SlideShare a Scribd company logo
1 of 61
Download to read offline
THE	
  GEORGE	
  WASHINGTON	
  UNIVERSITY	
  
SCHOOL	
  OF	
  MEDIA	
  AND	
  PUBLIC	
  AFFAIRS	
  
	
  
Twitter	
  and	
  the	
  2010	
  
Nevada	
  Senate	
  Race	
  
How	
  Senate	
  Majority	
  Leader	
  Harry	
  Reid	
  and	
  
Republican	
  Candidate	
  Sharron	
  Angle	
  incorporated	
  
Twitter	
  into	
  their	
  campaigns	
  
UNDERGRADUATE	
  HONORS	
  THESIS	
  
Rachel	
  Kottler	
  
4/28/2011	
  
	
  
	
  
	
  
Currently, little data exists on how Twitter, the newest social media phenomenon, is being used
in political campaigns. The following study explores the question of what role Twitter played in
the 2010 Nevada Senate race. This study seeks to understand how candidates are currently using
Twitter and determine how Twitter can be most effectively used to reach potential voters. It
compares campaign discussion on Twitter to discussion by in the mainstream media.
Specifically, it looks at all of the candidates’ tweets as well as all articles on the race in the Las
Vegas Review-Journal from October 1 through Election Day. It also looks at previous online
innovations to fully grasp how online mediums, including Twitter, can be used by political
campaigns. 	
  
  2	
  
Rachel Kottler
Thesis
4/26/11
Twitter and the 2010 Nevada Senate Race
I. Introduction
Just five years ago, the newest social media Web site, Twitter, was created. In 2006, Jack
Dosey, Evan Williams, and Biz Stone invented Twitter as a means for people to update each
other from anywhere and stay informed about news in their city (Sarno). Now, Twitter is used by
millions of people around the globe as a means to communicate with others, share information,
get news, or follow their favorite celebrities. All messages, more commonly referred to as tweets,
must be 140 characters or less and they can be directed at people with use of @. Conversations
are sorted with voluntary use of hashtags (#) that help track topical discussion. Users can
promote other people’s tweets by using a function called retweets.
As the number of people on Twitter grows, so does the site’s ability to serve as a political
platform. It hit the political scene in April 2007 when President Obama activated a Twitter
account for in his Democratic Presidential nomination campaign (Allen Hendricks 40). Since
Obama’s run for Presidency, Twitter has gained popularity among eligible American voters.
According to an August 2010 Pew Internet and American Life Project study, 17 percent of
voting age adults use Twitter in some regard (Madden 8). This suggests rapid growth as this rate
was only 11 percent in 2009 (Madden 8). Many Twitter users used the site to find out
information about the 2010 midterm elections (Smith 10). According to Pew, “28% of Twitter
users used the site to engage with the 2010 election in one way or another,” (Smith 10). When
asked about all social media sites combined, Pew found that over one in five adult Internet users
used these sites for political purposes during the 2010 election cycle (Smith 1).
  3	
  
Although Twitter usage by voting-age adults and political campaigns has increased, few
studies have been conducted about the function Twitter has in politics. Thus a gap exists in the
literature that would help us understand social media’s role in politics. This study will contribute
to the understanding of the role Twitter plays through an analysis of the role Twitter played in
the 2010 Nevada Senate race. Since a sizeable amount of research is available on how YouTube,
Facebook, Web sites, and blogs have been used in political campaigns, I set this study into the
broader context, by reviewing these Internet technologies that have influenced campaigns in the
last decade. In the findings section of this paper, I will discuss whether Twitter is used differently
than other technological advancements or used in a similar fashion.
The Senate race in Nevada was selected for a variety for reasons. First, it is one of the 32
states in the country that entertain early voting (Absentee and Early Voting). Second, the
drastically different candidates well represent the mood of the 2010 Midterm elections. Senate
Majority Leader Harry Reid, who was first elected to Senate in 1986, was up for reelection in
2010 against the Tea Party-endorsed candidate Sharron Angle, a candidate with no national
political experience. This Senate race was described as “highly contested” by both polls and the
news media prior to Election Day (Nevada Senate—Angle vs. Reid).
This race was unique because of Senator Reid’s position in the Senate, and also because
Sharron Angle ran on more radical and conservative positions than most Republican Senators in
office hold (Fournier). Due to these circumstances, the campaign was widely discussed in the
media. Its notoriety through the mainstream media makes it easier to compare and analyze the
way the candidates used social media to discuss the campaign to how the mainstream media
discussed the campaign.
  4	
  
This study will look into a variety of questions in order to propose an answer to the
research question: What was the role of Twitter in the highly contested 2010 Nevada Senate
race? Questions include whether Twitter was used to jumpstart the discussion of campaign
issues, or continue existing discussions? What type of information are campaigns tweeting
about? Are they using Twitter as a means to attack their opponent or to relay positive
information about themselves? Are the attacks direct or indirect? Are they using Twitter more
frequently to inform people about campaign events, share their positions, reinforce campaign
messages, or to connect individual voters with the campaign through various campaign asks (i.e.
donations, gotv, signing up for events)?
To give a background for the study and the questions it seeks to answer, theories about
whom the Internet can reach and how the news impacts voters must be discussed. Once these
theories are understood, the history of changing technologies and how they follow or deviate
from these theories must be studied. This will clarify how previous new innovations online have
affected campaigns and also help formulate hypotheses about Twitter. Noteworthy technologies
include candidate Web sites, email, blogs, and other social media sites. The current knowledge
and ideas about Twitter usages will also be discussed. Finally, hypotheses will be laid out and the
findings of the content analysis will be presented.
II. Theories
i. Theories about the role of the Internet
To fully grasp the impact of online technologies, it is important to discuss the different
approaches and theories that have been developed on the Internet’s role in campaigns.
Understanding these theories will help determine if Twitter use follows or deters from these
theories. Most importantly, analyzing these theories and history helped formulate the hypotheses
  5	
  
about the role of Twitter in campaigns that this study seeks to test. Two different approaches
taken, the instrumental approach and psychological approach, the must be explained in order to
explain the two major theories, the reinforcement theory and the mobilization theory, that have
been created. These theories attempt to answer the question of whom campaigns should target
their online engagement efforts towards based on the question of how Internet works to affect
voters. Each theory lies in one of the two broader points of view political scientists have about
the Internet and its possible affects on political engagement.
The instrumental approach explains that as the cost of communication deceases, people
are more apt to participate. (Xenos and Moy 706) According to Bimber ‘‘changes in the cost and
variety of sources of information directly affect levels of political participation’’ (Bimber. 2000).
In other words, as new, inexpensive technologies emerge, people will be more likely to
participate in the political process. If this is correct, since Twitter is free, it will in fact increase
citizen engagement in politics.
Multiple studies have been conducted that support the instrumental approach. A 2003
study by Johnson and Kaye found a relationship between Internet use and political engagement.
(Johnson and Kayne 9-34) Another study in 2003 done by Tolbert and McNeal found that there
was a positive correlation between access to campaign news online and political participation.
(Tolbert and McNeal 175-185) A 2006 study by Drew and Weaver found that there was a
positive relationship between exposure and attention to campaigns and interest level. (Drew and
Weaver 25-42). However, these studies fail to address the issue that those who already use the
Internet may be more likely to participate in the political realm online.
The psychological approach focuses more on motives and attributes of Internet users
(Xenos and Moy 706). This approach is based on the media niche theory that concludes, “the
  6	
  
Internet offers the broadest array of opportunities for gratifying various needs,” (Xenos and Moy
707). Specifically, this theory adheres to the belief that the differences in “individual’s level of
sophistication, characteristics, and social context” impact online engagement (Xenos and Moy
707). In other words, people are more likely to use the Internet to engage in the political process
if they are educated, of higher income, and concerned with current affairs and politics.
A 2005 study by Prior supports this theory’s claim that motive matters. It found that those
who used media for political information were more likely to be engaged than those who used
the media for entertainment purposes (Xenos and Moy 708). If this were correct, political
discussions on Twitter would only be from those highly informed and highly interested in the
campaign, thus agreeing with the reinforcement theory. Therefore, in terms of Twitter, those
retweeting would fall into this category.
The reinforcement theory believes that more politically active people will engage in
politics through online participation. The essential component to the reinforcement theory is that
“those with established interests retain the strongest incentive for online participation,” (Klotz
111). Active users would include journalists, lobbyists, party members, and grassroots activists
(Norris 218). Pippa Norris applied this overarching theory about campaigning to Internet
technologies in 2001 as a contrast to the mobilization theory. She used Pew data on Internet
users from 1996 and 1998 to support this theory (Norris 218). Findings by a study done by Bruce
Bimber and Richard Davis also support this theory. They found that in 2000, the majority of
Web site viewers were already supporters of the candidate (Bimber and Davis 108). As Bimber
states in his book Information and American Democracy,
“On the whole, those who pay the most attention to the media of previous information
revolutions are also paying the most attention to new media, and those most likely to be
active in democratic processes in earlier information regimes are those engaged with the
new organizational structures of the emergent information regime” (Bimber 230).
  7	
  
Bimber and other proponents of this theory do not think that the Internet can engage people who
are not already interested in the political process. Instead, it will engage people who were active
already.
This theory is related to earlier works on political communication effects. Specifically, it
applies Converse’s theory of non-attitudes. According to Converse, the majority of Americans do
not have a true political ideology. The few that do have a true ideology seek out information that
aligns with their partisan preference and reject information that contradicts it (Converse 228-
256). If this theory is correct, it would support a hypothesis that the majority of citizens engaged
in campaigns via social media, including Twitter, are already strong supporters of the candidate.
This theory could also explain why the number of people following a candidate on Twitter is
relatively low since few people hold such strong partisan preferences and are active in politics.
By contrast, the mobilization theory argues the Internet is a means to gain supporters.
Norris defines this theory as:
“The Internet may serve to inform, organize and engage those who are currently
marginalized from the existing political system—such as younger generation, people
living in isolated peripheral communities, or fridge political minorities disaffected by the
traditional system—so these groups will gradually become drawn into public life and
civic communities,” (Norris 218).
It suggests that the Internet can create incentives to garner new participants into campaigns and
politics since the Internet makes it easier to learn candidate positions, reach government officials,
and network about issues. This is because it makes it easier for people to have access to political
information who may have previously been isolated and easier for those who did not want to
spend time and money learning about politics. Thus, the Internet lowers the costs of political
engagement. It would focus more on gaining support from those who do not have true ideologies
and are still malleable.
  8	
  
The mobilization theory is supported from the 2008 Presidential election when Obama
used the Internet to connect with young voters and minorities (Kayne). Since Pew reported that
as of May 2010 over a quarter of 18-29 year olds were using Twitter, it may act as a means to
engage the younger generation as other social networking sites did in 2008 (Madden 8).
Evidence in support of this theory through Twitter use could come from retweets and hashtags. If
partisans are retweeting they can engage non-partisans by retweeting links and campaign
information that their non-partisan followers may not have sought out on their own. People, who
feel a stake in the election for specific reasons, like illegal immigrants whose families fates in
America would be affected by the winner of this election, may come across the campaign
information because of hashtags or retweets and engage themselves. They are more likely to
come across these hashtags if they are issue-centered, like #DREAMAct. In either of these
situations, the mobilization theory is supported.
Both of these theories stem back to the Internet Engagement Model, which seeks to
explain how the Internet can be used to engage voters. Each theory uses this model but makes
different conclusions about who is engaging with campaigns through the Internet. This model is
used to study Internet activity on three levels of analysis: the national context, political
institutions, and the individual level. In terms of political campaigns in the US, the national
context remains consistent. The part of this model that relates to this topic are the other two
levels. Norris explains,
“the core institutions of the political system available in the digital world provide the
systematic context within which individual citizens have opportunities to participate
online. Which particular citizens take advantage of these opportunities is determined by
their personal resources and their motivation,” (Norris 15).
This model suggests that citizens can become more involved with politics through using the
Internet to connect with the system. Their relationships with political groups, parties, and the
  9	
  
news media can strengthen, thus allowing individuals to contribute more to governmental
decisions.
With Twitter, people have a new resource that allows them to communicate directly with
those running for office and an easy avenue to become engaged with the campaign. Candidates
and news sources can see if people like their messages based on what is retweeted or by looking
at hashtags. It also presents itself as an intensely personal medium. People can voice their
support for candidates and causes to their immediate circle of followers and with the general
twitersphere via hashtag use. Non-politically active voters can stay updated on campaigns simply
by following someone on Twitter who chooses to be engaged.
Related to this model is the concept of voter efficacy. Political efficacy is defined as “a
combination of one’s sense of competence in the political sphere and one’s assessment of the
responsiveness of the system,” (Valentino 308). Therefore, if citizens think their actions can
influence the political system, they are more likely to participate in politics. This is based on both
internal efficacy, which is determined by voter confidence in him/herself to understand the
system, and external efficacy, which is comes from the voter’s confidence that the political
system is open to everyone and he/she will be heard. Twitter has the potential to enhance
external efficacy because it forms an easy route of direct communication with politicians. This
way, candidates (or staff members acting on their behalf) can respond to voters and show their
constituencies that their opinions matter.
ii. Theories about News Coverage
As history has shown, Internet communication can affect the campaign news cycle.
Therefore, it is important to study whether the same is true for candidate tweets on Twitter.
Theories of news coverage must be understood in order to address the question about whether
  10	
  
