining the spread of information within a Twitter network
(Yang and Counts, 2010b). Moreover, initial work has demonstrated that information on
Twitter may serve as a proxy for general public opinion (O’Connor, Balasubramanyan,
Routledge, and Smith, 2010). However, current studies concerning influence on Twitter
have focused only on simple measures of indegree, retweets, and mentions (Cha,
Haddadi, Benevenuto, and Gummadi, 2010), and have failed to examine the unique
4
Much of the research on the political use of Twitter has also been limited to a focus on
most notably both what encourages members of Congress to adopt use of Twitter, and
what helps them to be “successful” in such use. For instance, Lassen and Brown (2011)
found that members are more likely to adopt Twitter if their party leaders urge them to, if
they are young, or if they serve in the Senate. Gulati and Williams determined that party
and campaign resources were the most important predictors, and later work demonstrated
Downloaded by [North Dakota State University] at 23:54 28 February 2015
that competition and the education level of a district also played a role in early adoption
(Gulati & Williams 2010; Williams & Gulati 2012). And, Chi and Yang (2010b) suggest
that adoption is driven by a desire for constituency outreach, rather than a transparency
motivation, and adoption may be accelerated by evidence of past users’ success with the
medium (Chi and Yang, 2010a). Considering these findings, factors including vote share,
funding, usage and influence may help to explain why some congressional users have
Members of Congress have been shown to use Twitter and other new media principally to
promotion (Golbeck, Grimes, & Rogers, 2010, Williams & Gulati, 2012). Additional
lines of research have challenged the idea that Web 2.0 techniques are of much use,
suggesting that “mundane internet tools (like email) are much more deeply integrated
into mobilizing practices today than emerging tools (like social networking sites)”
(Nielsen 2011). However, little research has focused on the specific role that Twitter may
5
A single study to date of which the authors are aware has examined Twitter use within
the electoral context, in an attempt to predict election outcomes. This study was limited to
a general search for mentions of political candidates and political parties in tweets
(Tumasjan, Sprenger, Sandner, and Welpe, 2010). Word count analysis of this sample of
explicitly political tweets revealed that the more frequently a candidate or party was
mentioned, the more likely electoral victory for that entity. Although this is an important
contribution to the political Twitter literature, it fails to consider the specific dynamics by
Downloaded by [North Dakota State University] at 23:54 28 February 2015
which such information is shared. Moreover, the idea of what constitutes a “political”
tweet is operationalized and restricted by the author, rather than the users themselves.
Moreover, with the exception of the recent report from Pew, asking a single question
regarding political use of Twitter combined with other social networks, no study has
considered how political Twitter use is occurring outside the realm of political elites
(Pew Internet & American Life, 2011). This is problematic for two reasons. First, the
majority of Twitter users certainly are not political elites themselves (Twitter had about
200 million users worldwide as of January 2013, with roughly 30%, or 60 million,
residing in the United States (Twitter 2013)). As a result, considering the use of Twitter
by such elites neglects a large part of the population using Twitter for political purposes.
understand the relationship between political elites and their constituents, supporters, or
opponents within the realm of Twitter. We cannot fully comprehend how elites use
6
CONSIDERING OTHER HIGHLY POLITICALLY INTERESTED
SUBPOPULATIONS
other new media outlets adopted by politicians in recent memory. As a comparative case
study, we think political blogs are particularly relevant. Like Twitter, most blog creators
and blog readers are not creating or reading for political purposes. Even so, political
topics found their way into blogs, eventually resulting in a robust universe of widely-read
Downloaded by [North Dakota State University] at 23:54 28 February 2015
political blogs. i Similarly, we can imagine politics finding its way into the Twitterverse
The population of political blog users, not surprisingly, is extremely high in terms of
political interest and engagement (Gil de Zuniga, Veenstra, Vraga, and Shah, 2010).
Based on this previous research, we question whether the political Twitterverse might
also be tapping into the same subpopulation of people who originally flocked to political
blogs:
RQ1: Will Twitter users who follow political candidates be higher in terms of (a)
political interest and (b) political knowledge than the general population?
The question of whether political participation correlates with blog exposure is also
relevant to our study. Research suggests that political blog readers are highly
participatory, and that this participation occurs in both new online expressive styles of
participation, as well as more traditional offline formats (Gil de Zuniga, Veenstra, Vraga,
7
and Shah, 2010). Although we might expect that exposure to political blogs is merely a
proxy for political interest in general, in fact, political blog use predicts political
participation above and beyond the usual suspects, including political interest. Because
H1: Active political use of Twitter will positively predict online and offline political
participation.