Twitter starts the discussion of campaign information or lengthens it. Specifically, the role of
news coverage plays in political campaigns and how news affects people must be defined. This
will give important background information to help understand why politicians may want to use
Twitter to jumpstart or participate in conversations about the campaign that are occurring in the
mainstream media.
Priming is related to controlling the salience of issues and how important issues are
judged. According to Benoit, priming “concerns a voter’s expectations, which can influence
whether they attend to this information and how what he or she learns is perceived, construed, or
interpreted,” (Benoit 207). In terms of elections, priming will determine on what citizens judge
politicians. This judgment is based on their stances on the more salient issues. For example, since
in the 2010 midterm elections healthcare was important, priming would cause people to judge
Senator Reid based on whether or not they agree with the healthcare measures he fought to enact.
Agenda setting is extremely important to news formation and has been influenced by the
Internet. Agenda setting is when what gets covered aligns with what citizens consider important.
It says that the media influences what issues are salient, or top of mind, for most people. The
term agenda setting came from McCombs and Shaw’s 1972 study where they found a positive
correlation between perceived importance of a topic and number of mentions of that issue topic
in the media (McCombs and Shaw 176-187). In terms of elections, agenda setting influences
what issues people think are important for the campaign and this primes what issues a candidate
is evaluated in terms of. For example, in 2010 issues included healthcare reform, the stimulus
plan, and the economy.
Understanding priming is important to this study because Twitter is a place where
candidates can release information about issues. If an issue that the media chooses to broadcast
  11	
  
heavily is negative for a candidate, they may use Twitter to try and divert the way citizens view
them on that issue. They may also use Twitter to try and divert attention from one issue to
another or to further enhance issue discussion as a way to keep the issue at the forefront of
discussion. If the issue is negative for their opponent, a candidate may use Twitter to try and
further control how people judge their opponent on that issue through the use of linking articles
and releasing news. For example, during the Nevada Senate race, getting the Hispanic vote and
immigration policies were actively discussed through the mainstream media. Sharron Angle had
a few major campaign gaffes in reference to Hispanic voters, which Harry Reid could tweet
about to try and prime immigration as the issue on which Angle was judged by voters as well as
frame the issue.
Candidates may attempt to use Twitter and other social media sites to influence agenda
setting. Candidates can post about whatever they want via social media in an effort to control
campaign discussions. They can then hope that the topics they discuss on social media are picked
up by the mainstream media. Understanding agenda setting will help understand why campaigns
might want to use Twitter to try and set the agenda of the campaign. Also, knowing whether
Twitter can jumpstart discussion could help to show if there is a way for them to successfully set
the agenda via social media.
Lastly, Zaller’s theory of elite discourse in opinion formation is also relevant for trying to
figure out the role Twitter is playing in elections. According to this theory, in large societies such
as the United States, people depend on “unseen others” to get information that leads them to
opinion formation. These “unseen others”, include political elites such as politicians, government
officials, policy experts, activists. They also include family and friends (Zaller 288-289). This
theory is relevant because if people are getting information from Twitter, it most likely comes
  12	
  
from one of these groups of people. Because of the retweet function, elites, such as politicians,
can share their messages to a larger audience than simply their followers. Therefore, even if
some people only follow their family and friends, they can indirectly receive campaign messages
and help spread them.
III. The Rise of the Internet
The rise of the Internet has played an important role in political campaigning. For the last
decade, new Internet technologies have allowed its role to continue expanding and changing.
Similar to other forms of political engagement, those who use the Internet for politics are self-
selected. However, this self-selected group is quite large. Pew reported that 2008 “marks the first
time the Pew Internet & American Life Project has found that more than half the voting-age
population used the Internet to connect to the political process during an election cycle,” (Smith
2). During the 2010 election cycle, it found that this number stayed at about the same percent and
represented 73 percent of adult Internet users. (Smith the Internet and Campaign 2010). Since
use of the Internet to gather campaign information has increased so rapidly, it is necessary to
study how Internet use has changed during the last five election cycles. This will create
background data for the 2010 election cycle, allowing the use of Twitter to be effectively
analyzed.
i. The Year 2000—“The Year of the Internet”
Although the World Wide Web was invented in 1989 and went public in 1996, the 2000
election cycle marks the first time in which candidates effectively used the Internet to benefit
their campaigns (Lange 108). In their book Campaigning Online, Bruce Bimber and Richard
Davis point out that the election of 2000 was not merely unique because of the way it was
decided, but also because the 2000 elections represented a “leap forward by candidates in the
  13	
  
degree of effort, money, and innovation dedicated to the Internet,” (Bimber and Davis 1).
Specifically, for the first time in 2000 use of candidate Web sites and email lists by Presidential
candidates became the norm. A study by S.J. Farnsworth and D. Owen found that about a third
of the voting-age adult population used the Internet to get campaign news (Farnsworth 420). A
Juniper Communications study found that that percentage almost doubled when limited to adult
online users (Farnsworth 421).
According to Bimber and Davis, in 2000, the most common uses of the Internet by
political campaigns were as a means of direct communication with voters, fundraising, voter
activism, and voter mobilization (Bimber and Davis 48). In fact, the year 2000 was the first
major election cycle in which candidates could fundraise online (Bimber and Davis 38). This
was imperative during both the primary season, and the election season. Senator John McCain
understood the potential advantages of the Internet and raised over $5.6 million online
throughout the Presidential primary season and garnered 18,000 volunteers in California and
Michigan alone (Farnsworth 416). These large numbers suggest that the Internet was able to
activate voter participation in elections.
A key component of Internet use in 2000 was the demographics of people who used it.
The Internet played a much larger role in the election for citizens who directly sought out
campaign information. More politically active citizens, those who frequently engage in politics,
were more likely to visit campaign Web sites and news sites online, thus offering early support
for the reinforcement theory (Bimber and Davis 108). In the case of Republicans running for
President, partisan use of the Internet saw substantial increases from 1996. “The average
percentage of supporters who signed up online spiked by 318%,” and the percentage of
Republicans who had been emailed by one of the candidates increased by over 1000 percentage
  14	
  
points a study by Chris Hull reported (Williams 61-62). These findings support the reinforcement
theory since those reached the most were already supporters.
Lastly, Farnsworth and Owens found that nearly 70 percent of respondents who used the
Internet to find out about campaigns did so to search for issue-based information (Farnsworth
and Owens 422). This is similar to Bimber and Davis’s finding that searching for issue-based
information was the second most frequent use for visiting campaign Web sites (Bimber and
Davis 115). Both studies found that education, strength of ideology, and age influenced the
amount people used the Internet to find out campaign information. This data suggests that the
Internet and campaign Web sites may be used by candidates to prime voters on how they are
judged on salient issues. It gives candidates a chance to weigh in on the discussion of the issues.
ii.	
   The	
  Year	
  2002—Blogging’s	
  Watchdog	
  Role	
  
In 2002, campaigns made no major strides in the use of the Internet for politics, but
citizens did. The use of blogs and the emergence of citizen journalists were important advances
that arose in the midterm election cycle. The 2002 election cycle marked the first time when a
politician suffered severe consequences after blogs picked up a controversial statement that the
mainstream media had ignored. Specifically, Senate Majority leader Trent Lott eventually had to
resign from his position after his comment at Senator Strom Thurmond’s birthday dinner
expressing his belief that if Senator Thurmond had won the Presidency in 1948, "we wouldn't of
had all these problems," (Glaser). Since Senator Thurmond ran on a segregation list platform,
this comment was interpreted to mean that Senator Lott supported segregation and was
racist.1
This incident demonstrated the influence blogs could have on the news. The mainstream
	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  
1	
  The bloggers accredited were John Marshall who posted a story the next day on talkingpointsmemo.com, along
with Andrew Sullivan and David Frum (Glaser). 	
  
  15	
  
media no longer had full control over what stories get covered and what stories are ignored, nor
control over what frame dominated coverage.
Although no major advances by campaigns occurred in 2002, it was the first midterm
elections where a majority of candidates had some sort of Web site. In fact, a study done by
Danielle Endres and Barbara Warnick found that 63 percent of House candidates had campaign
Web sites, compared with about 40 percent in the 1998 midterm election (Endres 322). Similar
to the 2000 Presidential election, in 2002, Web sites were primarily used to connect voters to the
candidate and the campaign rather than to each other. Specifically, a study by W. Lance Bennett
and Michael Xenos found that the most common items found on Web sites were candidate
biographies and a page for news clips and press releases (Xenos and Bennett 457). They also
found that campaign sites were posting contact information, included a place to submit
questions, a place to submit donations, a place to sign up to volunteer, and a link to be added to
the email lists (Xenos and Bennett 457). In short, candidates were using their Web sites in
attempt to impact what was considered newsworthy and activating supporters.
Even though Web sites were being used to communicate with voters, both studies found
basic errors in Web strategy. Bennett and Xenos concluded that although about a quarter of 18-
29 year olds received the bulk of their election information online, candidates failed to market to
their age group via their Web sites (Xenos and Bennett 460). This was based on their content
analysis of the information available on 1200 candidate Web sites during 2002. Endres and
Warnick find that “although political campaigns see the necessity of Web-based campaigning,
campaign Web sites have yet to realize their full potential as a medium to improve
communication between candidates and Web users and to influence voters,” (Endres 323). Thus,
  16	
  
candidates failed to incorporate the mobilization theory into their Web strategy and use the Web
as a way to engage young voters who are a large part of the online community.
ii. The Year 2004—Howard Dean’s Online Miracle
Monumental leaps for Internet usage in campaigns occurred in 2004. The new and
innovative ways it was used went well beyond creating an informational campaign Web site.
Howard Dean’s campaign manager, Joe Trippi, and his campaign staff are largely responsible for
the tremendous growth in Internet technology in the 2004 Presidential campaign. Not only was
Dean able to generate a vast number of followers and fundraise online, but also he was able to
create an online network that allowed supporters to meet up and created the first candidate blog
(Trippi 89).
Although Howard Dean lost the Democratic Presidential bid, his use of the Internet was
more advanced than any other primary candidate. When the campaign started, Dean not only had
little money, but little name recognition and support. The Dean campaign partnered with the
Web site meetup.com to create a place online where supporters could communicate with each
other and plan rallies or meetings. Due to this arrangement, a network for 190,000 Dean
supporters who could interact was created (Trippi 84). Unlike with prior studies on campaign
Web sites, this network’s success supports the viability of the mobilization theory because a-
typical voters became involved. In fact, according to Pew, “About four in ten (42%) Dean
supporters were participating in their first political campaign,” (Keeter, Funk, and Kennedy 1).
Dean’s Web site also included the first candidate blog as well as a “TV channel”
HowardDean.TV (Trippi 109).
The Dean campaign understood the power of creating a grassroots organization and the
way the Internet could accomplish this goal. It used the grassroots organization it created online
  17	
  
for campaign events, as well as for online fundraising. When Dean formally announced his
candidacy, the campaign was able to raise over two million dollars online in a week, the average
donation being about 100 dollars (Trippi 131). During this week, the Web site posted a
fundraising bat that supporters mimicked and posted elsewhere online in attempts to help the
campaign reach its fundraising goal. The Dean campaign surpassed their goal for the quarter, and
even the fundraising totals that quarter of Kerry and Edwards and they then got a lot of media
coverage for their success (Trippi 134).
The Dean campaign was the first to use these techniques, but other campaigns followed
in its footsteps. A study conducted by Steven M. Schneider and Kirsten A. Foot found that in
2004, all of the Presidential candidate Web sites observed “had established a practice of
involving” (i.e. using it for campaign asks) compared with only five of eight sites that were
studied from 2000. They also found that using Web sites for mobilization had increased
significantly (Williams 28-29). Both President George W. Bush and Democratic nominee John
Kerry also had blogs on their Web sites (Williams 133).
iii. The Year 2006—Gaffes on YouTube
The year 2006 marked another crucial step for the Internet and politics because it was the
first campaign cycle where YouTube and MySpace were available. The emergence of these
social media sites was both positive and negative for campaigns. In some cases, exemplified by
Representative George Allen, social media enabled voters to see campaign gaffes that may have
not been picked up by the mainstream press. In other cases, such as with Ned Lamont’s Senate
campaign in Connecticut, the free communication tool was an extremely important means for a
money-strapped campaign to communicate with voters (Cohen). These benefits and weaknesses
  18	
  
of social media made the Internet’s use in campaign strategies more vital to campaigns with little
funding and close, highly broadcasted races.
Although many candidates feared YouTube in 2006, campaign strategists saw its
benefits. For example, The Montana Democratic Party taped Senator Conrad Burns of Montana
closing his eyes during a hearing for the Montana farm bill (Jalonick). The video was quickly
launched on YouTube and even got earned media in newspapers such as USA Today and the
Wall Street Journal. The USA Today article from August 22, 2006 quoted Phil Singer, the
spokesman for the Democratic Senatorial Campaign Committee, saying that campaigns "have
always used video cameras, but the difference this cycle is that YouTube allows information to
get uploaded and out to the public a lot faster than it has in the past," (Jalonick). This example
shows that campaigns could use YouTube to influence agenda setting in the mainstream media.
The same strategy was famously employed against Senator Allen, who was caught
calling a young Indian man “Macaca” during a campaign rally. Although the man worked for his
opponent, Senator Jim Webb, who was following George Allen and videotaping him at every
campaign event, all that mattered was that Alan was caught on camera making a racial slur. After
this, these gotcha moments caught on tape became known as “Macaca moments,” (Karpf 144).
The year 2006 marked a further expansion of Internet use as the main source of campaign
information for Americans. In fact, Pew found that twice as many people reported that they used
the Internet as their main source of election information as compared with the 2002 midterm
election. Overall, Pew used the name “campaign Internet users” to represent over 60 million
people who received and/or shared information about the election online (Election 2006 Online
3). This meant that priming, framing, and agenda setting of the campaigns could be impacted
online.
  19	
  
Pertinent to this study, the year 2006 marked a drastic increase in the use of the Internet
to learn about congressional races. According to Pew, “the number of Americans getting
information about Senate races rose fivefold since 2002, and at least doubled in the case of
House races, local contests and gubernatorial campaigns,” (Election 2006 Online 20). This may
have been influenced by the increased availability of campaign information online and it was
becoming more of a common practice to seek campaign information online. With the emergence
of social media and the expansion of Internet use for midterm elections, 2006 is an important
year to compare with 2010.
iv. The Year 2008—“Yes We Can”
The Obama campaign in 2008 expanded Internet use drastically from the primary season
onward. According to a Pew study from April 2008, during the primary season 40 percent of
Americans adults were using the Internet to get campaign information. This indicates a 2.5 times
expansion in adult Internet use from the primary season in 2000. Pew also found that the use of
social media was gaining in popularity. At this time, 35 percent of adults had watched political
videos online and 10 percent had used either MySpace or Facebook to receive campaign
information by following candidates or being their “friends” (The Internet and the 2008 election
3).
By 2008, the use of YouTube had exploded. A study by Williams and Gulati found that
there was a significant increase in Senate campaigns with a YouTube channel between 2006 and
2008. The percent of campaigns with this jumped from about 10 percent of total campaigns to
72.7 percent of Republicans and 70.6 percent of Democrats (Gulati 99). These channels were
successfully used to connect campaigns to voters. Pew found that 35 percent of voting-age
Americans watched political videos online during the 2008 cycle (Rainie ii).
  20	
  
However, campaigns were no longer the only ones using YouTube to post videos. Rather,
citizens who felt a stake in the election created independent campaign videos. The two most
famous were “Obamagirl”, which featured an attractive female singing about her crush on
Obama and Will.i.am’s “Yes We Can” video, which featured a number of famous celebrities
singing along with an Obama speech. According to a study done by Dr. Emily Metzgar and
Albert Maruggi, “Obamagirl” was viewed 11.6 million times and “Yes We Can” was viewed 13
million times during the campaign (Metzgar and Maruggi 153). Spending money on television
airtime no longer was the only way to showcase campaign videos. These two videos exemplify
the viral effect of the Internet and highlight that this effect can be utilized by anyone interested in
the campaign. YouTube has created a way for citizens to become included in the campaign
process by allowing them to make and legally post their own videos for anyone to see. When
citizens feel as though they can impact the election, they become more engaged, thus a potential
means for increasing voter efficacy is created though social media outlets.
Another major advancement by Obama was improving Dean’s meetup.com platform
from 2004 through the creation of Obama’s own online network of followers. His online
grassroots network was called Obama for America, later known as Organizing for America
(OFA). OFA used a “bottom up” campaign approach to get potential supporters involved in the
campaign. Like meetup.com, it connected supporters to each other via the formation of local
groups, blogging, and campaign events (Organizing for America). OFA targeted minority groups
and young voters who are not typically active in elections. It raised an unprecedented amount of
money online, totaling a half a billion dollars with small donations (Vargas). Organizing for
America thus adhered to the institutional approach focused on involving supporters who may
have not had the financial means to become involved before the Internet age. It gave people who
  21	
  
did not have the financial means to donate large sums of money to the campaign other ways to
get involved, via small donations, different degrees of volunteering, and ways to expand
Obama’s reach online. Like via social media, it was a way for people to feel as though they could
partake in spreading Obama’s message.
2008 also marked the first year in which Twitter was used by campaigns. Specifically,
President Obama used Twitter as dimension of his social media strategy. According to
Communicator-in-chief, a book about how President Obama used new media technology to win
the election, Fradric I. Solop states that “Obama used Twitter in tandem with other social media
tools to fine-tune his message to a technology-savvy population,” (Allen Hendricks 40).
Although only 3.5 million people had Twitter accounts in November 2008, Obama managed to
obtain 118,000 followers. According to Solop’s study, Obama mainly used Twitter to announce
his location and to get people to go to www.barackobama.com. (Allen Hendricks 41). Since
Obama used Twitter as a means to drive people back to older campaign technologies, it will be
interesting to see if the Nevada candidates for Senate are also using Twitter mainly as a way to
communicate their location to voters and gain Web site traffic.
IV. What Experts are saying about Twitter and Campaigns
As mentioned previously, there is a very small collection of studies on Twitter use by
campaigns. Despite the lack of systematically tested data, experts in political communications
are actively discussing Twitter and how campaigns should use it. According to Steve Pearson
and Ford O’Connell of Campaigns and Elections, “a campaign's tweets need to reflect the
candidate's voice.” In other words, they believe tweets will be more effective if they seem as
though they are coming from the candidate personally. This is because Twitter’s main function is
to give individuals a place to share their thoughts.
  22	
  
Experts believe that social media should be incorporated into other campaign strategies.
In an interview with Mindy Finn, co-partner at the political consulting firm Engage DC, she said
that “the most effective use of Twitter for close Senate races relies on frequent use that mirrors
the most important aspects of the campaign, what I call the three "M"s: message, money and
mobilization,” (Finn). In other words, Twitter should be used as another platform to advance
campaign messages, increase fundraising totals, and mobilize potential supporters in order to be
most effective. Finn believes that Twitter is more often used to promote candidates in a positive
manner than attack opponents (Finn). This is because being positive helps raise money and
mobilize supporters. Thus, her thinking supports the reinforcement theory. Sticking to campaign
messages and using campaign asks is one way to follow this advice. This study will attempt to
look into whether campaigns are using Twitter for the “three M’s” by categorizing tweets based
on what they are trying to accomplish and by studying how they are being picked up and spread
via retweets.
V. Current Studies about Twitter
Since 2008, a few studies have been conducted to try and determine how candidates,
politicians, and the public are using Twitter. Their findings were used to formulate the
hypotheses that will be presented in this thesis and to create the proper research design. This
study will expand upon techniques used in these previous works and the coding definitions they
used to propose an answer to the question of what is the role of Twitter in Senate campaigns for
the candidates, the news, and the public.
One particularly helpful study was that of Wyth Ruthven, Twitter Trends in Off-Year
Elections: Virginia, New Jersey, and Massachusetts. This study analyzed the tweets for a three-
month period prior to the 2009 elections for Governor in Virginia and New Jersey and Senator in
  23	
  
Massachusetts. Ruthven categorized the tweets on the basis of being calls to action, self-
promotion, personal observation, conversation, or passing information. He found that over 50
percent of tweets from the sample of 3,521 tweets observed fell into the category of “self-
promotion”, which was defined as “off-message tweets, personal observations not related
to the campaign” (Ruthven 2).
In Massachusetts, Ruthven found that Senator Scott Brown’s surge in Twitter followers
correlated with the surge of donations, interest, and growing popularity in his campaign. He
concluded that Twitter can be used as an effective means for calls to action, including
fundraising and increasing GOTV efforts. (Ruthven 3-6). These findings suggest that Twitter
may be affectively used for self-promotion and helped me formulate the hypothesis that tweets
will more often be self-promotional than attacking opposition. Like Finn, Ruthven concluded
that candidates should use preserve authenticity in their tweets (Ruthven 23). These findings also
helped me decide to code for authenticity, use of external content, and calls to action.
Overall, Rythven concluded that having multiple Twitter accounts can be problematic for
campaigns. In Nevada during the 2010 Senate race, Senator Reid had two Twitter accounts, one
for his position as Majority Leader, and one for his reelection campaign. This research suggests
that having the multiple accounts may hurt Reid.
Other useful studies have recently been released that rank effectiveness of current elected
officials on Twitter. One example is a 2010 Burson-Marsteller study on how House and Senate
members are using Twitter to connect with their constituencies. Like Ruthven, they found that
more authentic Twitter accounts are more successful than accounts that did not tweet in a
personal manner (Burson-Marsteller 23).
  24	
  
Another example, the 2010 L2 Senate Digital IQ test by George Washington University
Dean Doug Guthrie and NYU professor Scott Galloway, hypothesized that “digital competence
provides an opportunity for senators to authentically engage and mobilize voters and
constituents,” (Galloway 4). They created a metric to measure digital competence of Senators
and ranked all Senators in office. This metric was based on presence number of likes, and growth
on Facebook; presence, followers, velocity of Tweets, and follower growth on Twitter; Presence,
number of upload,s and number of channel/upload views on YouTube; Velocity of mentions on
Blogs and other 2.0 sites and sentiment of blogs; and site traffic to candidate Web sites.
According to their rankings, Senator Reid was categorized as a genius and ranked tied in fourth
place with Al Franken and John Cornyn (Galloway 5). Reid’s categorization as one of the most
effective users of social media supports the choice of analyzing Reid’s Twitter use to gain insight
into how Senate candidates are using the site.
Both of these works concluded that Republican Senators were using the Internet more
effectively than Democrats. The Burson-Marsteller study concluded, “while there are
fewer Republicans in Congress, a larger percentage of them are using Twitter to communicate,”
(Burson-Marsteller 9). The L2 Senate Digital IQ study added that Senators up for reelection were
utilizing social media more effectively than those not up for reelection, and that November
challengers in close 2010 Senate races were often using social media more effectively than the
incumbents. One example was Carly Fiorina in California who had over 11 times as many
Twitter followers as Senator Barbara Boxer (Galloway 9). Despite these numbers, Boxer won the
election.
Although Facebook and others suggest that social media presence can be a reliable
indicator of who will win an election, the California and Nevada Senate races show that the
  25	
  
overall number of followers may not be the best indicator of who is successfully using social
media to increase their chances of winning (Coyle). On November 2, Sharron Angle had over
2,000 more Twitter followers than Harry Reid’s campaign account, yet she lost the election.
According to PBS.org, from September to October 21, Angle increased her Twitter following by
39.5%. Reid’s only increased by 13.8% (Leher). If gaining followers toward the end of an
election was the best way to use Twitter, than her increased momentum should have meant that
she would win. Thus, this reinforces the need for this study that will look more in depth into
Twitter usage and give insight into current strategies. It will seek to figure out how Twitter can
best be utilized by campaigns.
VI. Hypotheses
The main research question I propose is how did candidates in the 2010 Nevada Senate
campaign use Twitter? To figure out the answer to this question, it has been divided into three
areas of sub-questions. For each area, hypotheses have been formulated.
The first area is the interaction between Twitter and the mainstream media. My first
research question is was Twitter used to jumpstart the discussion of campaign issues, or continue
existing discussions from media coverage? Based on the literature review presented above, I
hypothesize that Twitter is being used as a means to continue existing discussions. I came to this
conclusion mainly because history shows that it is rare for stories to break on social media or
blogs and then get picked up by the mainstream media. Instead, usually the mainstream media
sets the agenda and people turn to the Internet to find out more information about the issues that
are salient to the campaign. Therefore, I think that stories will show up in both Twitter and the
mainstream media around the same time, but will get more coverage on Twitter. Twitter is not
able to control the agenda of the mainstream media. Specifically, I believe that when there is
  26	
  
negative news coverage of one candidate, the opposing candidate will use Twitter to lengthen the
discussion of the issue. I have come to these conclusions based on my understanding of priming,
agenda setting, and understanding history of the web. Since the web has been used previously for
candidates to share their points of view, I believe Twitter is the same.
The second set of questions is related to how Twitter is being used by the campaigns. Are
Senate candidates using Twitter as a means to attack their opponent or to relay positive
information about themselves? And if attacks are present, are they direct or indirect? Indirect
attacks will use links to sources and may reference news articles whereas direct attacks will
come from the campaign. Based on the 2009 study’s results, I hypothesize that candidates are
using Twitter mainly as an outlet for releasing positive information about themselves. However,
I believe that I will find that when attacks are made using Twitter, they are presented in an
indirect fashion, via the usage of links to external sites and news stories. This is based on my
understanding of the current state of the Internet and social media. Candidate Web sites,
Candidate Facebook pages, blogs and even online news stories often use links to the original
sources, so I expect Twitter use to do the same. Also, political attack ads often use someone
other than the candidate running for office to make the attack. It is not common practice in
campaigning to have the candidate directly attack their opponent via any outlet and the literature
review does not present reasons to think it will be different with Twitter.
Another aspect of this question refers to the types of information campaigns tweet about.
Are candidates using Twitter more frequently to inform people about campaign events, post
news, discuss issues, reinforce campaign messages, or for campaign asks (i.e. donations, GOTV,
signing up for events)? Does this change over the course of a campaign? I hypothesize that the
most common thing candidates are tweeting about are ways for voters to get involved. This
  27	
  
includes GOTV efforts, as well as donations, event information, and random campaign asks. I
came to this hypothesis based on what candidates have used other online outlets for in the past,
such as was done by Barack Obama and Howard Dean who fundraised and created online,
grassroots networks. I also hypothesize that the main type of information campaigns put out on
Twitter changes throughout the campaigns. I think at the beginning campaigns use Twitter to
discuss issues whereas close to Election Day they use Twitter for GOTV. This is because overall
campaign strategy shifts to a GOTV focus close to Election Day.
The third area of questions is related to the viral nature of Twitter. To examine this
concept, I will look at retweets and hashtags. Specific questions include whether inclusive
language makes people more likely to retweet the candidates? Does using consistent hashtags
lead to more retweets? I hypothesize that tweets with inclusive language are more likely to be
retweeted. This is because personalized messages are more appealing because they make people
feel as though they can influence the results and are a part of the political process. This relates to
the concept of external efficacy. I think that using consistent hashtags will also lead to more
retweets because it makes the tweets easier to find. I chose to look at retweets because number of
followers proved to not be sufficient to show who is reaching the broadest audience on Twitter
and activating potential supporters. I decided retweets was better because depending on who
retweets the candidates, their messages potentially can reach a much larger and more diversified
audience. Since the instrumental approach suggests that as cost of participation decreases, people
will become more apt to participate, and retweeting is free, I thought it was a good way to look at
audience engagement.
  28	
  
VII. Research Design
This study used a content analysis of all of the tweets posted by Sharron Angle on her
campaign Twitter account, @SharronAngle, and Senator Reid on his campaign Twitter account,
@HarryReid, from October 1 2010 through Election Day (November 2, 2010) to understand how
Twitter was used in this campaign. This includes 127 tweets from Sharron Angle and 458 tweets
from Harry Reid.2
It will also compile data on the retweeting of these tweets. Lastly, it includes a
content analysis of news stories related to the campaign in the Las Vegas Review-Journal during
this time frame. Analyzing all of this data will pave the way for an analysis of how candidate
Twitter usage relates to the news coverage of the campaign, how it relates to campaign
strategies, and how it relate to the viral nature of Twitter.
The tweets were numbered and their date and time was recorded. They were then be
coded for usage of common hashtags. I chose to code for the three most frequently used hashtags
per candidate. For both, this includes #nvsen. For Reid, the other two include #p2, which means
progressives 2.0, and #GOTV. For Angle, the other two include #tcot, which means top
conservatives on Twitter, and #dumpreid. I coded for these hashtags since hashtags are used to
categorize tweets and make them more accessible to people on Twitter. I compared the hashtag
use to the amount of retweets.
Next, I coded based on the messages the tweets attempt to relay. Since it is possible to
relay more than one type of message per tweet, this is not mutually exclusive. Categories include
	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  
2	
  I thought a census of all the tweets from the last 33 days of the election would allow me to have a better
understanding of how campaigns were using Twitter. I came to this conclusion after using the Web site
www.tweetstats.com to get data on total tweets from each candidate. The data provided showed a sharp increase in
twitter usage as Election Day neared. Therefore, I thought analyzing the tweets from the time period where
candidates were tweeting the most would be more useful. It also allows me to see if specific uses of Twitter change
as Election Day nears.
	