Downloaded by [North Dakota State University] at 23:54 28 February 2015
We might further expect Twitter to serve as a conduit for the flow of information,
Crutchfield, 1955; Lakin & Chartrand, 2003; Mackie, 1987; Katz & Lazarsfeld, 1955;
Keller & Berry, 2003). The role of opinion leaders in the transmission of information and
encouragement of behavior adoption is highlighted in both the two-step flow theory and
in the theory of planned behavior (see, Ajzen, 1988, 1991, 2002; Katz & Lazarsfeld,
social networks may serve as a better way of informing the public than more formal
8
This literature suggests that Twitter users are likely to influence and be influenced by
their respective networks. As a result, it is likely that political candidates have a great
amount of influence over the constituents that follow them, and those users may
subsequently influence their own followers. While we do not have clear directional
expectations for the way in which this might occur, we propose a research question based
on these theoretical frameworks in order to begin to address the important potential for
Methods
In order to understand the population of users who employ Twitter for political
purposes, we first must identify them. For studies dealing with past communication
technologies, such as blogs or the Internet in general, the standard protocol for
determining users of such media has been the self-report (see, Pew Internet & American
Life, 2011). This is generally an effective strategy, but suffers from the problem of self-
identification. That is, we have no objective way of distinguishing between someone who
claims to read political blogs but does not, and someone who actually reads political
blogs. We rely on the respondents to honestly and accurately describe the extent to which
9
In order to combat this problem, we constructed our sample from within the
context of Twitter, rather than using an externally conducted survey. We first created a
universe of nearly all political candidates for national office in the 2010 midterm
elections (all those running for Senate (75 candidates) or Governor (76 candidates) and a
of over 500 candidatesii. We then restricted this sample to only those who had active
Twitter accountsiii that we could identify from some objective source (usually the
Downloaded by [North Dakota State University] at 23:54 28 February 2015
candidates for Senate, 74 candidates for governor, and 249 candidates for the House of
Representatives.
After generating this sample of candidates for public office using Twitter, we then
randomly sampled from their followers on Twitter. Over 10,000 followers of political
candidates were contacted within Twitter and invited to complete an online survey with
the following prompt: “@username: U Wisconsin study on political tweets. You've been
Participation was entirely voluntary, but the response rate was surprisingly high, with
1076 respondents completing the survey in its entirety (1654 respondents began but did
not complete the survey, representing a 16.5% response rate and a 10.8% overall
completion rate). Among those surveyed, 837 volunteered their Twitter user name,
allowing us to pair their survey responses with certain behavioral and network measures,
10
which were obtained the week before the election (October 30, 2010) unless otherwise
noted.
Measures
Demographics
characteristics, including age (mean = 36.8 years), race (sample is 88.1% white), gender
Downloaded by [North Dakota State University] at 23:54 28 February 2015
questions, and scoring them as correct or incorrect. The measure is the percent answered
correctly.iv
Media Variables
respondents might be exposed to outside the context of Twitter, and to understand their
orientation toward that media, we include a large battery of 12 media use variables, v
scaled into a mean index of use (1 to 7, mean = 4.8, chronbach’s α = 78). Trust in the
media was also measured, to determine to what extent feelings about the mainstream
media affect behavior on Twitter. Two items: “I trust the information I find in the news”
and “Most news are biased against my views” were averaged to create an index of media
11
Twitter Variables
Four main variables of interest relate to how respondents report using Twitter.
First, we simply asked them, along the lines of the media use questions above, how often
in a week they read political tweets (1 to 7, mean = 6.07, SD = 1.30). We further asked
them questions about trust in Twitter, asking to what extent they agreed with the
In order to assess what types of political behaviors Tweeters may engage in, we asked
exploratory factor analysis suggested two main dimensions of political Twitter use. First
is what we call active political tweeting, which consists of three measures reflecting how
often the respondent had retweeted a tweet from a politician, tweeted about politics, or
followed a link from a tweet to a story at a news site (scaled 1 to 7, mean = 5.73,
chronbach’s α = .69). This high mean suggests political Twitter users are actively
engaging in Twitter, rather than merely following politicians but doing little else.
The second dimension of political Twitter activities reflects the decision to expose
oneself to political disagreement via Twitter. It consists of two measures indicating how
often a respondent reports reading or following tweets from someone whose political
views he or she disagrees with (scaled 1 to 7, mean = 5.12). Again, the mean is quite
high, suggesting that concerns of a Twitter echo chamber may be overstated. Generally,
12
it seems political Twitter users choose to expose themselves to a variety of political views
Outcome Variables
Twitter, and thus needed to measure traditional political participation. Past research has
constructs, so we conduct separate analyses of each (Gil de Zuniga, Veenstra, Vraga, and
Shah, 2010).