  
  29	
  
Get out the vote (GOTV), other calls to action, campaign events, news stories, and issue
mention.
GOTV messages include tweets that ask people to vote, ask people to get out the vote, or
relay information about voting. Included in this category are candidate retweets of Twitter users.
Examples of voting information include information about polling locations and when people
can vote.
Other calls to action focus on something other than getting people to simply vote for the
candidate or push others to do the same. Examples include fundraising asks, event sign-ups, and
email sign-ups. Tweets asking people to show their support on Twitter, such as through changing
your Twitter picture to a campaign badge, are also included in this category.
Campaign events are tweets that relay information about campaign events, including
dates, time, locations, topics discussed, type of events, and guest speakers. If images from events
are posted, they will be counted in this category.
News story messages are tweets that reference campaign news stories. They can include
links to actual stories or just reference the story. References to polling data released will be
included in this category. For example, a tweet that says ABC exit poll shows Angle up by 2.7
would be counted here.
Issue mention messages include tweets about a substantive issue. For example, mentions
of healthcare, the economy, immigration, jobs, and education initiatives would be included.
Tweets that do not give specific issues, but relay campaign promises such as “will fight for
Nevada” will also be counted here. I specifically coded for the national issues of healthcare,
immigration, the economy, government spending, jobs, big government, and education.3
I
	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  
3	
  Please note: I combined jobs with economy and government spending with big government when
analyzing the data.	
  
  30	
  
also coded for campaign-specific issues that arose over the course of the campaign. This includes
Angle’s ignoring of the mainstream media, the Hispanic vote, dairy farming, nuclear waste
in Yucca, and Ashjian recording his conversation with Angle. See the appendix A for
definitions and specifics about how these were coded.
Next, tweets were coded based on their “voice”. Categories looked at include personal
observation, retweeted, and inclusive language. Personal observation tweets are tweets that
sound genuine and like they are coming from the candidate. For example, a tweet like “great
turnout at the event” would count as personal observation. Another example could be “thank you
for helping spread the message.”
User Retweets are tweets from other Twitter users that the candidate reposts, with or
without adding their own message. Inclusive language includes tweets that use language to
include the reader of the tweets. The words that count in this category are us, we, you yours, and
our.
Next, I marked whether or not links are present. Analyzing the types of links will help
answer the overarching question about the role of Twitter and provide insight into answers for
my sub-questions. Types of links were separated into whether they link to an outside source, a
site that is not owned or managed by the campaign, link to the campaign, via a campaign Web
site, link to party, via a party-owned site, or video link, via YouTube. I also marked whether or
not there is a photo link.
Lastly, I looked at the general purpose of the tweet. Tweets will be categorized as either
self-promotion or attack and must be put into the category they fit best into. Self-promotion
tweets include all tweets that promote the campaign and put the candidate in a positive light.
Tweets that are categorized as GOTV, other calls to action, and campaign events were included
  31	
  
in this category. Attack tweets attack the opponent in some manner. Attack tweets could be
found in issue mention tweets.
One way that I looked for attacks is by coding separately for opponent mention, tweets
that mention the opponent by name in the 140 characters. If links are present, the attack will be
considered indirect, whereas if they are not mentioned, the attack will be considered direct.
The number of times each tweet is retweeted was then recorded. This helped me
formulate conclusions about what types of tweets and messages used were more successful in
getting followers involved and reaching a wider audience. Specifically, this will help me
determine if the use of inclusive language leads to more retweets.4
Campaign news stories in the print version of the Las Vegas Review-Journal, the largest
newspaper in Nevada, during this time frame was also coded. The stories were selected through a
Lexis Nexus search of stories in the Las Vegas Review-Journal that mentioned “Nevada Senate
race”, “Reid”, or “Angle”. In total, this includes 225 articles. However, the only stories in the
search that were coded were those that reference the Nevada Senate race in at least two
paragraphs. Therefore, any stories that are not focused on this specific race were discarded. In
total, 131 stories were coded.
The stories were categorized based on their type of story, the issues mentioned, and also
whether articles as a whole are positive, negative, or neutral toward each candidate. I coded for
both type of story and issues so that I could better compare the types of stories and the issues
they mention with candidate tweets at that time.
	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  
4	
  It is important to note some issues that arise with using the Twitter function to look at retweets. The retweets
recorded by Twitter only occur when people hit the retweet button. If people copied the message and added their
own words it would not be known.Therefore, there is no way to know how many people did not hit this button, but
still retweeted the message by manually typing the candidate’s message into their own tweets.
	
  
  32	
  
Types of story include process, horse race coverage, event stories, personal information,
endorsements, and issue pieces. Process includes information about the electorate, predicted
turnout, important dates, and campaign finance stories. Horse race coverage includes stories
about who is up or down in the polls. Event stories are stories about specific campaign events,
such as the Reid rally that Michelle Obama went to. Personal information stories are stories
about the candidates. These stories may allude to the morals of the candidates, personality traits,
and information about their lives. Endorsements are stories about whom the paper or other
figures endorse. Issue pieces are stories about specific campaign issues, like healthcare. Releases
of news stories compared to releases of stories on Twitter will then be compared to determine
whether stories most often appear first on Twitter or in the mainstream press.
Next, I coded for the issues. I used the same set of national and local issues that I used to
code the Tweets listed above so I can compare issue mention on Twitter and in articles. I coded
for whether the article is positive, neutral, or negative toward Angle and toward Reid. To
determine this, I looked at the words used to describe the candidates and their positions. For
example, if Reid is called a socialist or his policies are referred to as reckless, I would code the
story as negative. If Angle were described as a Washington outsider, I would code the story as
positive toward her.
I chose this specific research design for a variety of reasons. Since I wanted to figure out
how candidates were using Twitter, a content analysis would allow me to do a much-needed in-
depth analysis of a specific campaign. I also compared Twitter use to the mainstream media
environment. Because there is so little known about Twitter use in campaigns, looking at just
one race in depth allows me to look at more variables and have a deeper analysis than would
have been possible if I was looking at a variety of races more generally.
  33	
  
Nevada was a good state for examining how Twitter was used in the 2010 campaign for a
variety of reasons. First, it was labeled a toss-up from April through Election Day. Second,
Sharron Angle was a Tea Party-endorsed candidate who represented the disgruntled,
conservative wing the of electorate, while Harry Reid was the Senate Majority Leader, largely
responsible for decisions made by the Democratic Congress. Harry Reid represented a
Washington insider, while Angle was a fresh, new figure. Thus, this gave added media attention
to the race. I also thought this race would be interesting to study because Angle ignored the
mainstream media throughout the campaign.
VIII. Limitations of this Study
Using a content analysis does have some limitations that must be addressed. Through a
content analysis, there is no way of knowing if the candidates themselves are tweeting or if
someone on their campaign staff is tweeting for them. Therefore, there is no way to tell how
genuine the tweets are. This is relevant since I am coding for inclusive language and personal
observation. I also do not know what campaigns did with the information they had on their
retweeters or Twitter followers or what their strategies were. This limitation could be addressed
by talking to the campaigns directly and asking them who did the tweets and what goals the
campaigns had for Twitter use.
There are also limits in just studying the Nevada Senate race. Since this race was so
unique, it also presents limitations for my study. It may not be reflective of the use of Twitter for
other tight Senate races that did not have the same factors at play, thus limiting the scope of
conclusions I can make. Also, since Twitter is used more with younger demographics and
Nevada has a lot of retirees, the audience Twitter was able to reach may have been smaller than
  34	
  
in other states. Lastly, a majority of the research discussed in my literature review is about
presidential races, not congressional races. Therefore, uses may differ.
Another limitation is that I only looked at one form of mainstream media. The print
version of the Las Vegas Review-Journal, which only comes out once a day, makes it difficult to
know if Twitter jumpstarts the discussion, or if the discussion is already going on via online
news outlets such as blogs and stories that are just posted online. However, it is extremely hard
to compile all of the online sources that discuss the campaign and determine whether or not those
sources have a following. The readership of the newspaper is clear and the articles are well
documented and accessible.
VIII. Findings
The findings of this study deviated somewhat from what was expected. In regards to
some of the research questions, data provided substantial evidence in support of the hypothesis.
In regards to other questions, the sample size was too small to obtain high confidence level from
the results of this study. Finally, in some other cases, the data proved the hypotheses were
incorrect.
To check for reliability, a second person also coded 11 percent of the tweets and news
articles used in this study. The coding of tweet variables between both coders varied between
83% and 100%. It should be noted that the news stories and issue mentions had an 83 percent
agreement while every other variable had a 94% or higher agreement. For the news articles, the
coding of the issues varied between 77% and 100%. The three categories that only had 77%
agreement were Ashjian’s recorded conversation with Angle, jobs, and government spending.
This may have been because jobs were so similar to economy and big government was so similar
to government spending. For the exact breakdown, see Appendix C.
  35	
  
A. Findings on the Relationship Between Twitter and the Mainstream Media
In reference to the first subset of questions, the interaction between Twitter and the
mainstream media, there was not enough data to provide a high level of confidence in the
findings from this study. This was mainly because tweets were infrequently about campaign
issues for long periods of time. Overall, 201 of the total tweets (34 percent) referenced an issue
of some sort. However, the issues mentioned varied across eleven different topics. The topic that
got the most mentions was the economy with 44 total tweets. Within the timeframe of 33 days,
this is not enough data to be able to make definitive conclusions about the timing on Twitter
compared to the newspaper.
Although the dataset’s size was limiting, some patterns were evident when comparing the
way issues were covered in the Las Vegas Review-Journal and by the candidates on Twitter. For
overarching campaign issues that defined the 2010 Midterm election nationwide such as the
economy, healthcare, and big government, coverage in Twitter and the newspaper remained
fairly consistent over the observed time period. In all three cases, the issues were mentioned
more frequently on Twitter than in the Las Vegas Review-Journal. However, no clear pattern
emerged showing mention in one source preceded mention in the other source. At best, this data
supports the hypothesis that candidates use Twitter to continue issue discussions in order to
weigh into the discussion and try and get their point of view across. For example, both
candidates used Twitter to talk about the economy in about 7 percent of their tweets. See figure
one for a graph of economic mentions in both sources.
  36	
  
5
For localized issues, a different pattern emerged than was seen for national issues. Most
of these issues were covered more frequently in the Las Vegas Review-Journal than they were by
Harry Reid and Sharron Angle on Twitter. This was the clearly the case for the Hispanic vote,
immigration, nuclear waste in Yucca and Ashjian’s recorded conversation with Angle. Thus,
Twitter was not effectively used to lengthen discussion of these issues. As an example, see figure
2, which shows the mentions of Angle ignoring the mainstream media in tweets and articles.
Lastly, an interesting pattern emerged with issues that are bad for the opposition. As I
predicted, in these instances, Twitter was used by the opposing candidate to continue issue
discussion. For example, Reid tweeted about Ashjian leaking his conversation to the press six
times, tweeted about how Angle ignores the press twenty-one times, and mentioned the Hispanic
vote seventeen times, while Angle never mentioned any of these issues on Twitter. This study
still could not conclusively say what came first. However, figure two below indicates that Angle
ignoring the mainstream media was being discussed in the month of October the Las Vegas
Review-Journal before by Harry Reid on Twitter.
	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  
5	
  see appendix B for the rest of the issue graphs
	
  
  37	
  
B. Findings on How Campaigns are Using Twitter
In regards to the second set of questions about how the campaigns used Twitter, the data
showed that tweets were slightly more likely to be promotional than attack. Combined, 52
percent of the total tweets were promotional while 48 percent of tweets were attack tweets.
However, when differentiated by candidate, a clear difference in purpose arose. Angle’s tweets
were used as a means of promotion 66 percent of the time while only 49 percent of Reid’s tweets
were used as a means of promotion. In fact, slightly more of Reid’s tweets were attack tweets
than promotional tweets.
Since I thought attacks would be indirect via the mention news stories and/or use links, I
also compared whether or not news stories were present in them. I found that 52.5 percent of
attack tweets referenced news stories while only 40.8 percent of promotional tweets referenced
news stories. Thus, these findings support my hypothesis that attacks via Twitter are more likely
to be indirect since a majority of the time they reference a news story.
  38	
  
To better answer this question, I also looked at whether links were present in attacks. I
predicted that attack tweets would use links to outside sources as a way to curb negativity from
being attributed to the candidate. If they were directly linking to their site or not linking at all, the
attack would more directly come from the candidate, making him/her seem more negative. I
found that over three-fourths of attack tweets included links to either the campaign site or an
outside site. Interestingly, the vast majority of these links were to the campaign site, which
appears to show that attacks are direct. I found that 59 percent of all attack tweets went directly
to a campaign site while only 19 percent went to an outside source.
While this would seem to undermine the claim that attacks are indirect, this can be
partially attributed to the fact that Harry Reid had a news section on his Web site that displayed
articles from outside sources. This section included positive coverage about Reid and negative
coverage about Angle. This page had been counted as links to the campaign Web site.
Since I had not specifically coded the links to a news page on a campaign site, I cross-
referenced the tweets that included news stories against the tweets that linked to the campaigns’
web sites and found that there were 171 of such tweets. Then, I compared those tweets with the
tweets whose general purpose was to attack and found that 34 percent of the attack tweets
included a news reference and linked to the campaign. Therefore, candidates were frequently
tweeting news articles that they had posted on their campaign Web site to attack their opponent
indirectly and also drive potential voters to their site. This was noteworthy since Obama had used
Twitter in 2008 to drive people to his Web site, therefore suggesting a similar strategy.
To figure out how candidates were using Twitter, I wanted to see how they were
integrating their Twitter strategy with other campaign strategies. Links to their campaign site was
one obvious way to do this. In total, the campaigns linked to their own site 51 percent of the
  39	
  
time. Angle linked to her site only 25 percent of the time while Reid linked to his site 58 percent
of the time. Angle’s infrequent use of linking to her own site differentiates her strategy from
Reid and Obama in 2008. Since this did not account for all the campaign related links, I also
recorded the number of times tweets linked to a video or to a photo.
Both candidates used links to YouTube videos in about 14 percent of their total tweets.
The vast majority of these video links were to campaign advertisements posted on the
candidate’s YouTube page. Thus, Twitter was also being used to drive people to the candidates’
YouTube pages. Photo links were only used in about five percent of the total tweets. This made
sense since the photos linked were taken at campaign events and only fifteen percent of the
tweets referenced campaign events. See figure three to understand the total distribution of links.
C. Findings on Types of Information Candidate Tweet About
  40	
  
In regards to the second set of questions and the types of information campaigns tweet
about, this study found that the campaigns most frequently tweeted news stories. See figure four
for the complete breakdown. In total, 46 percent of tweets references news stories. The next most
frequent use of Twitter was to discuss an issue, which was over a third of the tweets. Therefore,
my hypothesis that campaigns were most frequently using Twitter to make calls to action was
incorrect. Only about a quarter of total tweets were GOTV efforts or other campaign asks.
Although less than a fifth of the tweets had a GOTV message, I was correct that toward
the end of the campaign cycle, GOTV efforts were more prevalent. In fact, over sixty percent of
GOTV tweets occurred in the last week of the campaign cycle. On Election Day, the GOTV
message took over. Out of the 24 Election Day tweets, 21 of them (88 percent) had a GOTV
message. This shows that candidates were integrating Twitter into their overall campaign
strategy to promote potential supporters to vote. See figure five to understand how GOTV tweets
increased drastically in popularity toward the end of the campaign.
  41	
  
*Please note: figure shows number of GOTV tweets on a give day
In order to fully understand how candidates were using Twitter, I also recorded the
number of their tweets that were retweets, the number of tweets that were personal observations,
and the use of inclusive language (we, your[s], our[s]). Retweets were recorded to help measure
how candidates were engaging with their followers. I found that 12 percent of the total candidate
tweets from October 1-November 2 were user retweets (13.5 percent of Reid’s tweets and 8.4
percent of Angle’s tweets). Thus, Angle and Reid were using Twitter to engage with followers
by retweeting them.
Since all of the current studies on Twitter suggest that using an authentic voice enables
politicians to most effectively reach followers, I expected that both Reid and Angle would use an
authentic voice. I coded for authenticity based on whether tweets were categorized as personal
Figure	
  5:	
  GOTV	
  Tweets	
  Across	
  Time	
  
  42	
  
observation or if they used inclusive language, or did both. Surprisingly, only 10 percent of the
total tweets were coded as personal observation and only 24 percent used inclusive language.
This overall percentage was greatly impacted by Harry Reid, who only used personal
observation four percent of the time. This may be because the Reid campaign used an outside
firm, Well & Lighthouse LLC, to do all of the campaign tweeting. Instead of making the tweets
sound like they were coming from Reid, they chose to make the Twitter account sound like it
was coming from the campaign office. Therefore, Reid’s tweets often referenced him in the third
person.
Sharron Angle on the other hand used personal observation 29 percent of the time. This
may have been because Sharon Angle was not accessible to supporters through the mainstream
media, and therefore needed to communicate directly with voters through alternative venues, like
Twitter. Angle’s tweets never referred to her in the third person. In sum, tweets that failed to be
seen as personal observation simply sounded generic for both candidates.
D. Findings on Viral Nature
The third area of questions relates to the viral nature of Twitter. To examine this concept,
I looked at the number times each candidate tweet was retweeted and the presence of hashtags in
the tweets. Candidate tweets averaged 14 retweets for the entire data set. However, it is
important to note that Angle’s average number of retweets was three times higher than Reid’s
average number of retweets. Angle averaged about 30 retweets while Reid averaged about 10
retweets. This may have to do with the frequency each candidate was tweeting. Reid tweeted an
average of fourteen times a day while Angle tweeted and average of five times a day. Reid also
would post the exact same tweet multiple times in a single day or two-day span. Therefore, his
followers may not have been as willing to retweet information since it was repetitive. Another
  43	
  
possibility is that people were less willing to retweet Reid’s tweets because they did not sound
like they were coming from him.
This content analysis suggests that there is a relationship between use of inclusive
language and number of retweets. Looking at both candidates, the mean number of retweets is 18
when inclusive language is present and 13 when it is not present. The difference in the mean
number of tweets is significant (t=3.82, p<.05). However, this relationship was highly driven by
Angle. Each candidate was individually analyzed based on number of retweets and presence of
inclusive language. For Reid, the difference between average number of retweets when inclusive
language was and was not used was less than one. On the other hand, for Angle the average
number of retweets was five more when inclusive language was used, (33 retweets) than when it
was not used (28 retweets).
Another factor at play may have been use of hashtags. Angle used #nvsen in 97 percent
of her tweets, thus making them easily accessible on Twitter, while Reid used #nvsen 88 percent
of the time. More noticeable was the difference in Angle’s inclusion of hashtags to be used by
likely supporters. Angle also used the hashtag tcot, “top conservative on twitter”, 91 percent of
the time, while Reid only used the hashtag p2, “progressives 2.0”, 6 percent of the time. She also
used the hashtag dumpreid 86 percent of the time. Reid failed to consistently use a similar slogan
hashtag.
I then looked at the tweets that used #tcot and those that did not use #tcot and a difference
emerged. The mean number of retweets was 31 when she used it and dropped to 27 when she did
not use it. However, this was not statistically significant because the sample size for tweets that
did not have #tcot is only 12. A similar pattern did not emerge with #dumpreid. In this case, the
difference in number of retweets was less than one.
  44	
  
IX. Conclusions
This study found that Sharron Angle and Harry Reid had different approaches to using
Twitter. Angle tweeted less frequently, sounded more genuine, and perhaps in return was
retweeted more and had more followers. Reid used Twitter to amplify the same campaign
messages he was using elsewhere and to drive people to his Web site and YouTube page.
In regards to answering the question,” was Twitter used to jumpstart the discussion of
campaign issues, or continue existing discussions”, the small dataset in this study gave limited
results, which limited my ability to speak to this question. The study suggests that for national
issues, Twitter is used to continue discussion about the issues being talked about in the
mainstream media. This study could not tell which came first because they were often on the
same day and also because these issues were already present in the campaign before October 1.
However, for smaller, more localized issues, it suggests this is not the case. The issues are often
discussed more in the paper. If they were discussed on Twitter, it was by the opposing candidate.
In regards to the question,” are candidates using Twitter as a means to attack their
opponent or to relay positive information about themselves”, this study found that candidates are
using Twitter for both. Overall, Twitter was used slightly more for promotional purposes.
However, the amount of tweets used for promotion and attack varied greatly across candidates.
Angle only used a third of her tweets to attack Reid while Reid used over half of his tweets to
attack Angle.
In regards to the question, “if attacks are present, are they direct or indirect”, this study
found that attacks were mostly indirect. In other words, the majority of the attacks referenced
news stories and/or had links present in them. More interestingly was the fact that over a third of
attack tweets mentioned news stories and linked back to the candidate’s own Web site. Thus,
  45	
  
Twitter can be used to drive traffic to the candidate Web sites by posting news stories. This may
be because referencing an outside source increased the credibility of the attack and because it is a
common practice in other aspects of campaigning.
In regards to the question, “are candidates using Twitter more frequently to inform
people about campaign events, post news, discuss issues, reinforce campaign messages, or for
campaign asks (i.e. donations, GOTV, signing up for events)”, this study found that candidates
most frequently use Twitter to post news and discuss the issues. Since Twitter is so
instantaneous, it is not surprising that it is being used to release campaign news.
In regards to the question, “does the purpose change over the course of a campaign,” this
study found that the answer is yes. At the beginning, Twitter was used more to discuss campaign
issues and news regarding issues. However, as Election Day neared, GOTV messages flooded
the Twitter accounts of both candidates. This suggests that like with other forms of social media,
Twitter was used at the end of the campaign to drive supporters to the polls. This finding along
with the finding that 62 percent of the tweets link back to a candidate’s Web site or YouTube
page suggests that Twitter use in campaigns may align better with the reinforcement theory than
the mobilization theory.
In regards to the question, “does inclusive language makes people more likely to retweet
the candidates,” this study’s findings suggest that it does. The difference in the mean number of
tweets is significant (t=3.82, p<.05) when inclusive language was used. However, this was
driven by Sharron Angle whose mean number of retweets rose by 5 when she used inclusive
language. On the other hand, for Harry Reid, who only used inclusive language 4 percent of the
time, there was no difference in mean retweets. Since Harry Reid averaged much fewer retweets
than Sharon Angle (10 vs. 30) and rarely used inclusive language, this supports the belief among
  46	
  
experts that sounding genuine on social media matters. Inclusive language is one way to sound
more genuine, and Sharron Angle shows that this also helps expand reach throughout the
“twittersphere”. This also supports the reinforcement theory because as President Obama
exemplified in the 2008 election cycle, using inclusive language is a way to activate your base.
In regards to the last question,” does using consistent hashtags lead to more retweets,”
this study suggests that the answer is yes, it does. Sharron Angle who consistently used hashtags
got retweeted an average of three times more than did Reid. Angle’s mean number of retweets
rose by four when she included #tcot, which she used over 90 percent of the time, than when she
did not. This was not statistically significant because of the small sample size in tweets without
#tcot. Unlike Angle, Reid only used #nvsen 88 percent of the time and #p2 6 percent of the time.
Sharon Angle’s consistent use of the hashtags nvsen and tcot are only one way her Twitter use
was different than Harry Reid. Therefore, it is important to take into account the other
differences in Twitter style between the candidates and realize hashtags may only be one
contributing factor to her ability to get retweeted.
As the Internet becomes more integrated into campaigns, it is important to continue
studying how new innovations online contribute to campaign strategies. As history shows,
strategies online are constantly evolving and studies must stay up to date. For example, YouTube
was not around prior to the 2006 campaign cycle and candidates did not use Twitter before the
2008 cycle. Since innovations online are happening so quickly, it is important that each election
cycle, questions are raised on how candidates are integrating the new techniques into their online
strategy and overall campaign strategy so that there is no gap in the literature on this subject.
The increasing trend in people turning to the Internet for campaign information proves
that no online source of information distribution can be forgotten. Before this study was
  47	
  
conducted, there were very few studies available on Twitter use in campaigns. This is partially
because before 2010, it had only been used in four major races. Unlike with older online outlets
like campaign Web sites and YouTube, there is no uniformity within Twitter strategy. Therefore,
this study shows that candidates still have a great deal to learn about effective Twitter use.
This study is important because it increases the available knowledge on how Twitter can
be most effectively used. It shows that Twitter can and should be incorporated into other online
strategies through use of links. This could be links to the candidate’s Web site, other social
media sites, or even outside news articles. Since posting on Twitter is so immediate, it can be a
useful outlet to release campaign information on or for posting breaking news stories that are
favorable to the candidate.
Based on this study, I conclude that even if the candidate is not the one tweeting, those
receiving the messages must believe that the tweets are coming from the candidate. If voters do
not think this is the case, they may be less susceptible to become activate supporters on Twitter
or elsewhere. This is mainly because there was a significant difference in number of retweets
when inclusive language was used in this study.
I believe that the best use of Twitter in campaigns is twofold. Candidates should use
Twitter to reach a wide audience that may be missed in other forms of campaign strategy and to
energize their base as Election Day nears. This audience specifically includes the younger voters
who may not actively seek out campaign information. If candidates can use Twitter to get
retweeted by people with large Twitter followings, their messages will be broadcasted to a large
audience of people who are not choosing to seek out political information. If they make their
tweets interesting and include links to their Web site or to their other social media pages, they
can drive potential voters to sites where they have complete control over the message. If they can
  48	
  
consistently use hashtags and inclusive language, they may engage more people and help their
messages go viral.
Secondly, Twitter should be used like other online forms of social media to energize and
activate the candidate base. This can be done by using hashtags that your base will be likely to
see, linking to your site, including campaign asks, and using inclusive language. For example, if
you are a conservative candidate, you can use the hashtag tcot to specifically reach conservative
voters. This can be used to reach conservatives in Nevada and elsewhere. Since campaign asks,
like fundraising, are not limited to state residents, Twitter may be used to target out of state
contributors.
This study creates a baseline to which Twitter usage in the future can be compared. It
should be noted that the number of people following these candidates was low and that the
number of retweets was also relatively low. However, since other forms of social media, like
YouTube and Facebook, did not catch on instantly, but grew over time, the same trend is likely
to occur with Twitter. For example, today, candidates are expected to invest in a Web site, but
just 10 years ago, less than a majority of candidates for Congress had Web sites. I expect that as
time passes, more candidates and more citizens will use Twitter. This study paves the way for
further analysis on how candidate use of Twitter is different than candidate use of other forms of
social media. Candidates must understand Twitter as a medium to use it to its maximum
potential.
This study could be expanded upon in a few different ways to give a more complete
picture of how Twitter use influenced the campaigns. First, this study did not look at actual
fundraising data and how it compared to Twitter use. Looking at this data would have helped
determine if Twitter can help candidates fundraise as other online technologies have done and
  49	
  
see support the notion that Twitter can be used to reach potential contributors nationwide. It also
does not compare increase in number of followers to fundraising patterns.
Secondly, this study was unable to compile date on the users that retweeted the candidate
such as how many followers they had and what their ties to the election were. Getting this
information would help further answer the question of how large of an audience the tweets are
reaching and help campaigns understand who they should target on Twitter. If celebrities are
retweeting candidates, they may have a much larger Twitter following than the candidates
themselves and be able to reach people who would not typically be motivated to search for
campaign information. Also, if those retweeting and/or following the candidates are journalists,
the candidates’ messages may have a wider reach because of the amount of influence journalists
have on agenda setting and because of the network of people those journalists can reach.
Thirdly, this study only looked at the Las Vegas Review Journal in print. Perhaps looking
at other mainstream news outlets in both print and their online publications would help determine
who discusses the issues first. To further address that question, a longer time period could also be
observed.
The Nevada Senate race was one of the most monumental races in the 2010 Midterm
elections because it captured the mood of the electorate during that election cycle. The fact that a
Tea Party candidate who had radical beliefs including abolishing the Department of Education
could be in a predicted dead heat with the Senate Majority Leader showed how disgruntled
voters were with the Democratic Congress and Obama administration. Studying how these two
extremely different candidates used the newest online outlet, Twitter, provides insight into how
candidates in general can use this outlet to reach voters and win their race. In close races, like
this one, every piece of the overall campaign strategy matters. Therefore, political
  50	
  