Building on previous research (see, Gil de Zuniga, Puig-i-Abril, and Rojas, 2009;
Mossberger, Tolbert, and McNeal, 2008; Tolbert and McNeal, 2003), online political
participation consists of twelve items mean indexed into a single measure vii (1 to 7, mean
= 4.17, chronbach’s α = .90), and offline political participation consists of five items
mean indexed into a single measure viii (1 to 7, mean = 4.78, chronbach’s α = .85).ix
participation across the blogosphere and Twitterverse (Gil de Zuniga, Veenstra, Vraga,
and Shah, 2010). We recognize that items within our online and offline participation
indices may be differentiated into distinct clusters of activities (Gibson and Cantijoch,
2013), however, due to the novelty of our sample we chose to examine the effects of
13
provide a baseline understanding of whether people replace behaviors with political
ANALYSIS
The analysis of our data consists of four main parts. First we consider the simple make-up
of this unique population of Twitter users. Then we proceed to question how Twitter
behaviors relate to traditional measures of online and offline political participation. After
Downloaded by [North Dakota State University] at 23:54 28 February 2015
When considering the impact of a new medium it is essential to first consider the
examine two questions: first, what does this audience look like? And second, in what
ways does the average member of the political Twitterverse compare to the general U.S.
population? As displayed in Table 1, differences between political Twitter users and the
Demographically, the story is mixed. Although the mean age is almost identical among
political Twitter users and the general population, political Twitter users are not reflective
of the United States when it comes to race or gender. Rather political users tend to be a
14
predominantly white population (88.1% versus 79.6%), and heavily male (36.3% female
versus 50.8% in the general population). This is a reversal from previous studies of
Twitter use in general, which have found that men and women are roughly equally likely
to use the medium (or that women are even more likely to use it – see Kimbrough,
Guadagno, Muscanell, and Dill 2013), and that black users are overrepresented on
Twitter as compared to their overall numbers in the population (Pew 2012, Hargittai &
Walejko 2008). Moreover, political users tend to be higher in both education (70.5% with
Downloaded by [North Dakota State University] at 23:54 28 February 2015
a bachelor’s degree versus 24.4% in the broader population) and income ($62,000 annual
household income versus $49,777 in the general population). This result is unsurprising
given the requirement to have Internet access in order to use Twitter, and the desire to
engage in a new medium like Twitter. That is, such demographics are very much
Politically, it is clear that our sample is on the extreme end of the spectrum in terms of
political interest, knowledge x, and voting behavior (as indicated in Table 1). Whereas the
average American tends away from such attitudes and behaviors, our sample is off the
between that use and other behaviors of interest. In a democratic society, these are
generally the behaviors that contribute to democracy in some way, such as voting,
15
In accordance with our hypothesis, the results shown in Table 2 indicate that
scholars have argued that conversational diversity is a key factor in the development of
an informed and active democracy (McKuen, 1990, Huckfeldt, Mendez, and Osborn,
Downloaded by [North Dakota State University] at 23:54 28 February 2015
2004, Scheufele, Nisbet, Brossard, and Nisbet, 2004), our findings support research that
suggests that exposure to opposing viewpoints may in fact have a demobilizing effect on
individuals (Hwang et al., 2007; Mutz, 2006). Considering this finding, it is important to
examine how the infrastructural aspects of Twitter may influence both conversational
diversity and participation. In this case, due to the grand social networking capabilities of
the Twitter interface, users may be more likely to encounter opposing viewpoints online
vs. offline. It is clear from our findings, however, that this diversity may drive users away
from online participation activities rather than encourage them to seek them out. This has
important implications for political marketers who are interested in incorporating social
political interest are positively associated with online political participation, whereas race
(where higher numbers indicate non-white) and media trust are negatively associated
16
Offline participation (Table 3) shows similar relationships. As predicted, actively
income and media use are positively related to participating offline, and race and media
trust are negatively related. Two interesting relationships are worth noting for this
considering past findings which indicate knowledge and awareness as positive predictors
of participation (Galston, 2001), and will be discussed further in the discussion section.
Finally, our findings indicate that age is negatively associated with participation,
suggesting older users in our sample are less likely to participate than their younger
counterparts. xii
Because we have determined that active political tweeting is an important predictor for
more detailed survey measures reflecting perceptions about Twitter and politics, as well
Additional Measures
1
It is possible this finding is a function of Twitter’s makeup in general, of the generally more affluent,
educated, and white.
17
We first thought political tweeting might be a function of the effects an individual feels
she is having on the Twitter community at large, and thus sought to measure perceptions
of self-Twitter influence. Twitter influence is a 7 point scale created from three measures
reflecting agreement with the following statements: “I think my Tweets would have a
significant influence on Twitter users,” and “I think my Tweets would have a significant
We further believed that motivations for using Twitter might also affect whether and how
analysis indicated separated into three main components. First, users may be motivated to
statements: “I use Twitter to give me more facts to back up my opinions,” “I use Twitter
to help me stay informed about what is happening in the world,” “I use Twitter to learn
new ideas,” “I use Twitter to find others who think like me,” and “I use Twitter to find
α=.74). Users might also be motivated specifically by politics when engaging on Twitter.