communications experts must continue studying new innovations as they begin to be integrated
into campaigns to figure out best practices and help candidates convince more citizens to vote.
Appendix A
Coding Sheet: Tweets
1. Tweet number
a. Write the ID of the tweet.
2. Date
a. Write the date of the tweet.
3. Time
a. Write down the time the tweet was recorded.
4. Hashtags (For Reid: #nvsen, #gotv, #p2) (For Angle: #nvsen, #dumpreid, #tcot)
a. See if each of the assigned hashtags are present or not present
b. If present, write a 1, if not present, write a 0
5. Message of the Tweet
Since it is possible to relay more than one type of message per tweet, this will not be
mutually exclusive, but rather each type will be marked with “yes” or “no”. Categories
include GOTV, other calls to action, campaign events, news stories, and issue mention
a. GOTV
i. Definition: GOTV messages include tweets that ask people to vote, ask
people to get out the vote, or relay information about voting. Included this
category are candidate retweets of Twitter users. Examples of voting
information include information about polling locations and when people
can vote.
ii. Is there a GOTV message in the tweet?
1. If yes, mark 1
2. If no, mark 0
b. Other Calls to Action
i. Definition: Other calls to action focus on something other than getting
people to simply vote for the candidate or push others to do the same.
Examples include fundraising asks, event sign-ups, email sign-ups, and
asking to take action on twitter. Tweets asking people to show their
support on Twitter, such as through changing your Twitter picture to a
campaign badge, are also included in this category.
ii. Is there an other call to action message in the tweet?
1. If yes, mark 1
2. If no, mark 0
c. Campaign events
i. Definition: campaign events are tweets that relay information about
campaign events, including dates, time, locations, topics discussed, type of
events, and guest speakers. If images from events are posted, they will be
counted in this category.
ii. Is there a campaign event message in the tweet?
1. If yes, mark 1.
  51	
  
2. If no, mark 0
d. News stories
i. Definition: News stories are tweets that reference campaign news stories.
They can include links to actual stories or just reference the story.
References to polling data released will be included in this category. For
example, a tweet that says ABC exit poll shows Angle up by 2.7 would be
counted here. When there are links to news stories, the links will also be
coded by the same schema used to code news stories from the Las Vegas
Review-Journal.
ii. Is there a news story mentioned in the tweet?
1. If yes, mark 1
2. If no, mark 0
e. Issue Mention
i. Definition: include tweets about a substantive issue. General campaign
rhetoric does not count here. Mark each specific issue as follows:
1. 1=healthcare; 2=immigration; 3=economy; 4=government
spending (includes stimulus and social security); 5=jobs (including
unemployment); 6=big government (includes taxes); 7=education;
8=Angle’s ignoring mainstream media; 9=the Hispanic vote;
10=farming; 11=nuclear waste in Yucca; 12=Ashjian recording his
conversation with Angle; 13=Pork barrel spending; 14=other
2. If no issue is mentioned, mark 0.
6. Personal Observation
a. Definition: Personal observation tweets are tweets that sound genuine and like
they are coming from the candidate. For example, a tweet like “great turnout at
the event” would count as personal observation. Another example could be
“thank you for helping spread the message.”
b. If present, mark 1; if not, mark 0.
7. User Retweets
a. Definition: User retweets are tweets from other Twitter users that the candidate
posts, with or without adding their own message.
b. If present, mark 1; if not, mark 0.
8. Inclusive language
a. Definition: Inclusive Language includes tweets that use language to include the
reader of the tweets. Specific words that count in this category are us, we, you,
your, yours, and our.
b. If present, mark 1; if not, mark 0.
9. Links
a. I will code for three mutually exclusive types of links. Types of links include:
i. Link to the campaign, via a campaign Web site or official social media
site (i.e. YouTube, facebook).
ii. Link to party site, a site that is owned by a particular party. If a tweet
links to the DNC, it would count here.
iii. Link to an outside source, a site that is not owned or managed by the
campaign.
iv. Photo link, a link to some sort of photo sharing site, like twitpic.
Kottler Thesis 2011
Kottler Thesis 2011
Kottler Thesis 2011
Kottler Thesis 2011
Kottler Thesis 2011
Kottler Thesis 2011
Kottler Thesis 2011
Kottler Thesis 2011
Kottler Thesis 2011
Kottler Thesis 2011

More Related Content

What's hot

Political participation and social media usage
Political participation and social media usagePolitical participation and social media usage
Political participation and social media usageFJWU
 
Research Proposal : Political Representation of Different types of voters on ...
Research Proposal : Political Representation of Different types of voters on ...Research Proposal : Political Representation of Different types of voters on ...
Research Proposal : Political Representation of Different types of voters on ...Joshua Wong
 
the complete draft about the CA election time tweets -- awaiting final weedin...
the complete draft about the CA election time tweets -- awaiting final weedin...the complete draft about the CA election time tweets -- awaiting final weedin...
the complete draft about the CA election time tweets -- awaiting final weedin...japokh
 
Triple helix 2012 president
Triple helix 2012 presidentTriple helix 2012 president
Triple helix 2012 presidentHan Woo PARK
 
Outhwaite turner chp44_final (1)
Outhwaite turner chp44_final (1)Outhwaite turner chp44_final (1)
Outhwaite turner chp44_final (1)Rohanbhatkal2
 
Predicting Elections with Twitter
Predicting Elections with TwitterPredicting Elections with Twitter
Predicting Elections with TwitterTimm Sprenger
 
E marketer the_new_political_influencers-social_medias_effect_on_the_campaign...
E marketer the_new_political_influencers-social_medias_effect_on_the_campaign...E marketer the_new_political_influencers-social_medias_effect_on_the_campaign...
E marketer the_new_political_influencers-social_medias_effect_on_the_campaign...AdCMO
 
Twitter Based Outcome Predictions of 2019 Indian General Elections Using Deci...
Twitter Based Outcome Predictions of 2019 Indian General Elections Using Deci...Twitter Based Outcome Predictions of 2019 Indian General Elections Using Deci...
Twitter Based Outcome Predictions of 2019 Indian General Elections Using Deci...Ferdin Joe John Joseph PhD
 
Information gathering strategies in online social networks
Information gathering strategies in online social networksInformation gathering strategies in online social networks
Information gathering strategies in online social networksAlexander Decker
 
POLITICAL OPINION ANALYSIS IN SOCIAL NETWORKS: CASE OF TWITTER AND FACEBOOK
POLITICAL OPINION ANALYSIS IN SOCIAL NETWORKS: CASE OF TWITTER AND FACEBOOKPOLITICAL OPINION ANALYSIS IN SOCIAL NETWORKS: CASE OF TWITTER AND FACEBOOK
POLITICAL OPINION ANALYSIS IN SOCIAL NETWORKS: CASE OF TWITTER AND FACEBOOKIJwest
 
Media access and exposure as determinants of the political
Media access and exposure as determinants of the political Media access and exposure as determinants of the political
Media access and exposure as determinants of the political Alexander Decker
 
lis 3201 Final presentation
lis 3201 Final presentationlis 3201 Final presentation
lis 3201 Final presentationMonte VanDyke
 
SocialMediaandPoliticalExpressionRoderickandPryde
SocialMediaandPoliticalExpressionRoderickandPrydeSocialMediaandPoliticalExpressionRoderickandPryde
SocialMediaandPoliticalExpressionRoderickandPrydeEmma Roderick
 
Doctoral Seminar in Digital Marketing
Doctoral Seminar in Digital MarketingDoctoral Seminar in Digital Marketing
Doctoral Seminar in Digital MarketingHanani Ahmad
 

What's hot (19)

Political participation and social media usage
Political participation and social media usagePolitical participation and social media usage
Political participation and social media usage
 
Research Proposal : Political Representation of Different types of voters on ...
Research Proposal : Political Representation of Different types of voters on ...Research Proposal : Political Representation of Different types of voters on ...
Research Proposal : Political Representation of Different types of voters on ...
 
the complete draft about the CA election time tweets -- awaiting final weedin...
the complete draft about the CA election time tweets -- awaiting final weedin...the complete draft about the CA election time tweets -- awaiting final weedin...
the complete draft about the CA election time tweets -- awaiting final weedin...
 
Triple helix 2012 president
Triple helix 2012 presidentTriple helix 2012 president
Triple helix 2012 president
 
Outhwaite turner chp44_final (1)
Outhwaite turner chp44_final (1)Outhwaite turner chp44_final (1)
Outhwaite turner chp44_final (1)
 
Predicting Elections with Twitter
Predicting Elections with TwitterPredicting Elections with Twitter
Predicting Elections with Twitter
 
E marketer the_new_political_influencers-social_medias_effect_on_the_campaign...
E marketer the_new_political_influencers-social_medias_effect_on_the_campaign...E marketer the_new_political_influencers-social_medias_effect_on_the_campaign...
E marketer the_new_political_influencers-social_medias_effect_on_the_campaign...
 
Ijrar19 j1432
Ijrar19 j1432Ijrar19 j1432
Ijrar19 j1432
 
Asymmetric polarization
Asymmetric polarizationAsymmetric polarization
Asymmetric polarization
 
Twitter Based Outcome Predictions of 2019 Indian General Elections Using Deci...
Twitter Based Outcome Predictions of 2019 Indian General Elections Using Deci...Twitter Based Outcome Predictions of 2019 Indian General Elections Using Deci...
Twitter Based Outcome Predictions of 2019 Indian General Elections Using Deci...
 
Information gathering strategies in online social networks
Information gathering strategies in online social networksInformation gathering strategies in online social networks
Information gathering strategies in online social networks
 
Wiedman_Op-Ed
Wiedman_Op-EdWiedman_Op-Ed
Wiedman_Op-Ed
 
POLITICAL OPINION ANALYSIS IN SOCIAL NETWORKS: CASE OF TWITTER AND FACEBOOK
POLITICAL OPINION ANALYSIS IN SOCIAL NETWORKS: CASE OF TWITTER AND FACEBOOKPOLITICAL OPINION ANALYSIS IN SOCIAL NETWORKS: CASE OF TWITTER AND FACEBOOK
POLITICAL OPINION ANALYSIS IN SOCIAL NETWORKS: CASE OF TWITTER AND FACEBOOK
 
Media access and exposure as determinants of the political
Media access and exposure as determinants of the political Media access and exposure as determinants of the political
Media access and exposure as determinants of the political
 
lis 3201 Final presentation
lis 3201 Final presentationlis 3201 Final presentation
lis 3201 Final presentation
 
SocialMediaandPoliticalExpressionRoderickandPryde
SocialMediaandPoliticalExpressionRoderickandPrydeSocialMediaandPoliticalExpressionRoderickandPryde
SocialMediaandPoliticalExpressionRoderickandPryde
 
Sm
SmSm
Sm
 
Fys
FysFys
Fys
 
Doctoral Seminar in Digital Marketing
Doctoral Seminar in Digital MarketingDoctoral Seminar in Digital Marketing
Doctoral Seminar in Digital Marketing
 

Viewers also liked

Viewers also liked (11)

Valparaiso conurbación o metástasis
Valparaiso conurbación o metástasisValparaiso conurbación o metástasis
Valparaiso conurbación o metástasis
 
Estado islamico
Estado islamicoEstado islamico
Estado islamico
 
Windows 8
Windows 8Windows 8
Windows 8
 
WebQuest
WebQuestWebQuest
WebQuest
 
Power
PowerPower
Power
 
Effort estimation for software development
Effort estimation for software developmentEffort estimation for software development
Effort estimation for software development
 
CSC124 word project (Autosaved)
CSC124 word project (Autosaved)CSC124 word project (Autosaved)
CSC124 word project (Autosaved)
 
Proyecto adm de diseño
Proyecto   adm de diseñoProyecto   adm de diseño
Proyecto adm de diseño
 
Aprendizaje ubicuo y el aula aumentada
Aprendizaje ubicuo y el aula aumentadaAprendizaje ubicuo y el aula aumentada
Aprendizaje ubicuo y el aula aumentada
 
Laura M Brumbaugh_Vitea_2016
Laura M Brumbaugh_Vitea_2016Laura M Brumbaugh_Vitea_2016
Laura M Brumbaugh_Vitea_2016
 
8 Tips To Create Epic Visual Presentations
8 Tips To Create Epic Visual Presentations8 Tips To Create Epic Visual Presentations
8 Tips To Create Epic Visual Presentations
 

Similar to Kottler Thesis 2011

Final Research Project
Final Research ProjectFinal Research Project
Final Research ProjectMollie Neal
 
Social Media and Politics
Social Media and PoliticsSocial Media and Politics
Social Media and PoliticsVince Carr
 
Did Social Media Really MatterCollege Students’ Use of Onli.docx
Did Social Media Really MatterCollege Students’ Use of Onli.docxDid Social Media Really MatterCollege Students’ Use of Onli.docx
Did Social Media Really MatterCollege Students’ Use of Onli.docxcuddietheresa
 
Did Social Media Really MatterCollege Students’ Use of Onli.docx
Did Social Media Really MatterCollege Students’ Use of Onli.docxDid Social Media Really MatterCollege Students’ Use of Onli.docx
Did Social Media Really MatterCollege Students’ Use of Onli.docxmariona83
 
2016Election Final Draft
2016Election Final Draft2016Election Final Draft
2016Election Final DraftSarah Abel
 
The Revolutionwillbe Networked
The Revolutionwillbe NetworkedThe Revolutionwillbe Networked
The Revolutionwillbe Networkedkylewelch
 
Impact of social media on voting behaviour
Impact of social media on voting behaviourImpact of social media on voting behaviour
Impact of social media on voting behaviourShoeb Khan
 