A second component was revealed to include three concepts: “I use Twitter to learn how
politicians stand on issues,” “I use Twitter to find ways to participate in the political
process,” and “I use Twitter to organize or engage in collective action for a political or
motivated to seek out entertainment on Twitter. This was captured by a single measure:
18
For a subsample of our total survey population (those respondents who offered their
Twitter handle during the course of the survey), we were further able to pair Twitter
behavioral measures with our survey data. The Twitter API provided us access to one
would expect that greater numbers of followers would encourage additional tweeting by
users. That is, the greater number of followers a tweeter has, the more likely she is to
Downloaded by [North Dakota State University] at 23:54 28 February 2015
endeavor to provide them content. xvi This subsample was fairly well-connected, with a
Analysis
regression in order to determine what predicts active tweeting. Results may be found in
Table 2.
In terms of basic results, we find that education and political interest predict active
political tweeting, as we might well expect them to. Political tweeting is not related to
political knowledge, suggesting it is not objective knowledge alone that prompts sharing
political thoughts and information with the Twitterverse. Media use is also positively
related to tweeting about politics, although trust in the media is negatively related,
suggesting that the less trusting one is of the mainstream media, the more likely she is to
tweet about politics. This might be a corrective measure – if the media cannot get the job
19
done, I will take its place and become a sort of citizen journalist. Interestingly, it is
distrust of the mainstream media but not trust in Twitter that drives tweeting about
politics. This suggests motivations for using Twitter are driven by its appeal as an
alternative media outlet, rather than as a particularly appealing platform in its own right.
political tweeting, and this is generally born out. Frequency of reading political tweets,
Downloaded by [North Dakota State University] at 23:54 28 February 2015
the number of political accounts followed, and exposure to disagreement within Twitter
Motivations for using Twitter also seem to play a role in motivating political tweeting.
Predictably, users motivated to use Twitter for political reasons are more likely to tweet
about politics. Less obvious is the negative relationship between information seeking
motivations on Twitter and political tweeting. Perhaps this measure captures the desire to
seek more general information, rather than the relatively narrow realm of political
tweeting, suggesting that users may or may not use Twitter recreationally, with no
having influence, and not the potential to actually influence a large number of others,
20
which seems to play a role in motivating political tweeting, suggesting perhaps that
perceptions of influence are overestimated by some users. This finds support in other
studies which have suggested that actual influence on Twitter does not only come from
users with the highest number of followers (Cha, Haddadi, Benevenuto, and Gummadi
2010).
Twitter, and what those communities look like in terms of realized influence. In order to
assess this, we rely on the behavioral measures gleaned from Twitter itself. We first
Twitter) by each of our respondents. We then aggregated these by respondent, and nested
each respondent in one or more candidate communities, depending upon whether or not
they were a follower of that politician at the time of our collection. xviii This results in both
a number of followers within our sample for each candidate, as well as the total number
of retweets by any of those followers, aggregated by candidate. These totals may be seen
in Table 3.
Influence could potentially manifest in two ways – either the number of followers
a candidate has, or the proclivity of the candidate or his followers to tweet useful or
interesting information, prompting it to spread to others. That is, a candidate may exert
influence directly – by simply reaching out to a large number of people with her tweets –
or indirectly – by reaching further into the Twitter network by virtue of the influence of
21
her followers. In comparing the number of followers and the number of retweets accrued
exert influence in 2010. Table 3 is sorted by the total number of retweets nested within
each candidate, and it can be seen that this is not directly parallel to the number of
followers. While followers and retweets are relatively highly correlated (.55), some of the
candidates with the largest following in our sample do not break the top ranks of
retweets. Rubio from Florida, for instance, is number one in terms of raw numbers of
Downloaded by [North Dakota State University] at 23:54 28 February 2015
campaign strategy, this suggests that candidates should target not only a large following,
but also an active and interested following, in order to maximize their influence on
Twitter.
CONCLUSIONS
The purpose of this study was to extend our understanding of the political use of Twitter,
the implications of this social medium for political participation, and how communities of
political influence may emerge within the political Twitterverse. Overall, our hypotheses
find strong support from our data. Specifically, our findings suggest that Twitter contains
a unique audience for political elites to target. Political Twitter users are in many ways
the ideal subpopulation with which elites might choose to communicate. They are
extremely interested in politics, very likely to turn out at the polls, and wealthy enough to
contribute to campaigns. Moreover, political Twitter use makes it even more likely for
22
Our novel sample gives us the advantage of clearly identifying political users of Twitter
in an objective manner. Rather than relying on self-reports, we can be assured that our
sample uses Twitter in political ways at least occasionally. However, this restricts our
sample to people who choose to follow politicians. There are certainly Twitter users who
tweet about politics and yet do not follow any politicians or candidates for public office
on Twitter, and these people would not be included in our sample. Moreover, our sample
was recruited only by voluntary participation. Accordingly, we are most likely capturing
Downloaded by [North Dakota State University] at 23:54 28 February 2015
the most politically interested possible sample, and excluding less intense political users
of Twitter. This might explain the strength of the relationships we find, but we believe
they would persist in other samples of political Twitter users. Future research should
findings.
While we have learned a great deal in this preliminary study, we implore future research
to examine these limitations in order to better assess the behaviors and demographics of
political tweeters, the influence they may exert, and the communities they embody.
Incorporating further behavioral measures with survey data is a logical and likely fruitful
Keeping these limitations in mind, there are still several conclusions that can be
drawn from our findings. For instance, our results point to an unexpected relationship
23
participation. Although previous research has indicated that the use of media to inform
oneself about politics is usually a positive predictor of political participation (see Galston,
2001), this finding suggests that using Twitter as a method of gaining information about
politics may not be as effective as more traditional media. This may be explained by the
passive nature of the medium. For instance, due to the 140-character limit of tweets,
newspaper. Similar to findings that have highlighted the limited effects of television use
Downloaded by [North Dakota State University] at 23:54 28 February 2015
on political knowledge and participation (Dalrymple, 2006; Chaffee and Kanihan, 1997),
Twitter users may not be actively processing the information that they read, but rather
Tweeters are exposed may or may not be retained for future use in political discussion or
Perhaps most importantly, our study points to some interesting findings regarding how
influential various follower groups may be on Twitter. Specifically, our results indicate
motivating active political tweeting than the potential to actually influence a large
Additionally, the results presented above highlight the importance in considering both a
candidate’s number of followers and the activity of those followers. More specifically,
our findings suggest that candidates with a large following may not be contributing much
24
to the two-step flow of information if their followers are not tweeting and being
retweeted in order to continue the flow of information deeper into the network. Although
candidates are engaging on Twitter with others beyond the candidate community, social
media campaigns may not be as successful in reaching less active and interested
populations, like swing voters or “soccer moms”. As a result, it is essential that future
campaigns focus on both recruiting of followers, and encouraging tweeting among those
Downloaded by [North Dakota State University] at 23:54 28 February 2015
followers. Such behavior will only contribute to the flow of information and may help
source of information and a form of political expression, it is crucial that future research
consider the potential effects of this medium as well as the types of citizens that are using
it. Using a novel sampling technique, this study identified the ‘average’ political tweeter
and further analyzed the potential effects of Twitter behaviors on political participation.
As candidate and constituent adoption continue, the political importance of Twitter will
likely increase. While this may change the dynamics of the universe of political tweeters,
it will also make a useful campaign tool even more essential to modern American
candidacies. As Karl Rove, in a recent editorial in the Wall Street Journal, suggested,
“campaigns are only starting to understand how to use the Web and social-networking
tools to make video and other data go viral—moving not just to those on a campaign's
email list but to the broader public…. It will have huge implications on how campaigns
are run, who we elect, and what kind of country we become” (Rove, 2011).
25
Works Cited
Ajzen, I. (1988). Attitudes, personality, and behavior. Chicago, IL: The Dorsey Press.
Ajzen, I. (1991). The theory of planned behavior. Organizational behavior and human
Ajzen, I. (2002). Perceived behavioral control, self-efficacy, locus of control, and the
http://www.electionstudies.org/studypages/download/datacenter_all.htm.
Asch, S.E. (1956). Studies of independence and conformity: I. A minority of one against
Bode, Leticia, Stephanie Edgerly, Emily Vraga, Ben Sayre, and Dhavan Shah. (2011).
Digital Democracy: How the Internet Has Changed Politics. In Media Effects, Scharer,
Erica (ed).
Carr, David. (2008). How Obama Tapped Into Social Networks’ Power. The New York
Case, C., & King, D. (2011). Twitter Usage in the Fortune 50: A Marketing Opportunity?
Cha, M.; Haddadi, H.; Benevenuto, F.; and Gummadi, K. P. (2010). Measuring user
influence in twitter: The million follower fallacy. Proceedings of the Fourth International
Chaffee, S. H., & Kanihan, S. F. (1997). Learning about politics from the mass media.
26
Chi, Feng, and Nathan Yang. (2010a). Twitter in Congress: Outreach vs. Transparency.
http://ssrn.com/abstract=1630943.
Chi, Feng, and Nathan Yang. (2010b). Twitter Adoption in Congress. Social Science
factbook/geos/us.html.
Downloaded by [North Dakota State University] at 23:54 28 February 2015
Cialdini, R.B. (2001). Influence: Science and practice (4th ed.). New York: Allyn &
Bacon.
Coleman, J.S., Katz, E. & Menzel, H. (1966). Medical Innovation: A Diffusion Study.