How and why social media has transformed campaigns for political office
How and why social media has transformed campaigns for political officeHow and why social media has transformed campaigns for political office
How and why social media has transformed campaigns for political officeService_supportAssignment
 
How social media used by politicians? 2016
How social media used by politicians? 2016How social media used by politicians? 2016
How social media used by politicians? 2016Susana Gallardo
 
Social Media and PoliticsLearning objectivesLearning objec.docx
Social Media and PoliticsLearning objectivesLearning objec.docxSocial Media and PoliticsLearning objectivesLearning objec.docx
Social Media and PoliticsLearning objectivesLearning objec.docxjensgosney
 
Twitter And Social Justice
Twitter And Social JusticeTwitter And Social Justice
Twitter And Social JusticeJodi Sperber
 
The Correlation Between Social Media and Voter Turnout
The Correlation Between Social Media and Voter TurnoutThe Correlation Between Social Media and Voter Turnout
The Correlation Between Social Media and Voter TurnoutGordon Gearhart
 
How the Presidential Campaigns Captivate Social Networking
How the Presidential Campaigns Captivate Social Networking How the Presidential Campaigns Captivate Social Networking
How the Presidential Campaigns Captivate Social Networking Real-Time OutSource
 
The effects of social networks as a public relations tool in political commun...
The effects of social networks as a public relations tool in political commun...The effects of social networks as a public relations tool in political commun...
The effects of social networks as a public relations tool in political commun...Gabriela Olaru
 

Similar to Kottler Thesis 2011 (20)

Final Research Project
Final Research ProjectFinal Research Project
Final Research Project
 
Social Media and Politics
Social Media and PoliticsSocial Media and Politics
Social Media and Politics
 
Did Social Media Really MatterCollege Students’ Use of Onli.docx
Did Social Media Really MatterCollege Students’ Use of Onli.docxDid Social Media Really MatterCollege Students’ Use of Onli.docx
Did Social Media Really MatterCollege Students’ Use of Onli.docx
 
Did Social Media Really MatterCollege Students’ Use of Onli.docx
Did Social Media Really MatterCollege Students’ Use of Onli.docxDid Social Media Really MatterCollege Students’ Use of Onli.docx
Did Social Media Really MatterCollege Students’ Use of Onli.docx
 
2016Election Final Draft
2016Election Final Draft2016Election Final Draft
2016Election Final Draft
 
The Revolutionwillbe Networked
The Revolutionwillbe NetworkedThe Revolutionwillbe Networked
The Revolutionwillbe Networked
 
Social media
Social mediaSocial media
Social media
 
Polinter07
Polinter07Polinter07
Polinter07
 
Polinter07
Polinter07Polinter07
Polinter07
 
IDIS399RoughDraft copy
IDIS399RoughDraft copyIDIS399RoughDraft copy
IDIS399RoughDraft copy
 
Impact of social media on voting behaviour
Impact of social media on voting behaviourImpact of social media on voting behaviour
Impact of social media on voting behaviour
 
How and why social media has transformed campaigns for political office
How and why social media has transformed campaigns for political officeHow and why social media has transformed campaigns for political office
How and why social media has transformed campaigns for political office
 
How social media used by politicians? 2016
How social media used by politicians? 2016How social media used by politicians? 2016
How social media used by politicians? 2016
 
Social Media and PoliticsLearning objectivesLearning objec.docx
Social Media and PoliticsLearning objectivesLearning objec.docxSocial Media and PoliticsLearning objectivesLearning objec.docx
Social Media and PoliticsLearning objectivesLearning objec.docx
 
The SMAG Project
The SMAG ProjectThe SMAG Project
The SMAG Project
 
Twitter And Social Justice
Twitter And Social JusticeTwitter And Social Justice
Twitter And Social Justice
 
Monitoring of the Last US Presidential Elections
Monitoring of the Last US Presidential ElectionsMonitoring of the Last US Presidential Elections
Monitoring of the Last US Presidential Elections
 
The Correlation Between Social Media and Voter Turnout
The Correlation Between Social Media and Voter TurnoutThe Correlation Between Social Media and Voter Turnout
The Correlation Between Social Media and Voter Turnout
 
How the Presidential Campaigns Captivate Social Networking
How the Presidential Campaigns Captivate Social Networking How the Presidential Campaigns Captivate Social Networking
How the Presidential Campaigns Captivate Social Networking
 
The effects of social networks as a public relations tool in political commun...
The effects of social networks as a public relations tool in political commun...The effects of social networks as a public relations tool in political commun...
The effects of social networks as a public relations tool in political commun...
 