Crutchfield, R.S. (1955). Conformity and character. American Psychologist, 10(5), 191-
198.
Dalrymple, K.E. (2006, November). Political knowledge and participation gaps in 2004:
Does Internet use bridge the gap between high and low education groups? Paper
presented at the annual convention of the Midwest Association for Public Opinion
27
Garrett, R. K. (2009). Echo chambers online? Politically motivated selective exposure
age of the internet: Is online political engagement really different to offline? The Journal
Gil de Zuniga, H., Veenstra, A. S., Vraga, E. K., and Shah, D. V. (2010). Digital
Gil de Zuniga, H., Puig-i-Abril, E., and Rojas. (2009). Weblogs, traditional sources
online, and political participation: an assessment of how the internet is changing the
Golbeck, J., Grimes, J.M., & Rogers, A. (2010). Twitter use by the U.S. Congress.
Journal of the American Society for Information Science & Technology, 61(8), 1612-
1621.
Gulati, Girish J., and Christine B. Williams. (2010). Communicating with Constituents in
140 Characters or Less: Twitter and the Diffusion of Technology Innovation in the
United States Congress. Social Science Research Network Working Paper Series.
Hargittai, E., & Walejko, G. (2008). The Participation Divide: Content creation and
sharing in the digital age. Information, Communication, & Society, 11(2), 239-256.
http://www.hubspot.com/blog/bid/5503/HubSpot-Releases-Third-State-of-the-
Twittersphere-Report-SOTwitter.
28
Huckfeldt, R., Mendez, J.M., and Osborn, T. (2004). Disagreement, ambivalence, and
Hwang, H., Thorson, K., Borah, P., Cleland, R., and Perlmutter, D. D. (2007). The
blogosphere and participatory democracy: The role of hostile media perception in blog
users’ news source selection and expressive participation. Presented at AEJMC 2007 in
Washington, DC.
Downloaded by [North Dakota State University] at 23:54 28 February 2015
Katz, E. & Lazarsfeld, P.F. (1955). Personal Influence. Glencoe, IL: Free Press.
Keller, E.B. & Berry, J.L (2003). The influentials: One American in ten tells the other
nine how to vote, where to eat, and what to buy. New York: Simon & Schuster.
Kimbrough, A.M., Guadagno, R.E., Muscanell, N.L., and Dill, J. (2013). Gender
Krueger, B.S. (2002). Assessing the potential of Internet political participation in the
Lakin, J.L., & Chartrand, T.L. (2003). Using nonconscious behavioral mimicry to create
Larsson, A.O. (2013). Rejected bits of program code: Why notions of ‘politics 2.0’
remain (mostly) unfulfilled. Journal of Information Technology & Politics, 10: 72-85.
Larsson, A.O., & Moe, H. (2011). Studying political microblogging: Twitter users in the
2010 Swedish election campaign. New Media & Society, 14(5), 729-747.
Lassen, David. S. and Adam R. Brown. (2011). Twitter: The Electoral Connection?
29
Lerman, K., and Ghosh, R. (2010). Information contagion: An empirical study of the
spread of news on digg and twitter social networks. Proceedings of the Fourth
Mackie, D.M. (1987). Systematic and nonsystematic processing of majority and minority
Marwick, A., & boyd, d. (2011). To See and Be Seen: Celebrity Practice on Twitter.
17(2), 139-158.
opinion, and the prospects for deliberative democracy. In J. Ferejhon & J. Kuklinski
(Eds.), Information and democratic processes (pp. 59-99). Urbana: University of Illinois
Press.
Mossberger, K., Tolbert, C.J., and McNeal, R.S. (2008). Digital citizenship: The Internet,
Mutz, D. C. (2006). Hearing the other side: Deliberative versus participatory democracy.
Nie, N.H., Powell, G.B. Jr., & Prewitt, K. (1969). Social structure and political
Nielsen, R.K. (2011). Mundane internet tools, mobilizing practices, and the coproduction
30
O’Connor, B.; Balasubramanyan, R.; Routledge, B. R.; and Smith, N. A. (2010). From
tweets to polls: Linking text sentiment to public opinion time series. Proceedings of the
Pew Research Center for the People & the Press. (2008). Social networking and online
videos take off. Report from the Pew Research Center for the People & the Press.
Pew Internet and American Life Project. (2011). 8% of online Americans use Twitter.
2010.aspx.
Pew Internet and American Life Project. (2012). Twitter Use 2012. May 31, 2012.
Available at
http://networked.pewinternet.com/~/media/Files/Reports/2012/PIP_Twitter_Use_2012.pd
f.
Rogers, E.M. (1976). New Product Adoption and Diffusion. Journal of Consumer
Rove, Karl. (2011). Political campaigns go viral: The impact of the Internet on elections
has only begun to be felt. The Wall Street Journal. March 10, 2011. Available at
http://online.wsj.com/article/SB10001424052748704132204576190472697049288.html?
mod=WSJ_Opinion_LEFTTopOpinion
Scheufele, D.A., Misbet, M.C., Brossard, D., Nisbet, E.C. (2004). Social structure and
338.
31
Strabac, Z., & Aalberg, T. (2011). Measuring political knowledge in telephone and web
Trippi, Joe. (2004). The revolution will not be televised: Democracy, the Internet, and the
Tolbert, C.J., and McNeal, R.S., (2003). Unraveling the effects of the Internet on political
Tumasjan, A.; Sprenger, T. O.; Sandner, P. G.; and Welpe, I. M. (2010). Predicting
Downloaded by [North Dakota State University] at 23:54 28 February 2015
elections with twitter: What 140 characters reveal about political sentiment. Proceedings
Twitter. (2013). Connecting adverstisers to Twitter users around the world. Twitter Blog,
around-world
Williams, C.B., & Gulati, G.J. (2012). Social networks in political campaigns: Facebook
and the congressional elections of 2006 and 2008. New Media & Society, 15(1), 52-71.
Yang, Jiang, and Scott Counts. (2010a). Comparing Information Diffusion Structure in
Yang, Jiang, and Scott Counts. (2010b). Predicting the Speed, Scale, and Range of
32
Table 1. Descriptive Statistics for Sampled Political Twitter Users
(%female) (%
Education * * 70.5% * 24.4% b
Downloaded by [North Dakota State University] at 23:54 28 February 2015
college degree)
Income <$20,000 >$200,000 $62,000 $20,780 $49,777 b
Interest
Voting 2010 * * 94.2% * 34.2% c
33
Table 2. Ordinary Least Squares Regression Models
N=848 N=600
Reading
Disagreement
Political - - .241
Knowledge
Twitter - - .105*
34
Influence
Twitter - - -.100*
Information
Motivations
Motivations
Twitter - - -.028
Downloaded by [North Dakota State University] at 23:54 28 February 2015
Entertainment
Motivations
Followers - - .001
35
Table 3. Candidates sorted by number of followers in our sample and number of retweets
of followers
NC/Burr_Richard 40 39079
DE/Coons_Christopher A 49 33056
IA/Conlin_Roxanne 32 31406
LA/Melancon_Charlie 19 29857
Downloaded by [North Dakota State University] at 23:54 28 February 2015
WI/Feingold_Russell 70 28977
DE/O'Donnell_Christine 86 28754
NY/Gillibrand_Kirsten 57 28037
AK/McAdams_Scott 28 27767
FL/Crist_Charlie 35 27692
WA/Murray_Patty 26 27480
NV/Angle_Sharron 92 26685
CA/Boxer_Barbara 74 22949
KY/Conway_Jack 32 21148
NV/Reid_Harry 46 20762
MO/Carnahan_Robin 26 20512
LA/Vitter_David 45 19862
OH/Fisher_Lee 28 19524
PA/Sestak_Joseph A Jr 66 19031
36
NH/Hodes_Paul 21 18791
MO/Blunt_Roy 56 17880
OK/Coburn_Tom 93 17871
AK/Miller_Joe 68 16705
CT/Blumenthal_Richard 17 15492
Downloaded by [North Dakota State University] at 23:54 28 February 2015
IL/Giannoulias_Alexander 22 15361
CA/Fiorina_Carly 68 15235
NC/Marshall_Elaine 27 15075
IN/Ellsworth_Brad 12 14756
WV/Manchin_Joe 8 14143
CT/McMahon_Linda 28 12452
CO/Buck_Ken 22 12104
NH/Ayotte_Kelly 29 11796
IN/Coats_Dan 18 11039
WA/Rossi_Dino 32 9128
WI/Johnson_Ron 30 7927
IL/Kirk_Mark 50 6511
AR/Lincoln_Blanche 16 4740
KY/Paul_Rand 50 3172
37
OH/Portman_Rob 43 1535
AZ/McCain_John 26 1509
WV/Raese_John 16 1497
UT/Lee_Mike 16 1046
NY/Dioguardi_Joseph 15 991
AR/Boozman_John 13 972
NY/Schumer_Charles 15 870
Downloaded by [North Dakota State University] at 23:54 28 February 2015
HI/Inouye_Dan 14 831
OR/Huffman_Jim 8 368
MD/Mikulski_Barbara 9 310
KA/Moran_Jerry 9 241
AK/Murkowski_Lisa 8 178
MD/Wargotz_Eric 1 62
IA/Grassley_Chuck 6 52
ID/Crapo_Mike 1 36
GA/Isakson_John 7 34
ND/Hoeven_John 6 20
VT/Britton_Leonard 1 15
GA/Thurmond_Mike 1 0
KA/Johnston_Lisa 2 0
38
i
The growth of political blogs in the last decade has been enormous, as noted by Bode, Edgerly, Vraga,
Sayre, and Shah (2011): “from a relatively unknown medium in 2004, when Howard Dean’s blog ‘changed
politics’ to their heavy use in 2008, when over a quarter of American voters used blogs for political
information”.
ii
This list was generated primarily through publicly available lists of major party candidates, from the New
York Times and Roll Call.
iii
Active Twitter accounts, for our purposes, simply have to have tweeted at some point during the
campaign season.
iv
The questions were as follows: Which political party has control of the Senate after the 2010 elections?
Which political party has control of the House of Representatives? Which political party largely favors
repealing the recent health care bill? Do you happen to know…how many times can one be elected US
president? For how many years is a US senator elected? How many US senators are there from each state?
The percentage of votes needed to override a veto in the House of Representatives?
v
Respondents were asked, “How many days in the past week have you…” read a print copy of a local daily newspaper, read
Downloaded by [North Dakota State University] at 23:54 28 February 2015
a print copy of a national daily newspaper, read a news magazine (e.g., Time, Newsweek), visited a news Web site (e.g.,
CNN.com; NYTimes.com), visited a political blog (e.g., DailyKos.com; Instapundit.com), watched national evening news,
watched local evening news, watched cable news (CNN, MSNBC, FOX News), listened to news/talk radio (e.g., Rush
Limbaugh, Air America, watched a comedic news show, received political information from your Facebook network, visited
a political campaign website.
vi
The focus of this study was on political aspects of Twitter use. For reference, we asked respondents about
a number of motivations for their Twitter use, which fell into three rough categories – political motivations,
informational motivations, and entertainment motivations (nine questions in total). Means for political
motivations were generally the lowest (3 questions, mean=4.3), suggesting that although our sample is
highly engaged in politics, they still use Twitter primarily for other reasons such as finding information (5
questions, mean=4.8) or entertainment (1 question, mean=5.1).
vii
Sent an email to an editor of a newspaper/magazine, used email to contact a politician, contributed money
online to a political campaign that I follow on Twitter, contributed money online to a political campaign
that I do not follow on Twitter, forwarded a political email to friends, organized political activity via the
Internet, recruited friends via email to support a cause or campaign, subscribed to a political email list, met
in person with a political group I joined online.
viii
Did volunteer work, attended a political meeting, rally, or speech, displayed a campaign button, sticker,
or sign, worked for a political party or candidate, and contacted a politician.
ix
We combine measures of political participation as is standard in the field. See Nie, Powell, and Prewitt,
1969; Krueger 2002; Mossberger, Tolbert, and McNeal, 2008 for examples.
x
Although some might be concerned that respondents to an online survey might “cheat” on knowledge
measures such as these, evidence suggests they do not actually do so (Strabac & Aalberg 2011).
xi
It should be noted that we use different measures than ANES for each of these variables, so comparisons
are not precise. In particular, our knowledge measure is composed of more items, and our items tend to be
procedural (what percentage of legislatures to override a veto) rather than ANES’s questions which deal
more with current events (what is the current unemployment rate). Additionally, as a large number of
respondents began but did not complete the survey, we were forced to eliminate missing data, so our
measure likely inflates the knowledge of our sample slightly.
xii
While this may be surprising on its face, we feel it is related to our sampling strategy. In order for
respondents to be sampled, they must be following a candidate on Twitter. Implicitly, this means they must
be (a) on Twitter, and (b) relatively interested in politics. Both of those factors tend to skew to the older
population. Thus for an individual to to end up in our sample and NOT be older, the threshold of political
activity is likely higher. That is, we believe the finding that younger users are more likely to participate
offline is simply due to the fact that it’s less likely younger users end up in our sample to begin with.
xiii
We mean this statement in terms of how we have modeled the relationship. Given that our data are
cross-sectional, we cannot say definitively whether those who participate offline in politics are moved to
participate on Twitter, or whether participating on Twitter encourages participation in offline political
activities.
39
xiv
We also included a measure of perceptions of the influence of the tweets of others, including politicians.
The two measures were extremely collinear (corr=.97) so the additional measure was not included in
models.
xv
We collected data on followers the week before the election. However, such numbers are unlikely to
change much between data collection and survey data collection, just after the election.
xvi
Again, the causal direction here is unclear. The relationship is correlational (based only on cross-
sectional data), and thus it may also be true that those who tweet the most garner more followers.
xvii
Not surprisingly, this distribution roughly follows a power law and is thus quite skewed, as indicated by
the radical difference between mean and median. For perspective, almost 4/5 of our sample (657
respondents) had fewer than 1,000 followers, while only 20 users had 10,000 followers or more (about 2%
of our sample).
xviii
We restrict this analysis to candidates for Senate in 2010, mainly for ease of display.
Downloaded by [North Dakota State University] at 23:54 28 February 2015
40