Kottler Thesis 2011

  • 1. THE  GEORGE  WASHINGTON  UNIVERSITY   SCHOOL  OF  MEDIA  AND  PUBLIC  AFFAIRS     Twitter  and  the  2010   Nevada  Senate  Race   How  Senate  Majority  Leader  Harry  Reid  and   Republican  Candidate  Sharron  Angle  incorporated   Twitter  into  their  campaigns   UNDERGRADUATE  HONORS  THESIS   Rachel  Kottler   4/28/2011         Currently, little data exists on how Twitter, the newest social media phenomenon, is being used in political campaigns. The following study explores the question of what role Twitter played in the 2010 Nevada Senate race. This study seeks to understand how candidates are currently using Twitter and determine how Twitter can be most effectively used to reach potential voters. It compares campaign discussion on Twitter to discussion by in the mainstream media. Specifically, it looks at all of the candidates’ tweets as well as all articles on the race in the Las Vegas Review-Journal from October 1 through Election Day. It also looks at previous online innovations to fully grasp how online mediums, including Twitter, can be used by political campaigns.  
  • 2.   2   Rachel Kottler Thesis 4/26/11 Twitter and the 2010 Nevada Senate Race I. Introduction Just five years ago, the newest social media Web site, Twitter, was created. In 2006, Jack Dosey, Evan Williams, and Biz Stone invented Twitter as a means for people to update each other from anywhere and stay informed about news in their city (Sarno). Now, Twitter is used by millions of people around the globe as a means to communicate with others, share information, get news, or follow their favorite celebrities. All messages, more commonly referred to as tweets, must be 140 characters or less and they can be directed at people with use of @. Conversations are sorted with voluntary use of hashtags (#) that help track topical discussion. Users can promote other people’s tweets by using a function called retweets. As the number of people on Twitter grows, so does the site’s ability to serve as a political platform. It hit the political scene in April 2007 when President Obama activated a Twitter account for in his Democratic Presidential nomination campaign (Allen Hendricks 40). Since Obama’s run for Presidency, Twitter has gained popularity among eligible American voters. According to an August 2010 Pew Internet and American Life Project study, 17 percent of voting age adults use Twitter in some regard (Madden 8). This suggests rapid growth as this rate was only 11 percent in 2009 (Madden 8). Many Twitter users used the site to find out information about the 2010 midterm elections (Smith 10). According to Pew, “28% of Twitter users used the site to engage with the 2010 election in one way or another,” (Smith 10). When asked about all social media sites combined, Pew found that over one in five adult Internet users used these sites for political purposes during the 2010 election cycle (Smith 1).
  • 3.   3   Although Twitter usage by voting-age adults and political campaigns has increased, few studies have been conducted about the function Twitter has in politics. Thus a gap exists in the literature that would help us understand social media’s role in politics. This study will contribute to the understanding of the role Twitter plays through an analysis of the role Twitter played in the 2010 Nevada Senate race. Since a sizeable amount of research is available on how YouTube, Facebook, Web sites, and blogs have been used in political campaigns, I set this study into the broader context, by reviewing these Internet technologies that have influenced campaigns in the last decade. In the findings section of this paper, I will discuss whether Twitter is used differently than other technological advancements or used in a similar fashion. The Senate race in Nevada was selected for a variety for reasons. First, it is one of the 32 states in the country that entertain early voting (Absentee and Early Voting). Second, the drastically different candidates well represent the mood of the 2010 Midterm elections. Senate Majority Leader Harry Reid, who was first elected to Senate in 1986, was up for reelection in 2010 against the Tea Party-endorsed candidate Sharron Angle, a candidate with no national political experience. This Senate race was described as “highly contested” by both polls and the news media prior to Election Day (Nevada Senate—Angle vs. Reid). This race was unique because of Senator Reid’s position in the Senate, and also because Sharron Angle ran on more radical and conservative positions than most Republican Senators in office hold (Fournier). Due to these circumstances, the campaign was widely discussed in the media. Its notoriety through the mainstream media makes it easier to compare and analyze the way the candidates used social media to discuss the campaign to how the mainstream media discussed the campaign.
  • 4.   4   This study will look into a variety of questions in order to propose an answer to the research question: What was the role of Twitter in the highly contested 2010 Nevada Senate race? Questions include whether Twitter was used to jumpstart the discussion of campaign issues, or continue existing discussions? What type of information are campaigns tweeting about? Are they using Twitter as a means to attack their opponent or to relay positive information about themselves? Are the attacks direct or indirect? Are they using Twitter more frequently to inform people about campaign events, share their positions, reinforce campaign messages, or to connect individual voters with the campaign through various campaign asks (i.e. donations, gotv, signing up for events)? To give a background for the study and the questions it seeks to answer, theories about whom the Internet can reach and how the news impacts voters must be discussed. Once these theories are understood, the history of changing technologies and how they follow or deviate from these theories must be studied. This will clarify how previous new innovations online have affected campaigns and also help formulate hypotheses about Twitter. Noteworthy technologies include candidate Web sites, email, blogs, and other social media sites. The current knowledge and ideas about Twitter usages will also be discussed. Finally, hypotheses will be laid out and the findings of the content analysis will be presented. II. Theories i. Theories about the role of the Internet To fully grasp the impact of online technologies, it is important to discuss the different approaches and theories that have been developed on the Internet’s role in campaigns. Understanding these theories will help determine if Twitter use follows or deters from these theories. Most importantly, analyzing these theories and history helped formulate the hypotheses
  • 5.   5   about the role of Twitter in campaigns that this study seeks to test. Two different approaches taken, the instrumental approach and psychological approach, the must be explained in order to explain the two major theories, the reinforcement theory and the mobilization theory, that have been created. These theories attempt to answer the question of whom campaigns should target their online engagement efforts towards based on the question of how Internet works to affect voters. Each theory lies in one of the two broader points of view political scientists have about the Internet and its possible affects on political engagement. The instrumental approach explains that as the cost of communication deceases, people are more apt to participate. (Xenos and Moy 706) According to Bimber ‘‘changes in the cost and variety of sources of information directly affect levels of political participation’’ (Bimber. 2000). In other words, as new, inexpensive technologies emerge, people will be more likely to participate in the political process. If this is correct, since Twitter is free, it will in fact increase citizen engagement in politics. Multiple studies have been conducted that support the instrumental approach. A 2003 study by Johnson and Kaye found a relationship between Internet use and political engagement. (Johnson and Kayne 9-34) Another study in 2003 done by Tolbert and McNeal found that there was a positive correlation between access to campaign news online and political participation. (Tolbert and McNeal 175-185) A 2006 study by Drew and Weaver found that there was a positive relationship between exposure and attention to campaigns and interest level. (Drew and Weaver 25-42). However, these studies fail to address the issue that those who already use the Internet may be more likely to participate in the political realm online. The psychological approach focuses more on motives and attributes of Internet users (Xenos and Moy 706). This approach is based on the media niche theory that concludes, “the
  • 6.   6   Internet offers the broadest array of opportunities for gratifying various needs,” (Xenos and Moy 707). Specifically, this theory adheres to the belief that the differences in “individual’s level of sophistication, characteristics, and social context” impact online engagement (Xenos and Moy 707). In other words, people are more likely to use the Internet to engage in the political process if they are educated, of higher income, and concerned with current affairs and politics. A 2005 study by Prior supports this theory’s claim that motive matters. It found that those who used media for political information were more likely to be engaged than those who used the media for entertainment purposes (Xenos and Moy 708). If this were correct, political discussions on Twitter would only be from those highly informed and highly interested in the campaign, thus agreeing with the reinforcement theory. Therefore, in terms of Twitter, those retweeting would fall into this category. The reinforcement theory believes that more politically active people will engage in politics through online participation. The essential component to the reinforcement theory is that “those with established interests retain the strongest incentive for online participation,” (Klotz 111). Active users would include journalists, lobbyists, party members, and grassroots activists (Norris 218). Pippa Norris applied this overarching theory about campaigning to Internet technologies in 2001 as a contrast to the mobilization theory. She used Pew data on Internet users from 1996 and 1998 to support this theory (Norris 218). Findings by a study done by Bruce Bimber and Richard Davis also support this theory. They found that in 2000, the majority of Web site viewers were already supporters of the candidate (Bimber and Davis 108). As Bimber states in his book Information and American Democracy, “On the whole, those who pay the most attention to the media of previous information revolutions are also paying the most attention to new media, and those most likely to be active in democratic processes in earlier information regimes are those engaged with the new organizational structures of the emergent information regime” (Bimber 230).
  • 7.   7   Bimber and other proponents of this theory do not think that the Internet can engage people who are not already interested in the political process. Instead, it will engage people who were active already. This theory is related to earlier works on political communication effects. Specifically, it applies Converse’s theory of non-attitudes. According to Converse, the majority of Americans do not have a true political ideology. The few that do have a true ideology seek out information that aligns with their partisan preference and reject information that contradicts it (Converse 228- 256). If this theory is correct, it would support a hypothesis that the majority of citizens engaged in campaigns via social media, including Twitter, are already strong supporters of the candidate. This theory could also explain why the number of people following a candidate on Twitter is relatively low since few people hold such strong partisan preferences and are active in politics. By contrast, the mobilization theory argues the Internet is a means to gain supporters. Norris defines this theory as: “The Internet may serve to inform, organize and engage those who are currently marginalized from the existing political system—such as younger generation, people living in isolated peripheral communities, or fridge political minorities disaffected by the traditional system—so these groups will gradually become drawn into public life and civic communities,” (Norris 218). It suggests that the Internet can create incentives to garner new participants into campaigns and politics since the Internet makes it easier to learn candidate positions, reach government officials, and network about issues. This is because it makes it easier for people to have access to political information who may have previously been isolated and easier for those who did not want to spend time and money learning about politics. Thus, the Internet lowers the costs of political engagement. It would focus more on gaining support from those who do not have true ideologies and are still malleable.
  • 8.   8   The mobilization theory is supported from the 2008 Presidential election when Obama used the Internet to connect with young voters and minorities (Kayne). Since Pew reported that as of May 2010 over a quarter of 18-29 year olds were using Twitter, it may act as a means to engage the younger generation as other social networking sites did in 2008 (Madden 8). Evidence in support of this theory through Twitter use could come from retweets and hashtags. If partisans are retweeting they can engage non-partisans by retweeting links and campaign information that their non-partisan followers may not have sought out on their own. People, who feel a stake in the election for specific reasons, like illegal immigrants whose families fates in America would be affected by the winner of this election, may come across the campaign information because of hashtags or retweets and engage themselves. They are more likely to come across these hashtags if they are issue-centered, like #DREAMAct. In either of these situations, the mobilization theory is supported. Both of these theories stem back to the Internet Engagement Model, which seeks to explain how the Internet can be used to engage voters. Each theory uses this model but makes different conclusions about who is engaging with campaigns through the Internet. This model is used to study Internet activity on three levels of analysis: the national context, political institutions, and the individual level. In terms of political campaigns in the US, the national context remains consistent. The part of this model that relates to this topic are the other two levels. Norris explains, “the core institutions of the political system available in the digital world provide the systematic context within which individual citizens have opportunities to participate online. Which particular citizens take advantage of these opportunities is determined by their personal resources and their motivation,” (Norris 15). This model suggests that citizens can become more involved with politics through using the Internet to connect with the system. Their relationships with political groups, parties, and the
  • 9.   9   news media can strengthen, thus allowing individuals to contribute more to governmental decisions. With Twitter, people have a new resource that allows them to communicate directly with those running for office and an easy avenue to become engaged with the campaign. Candidates and news sources can see if people like their messages based on what is retweeted or by looking at hashtags. It also presents itself as an intensely personal medium. People can voice their support for candidates and causes to their immediate circle of followers and with the general twitersphere via hashtag use. Non-politically active voters can stay updated on campaigns simply by following someone on Twitter who chooses to be engaged. Related to this model is the concept of voter efficacy. Political efficacy is defined as “a combination of one’s sense of competence in the political sphere and one’s assessment of the responsiveness of the system,” (Valentino 308). Therefore, if citizens think their actions can influence the political system, they are more likely to participate in politics. This is based on both internal efficacy, which is determined by voter confidence in him/herself to understand the system, and external efficacy, which is comes from the voter’s confidence that the political system is open to everyone and he/she will be heard. Twitter has the potential to enhance external efficacy because it forms an easy route of direct communication with politicians. This way, candidates (or staff members acting on their behalf) can respond to voters and show their constituencies that their opinions matter. ii. Theories about News Coverage As history has shown, Internet communication can affect the campaign news cycle. Therefore, it is important to study whether the same is true for candidate tweets on Twitter. Theories of news coverage must be understood in order to address the question about whether
  • 10.   10   Twitter starts the discussion of campaign information or lengthens it. Specifically, the role of news coverage plays in political campaigns and how news affects people must be defined. This will give important background information to help understand why politicians may want to use Twitter to jumpstart or participate in conversations about the campaign that are occurring in the mainstream media. Priming is related to controlling the salience of issues and how important issues are judged. According to Benoit, priming “concerns a voter’s expectations, which can influence whether they attend to this information and how what he or she learns is perceived, construed, or interpreted,” (Benoit 207). In terms of elections, priming will determine on what citizens judge politicians. This judgment is based on their stances on the more salient issues. For example, since in the 2010 midterm elections healthcare was important, priming would cause people to judge Senator Reid based on whether or not they agree with the healthcare measures he fought to enact. Agenda setting is extremely important to news formation and has been influenced by the Internet. Agenda setting is when what gets covered aligns with what citizens consider important. It says that the media influences what issues are salient, or top of mind, for most people. The term agenda setting came from McCombs and Shaw’s 1972 study where they found a positive correlation between perceived importance of a topic and number of mentions of that issue topic in the media (McCombs and Shaw 176-187). In terms of elections, agenda setting influences what issues people think are important for the campaign and this primes what issues a candidate is evaluated in terms of. For example, in 2010 issues included healthcare reform, the stimulus plan, and the economy. Understanding priming is important to this study because Twitter is a place where candidates can release information about issues. If an issue that the media chooses to broadcast
  • 11.   11   heavily is negative for a candidate, they may use Twitter to try and divert the way citizens view them on that issue. They may also use Twitter to try and divert attention from one issue to another or to further enhance issue discussion as a way to keep the issue at the forefront of discussion. If the issue is negative for their opponent, a candidate may use Twitter to try and further control how people judge their opponent on that issue through the use of linking articles and releasing news. For example, during the Nevada Senate race, getting the Hispanic vote and immigration policies were actively discussed through the mainstream media. Sharron Angle had a few major campaign gaffes in reference to Hispanic voters, which Harry Reid could tweet about to try and prime immigration as the issue on which Angle was judged by voters as well as frame the issue. Candidates may attempt to use Twitter and other social media sites to influence agenda setting. Candidates can post about whatever they want via social media in an effort to control campaign discussions. They can then hope that the topics they discuss on social media are picked up by the mainstream media. Understanding agenda setting will help understand why campaigns might want to use Twitter to try and set the agenda of the campaign. Also, knowing whether Twitter can jumpstart discussion could help to show if there is a way for them to successfully set the agenda via social media. Lastly, Zaller’s theory of elite discourse in opinion formation is also relevant for trying to figure out the role Twitter is playing in elections. According to this theory, in large societies such as the United States, people depend on “unseen others” to get information that leads them to opinion formation. These “unseen others”, include political elites such as politicians, government officials, policy experts, activists. They also include family and friends (Zaller 288-289). This theory is relevant because if people are getting information from Twitter, it most likely comes
  • 12.   12   from one of these groups of people. Because of the retweet function, elites, such as politicians, can share their messages to a larger audience than simply their followers. Therefore, even if some people only follow their family and friends, they can indirectly receive campaign messages and help spread them. III. The Rise of the Internet The rise of the Internet has played an important role in political campaigning. For the last decade, new Internet technologies have allowed its role to continue expanding and changing. Similar to other forms of political engagement, those who use the Internet for politics are self- selected. However, this self-selected group is quite large. Pew reported that 2008 “marks the first time the Pew Internet & American Life Project has found that more than half the voting-age population used the Internet to connect to the political process during an election cycle,” (Smith 2). During the 2010 election cycle, it found that this number stayed at about the same percent and represented 73 percent of adult Internet users. (Smith the Internet and Campaign 2010). Since use of the Internet to gather campaign information has increased so rapidly, it is necessary to study how Internet use has changed during the last five election cycles. This will create background data for the 2010 election cycle, allowing the use of Twitter to be effectively analyzed. i. The Year 2000—“The Year of the Internet” Although the World Wide Web was invented in 1989 and went public in 1996, the 2000 election cycle marks the first time in which candidates effectively used the Internet to benefit their campaigns (Lange 108). In their book Campaigning Online, Bruce Bimber and Richard Davis point out that the election of 2000 was not merely unique because of the way it was decided, but also because the 2000 elections represented a “leap forward by candidates in the
  • 13.   13   degree of effort, money, and innovation dedicated to the Internet,” (Bimber and Davis 1). Specifically, for the first time in 2000 use of candidate Web sites and email lists by Presidential candidates became the norm. A study by S.J. Farnsworth and D. Owen found that about a third of the voting-age adult population used the Internet to get campaign news (Farnsworth 420). A Juniper Communications study found that that percentage almost doubled when limited to adult online users (Farnsworth 421). According to Bimber and Davis, in 2000, the most common uses of the Internet by political campaigns were as a means of direct communication with voters, fundraising, voter activism, and voter mobilization (Bimber and Davis 48). In fact, the year 2000 was the first major election cycle in which candidates could fundraise online (Bimber and Davis 38). This was imperative during both the primary season, and the election season. Senator John McCain understood the potential advantages of the Internet and raised over $5.6 million online throughout the Presidential primary season and garnered 18,000 volunteers in California and Michigan alone (Farnsworth 416). These large numbers suggest that the Internet was able to activate voter participation in elections. A key component of Internet use in 2000 was the demographics of people who used it. The Internet played a much larger role in the election for citizens who directly sought out campaign information. More politically active citizens, those who frequently engage in politics, were more likely to visit campaign Web sites and news sites online, thus offering early support for the reinforcement theory (Bimber and Davis 108). In the case of Republicans running for President, partisan use of the Internet saw substantial increases from 1996. “The average percentage of supporters who signed up online spiked by 318%,” and the percentage of Republicans who had been emailed by one of the candidates increased by over 1000 percentage
  • 14.   14   points a study by Chris Hull reported (Williams 61-62). These findings support the reinforcement theory since those reached the most were already supporters. Lastly, Farnsworth and Owens found that nearly 70 percent of respondents who used the Internet to find out about campaigns did so to search for issue-based information (Farnsworth and Owens 422). This is similar to Bimber and Davis’s finding that searching for issue-based information was the second most frequent use for visiting campaign Web sites (Bimber and Davis 115). Both studies found that education, strength of ideology, and age influenced the amount people used the Internet to find out campaign information. This data suggests that the Internet and campaign Web sites may be used by candidates to prime voters on how they are judged on salient issues. It gives candidates a chance to weigh in on the discussion of the issues. ii.   The  Year  2002—Blogging’s  Watchdog  Role   In 2002, campaigns made no major strides in the use of the Internet for politics, but citizens did. The use of blogs and the emergence of citizen journalists were important advances that arose in the midterm election cycle. The 2002 election cycle marked the first time when a politician suffered severe consequences after blogs picked up a controversial statement that the mainstream media had ignored. Specifically, Senate Majority leader Trent Lott eventually had to resign from his position after his comment at Senator Strom Thurmond’s birthday dinner expressing his belief that if Senator Thurmond had won the Presidency in 1948, "we wouldn't of had all these problems," (Glaser). Since Senator Thurmond ran on a segregation list platform, this comment was interpreted to mean that Senator Lott supported segregation and was racist.1 This incident demonstrated the influence blogs could have on the news. The mainstream                                                                                                                 1  The bloggers accredited were John Marshall who posted a story the next day on talkingpointsmemo.com, along with Andrew Sullivan and David Frum (Glaser).  
  • 15.   15   media no longer had full control over what stories get covered and what stories are ignored, nor control over what frame dominated coverage. Although no major advances by campaigns occurred in 2002, it was the first midterm elections where a majority of candidates had some sort of Web site. In fact, a study done by Danielle Endres and Barbara Warnick found that 63 percent of House candidates had campaign Web sites, compared with about 40 percent in the 1998 midterm election (Endres 322). Similar to the 2000 Presidential election, in 2002, Web sites were primarily used to connect voters to the candidate and the campaign rather than to each other. Specifically, a study by W. Lance Bennett and Michael Xenos found that the most common items found on Web sites were candidate biographies and a page for news clips and press releases (Xenos and Bennett 457). They also found that campaign sites were posting contact information, included a place to submit questions, a place to submit donations, a place to sign up to volunteer, and a link to be added to the email lists (Xenos and Bennett 457). In short, candidates were using their Web sites in attempt to impact what was considered newsworthy and activating supporters. Even though Web sites were being used to communicate with voters, both studies found basic errors in Web strategy. Bennett and Xenos concluded that although about a quarter of 18- 29 year olds received the bulk of their election information online, candidates failed to market to their age group via their Web sites (Xenos and Bennett 460). This was based on their content analysis of the information available on 1200 candidate Web sites during 2002. Endres and Warnick find that “although political campaigns see the necessity of Web-based campaigning, campaign Web sites have yet to realize their full potential as a medium to improve communication between candidates and Web users and to influence voters,” (Endres 323). Thus,
  • 16.   16   candidates failed to incorporate the mobilization theory into their Web strategy and use the Web as a way to engage young voters who are a large part of the online community. ii. The Year 2004—Howard Dean’s Online Miracle Monumental leaps for Internet usage in campaigns occurred in 2004. The new and innovative ways it was used went well beyond creating an informational campaign Web site. Howard Dean’s campaign manager, Joe Trippi, and his campaign staff are largely responsible for the tremendous growth in Internet technology in the 2004 Presidential campaign. Not only was Dean able to generate a vast number of followers and fundraise online, but also he was able to create an online network that allowed supporters to meet up and created the first candidate blog (Trippi 89). Although Howard Dean lost the Democratic Presidential bid, his use of the Internet was more advanced than any other primary candidate. When the campaign started, Dean not only had little money, but little name recognition and support. The Dean campaign partnered with the Web site meetup.com to create a place online where supporters could communicate with each other and plan rallies or meetings. Due to this arrangement, a network for 190,000 Dean supporters who could interact was created (Trippi 84). Unlike with prior studies on campaign Web sites, this network’s success supports the viability of the mobilization theory because a- typical voters became involved. In fact, according to Pew, “About four in ten (42%) Dean supporters were participating in their first political campaign,” (Keeter, Funk, and Kennedy 1). Dean’s Web site also included the first candidate blog as well as a “TV channel” HowardDean.TV (Trippi 109). The Dean campaign understood the power of creating a grassroots organization and the way the Internet could accomplish this goal. It used the grassroots organization it created online
  • 17.   17   for campaign events, as well as for online fundraising. When Dean formally announced his candidacy, the campaign was able to raise over two million dollars online in a week, the average donation being about 100 dollars (Trippi 131). During this week, the Web site posted a fundraising bat that supporters mimicked and posted elsewhere online in attempts to help the campaign reach its fundraising goal. The Dean campaign surpassed their goal for the quarter, and even the fundraising totals that quarter of Kerry and Edwards and they then got a lot of media coverage for their success (Trippi 134). The Dean campaign was the first to use these techniques, but other campaigns followed in its footsteps. A study conducted by Steven M. Schneider and Kirsten A. Foot found that in 2004, all of the Presidential candidate Web sites observed “had established a practice of involving” (i.e. using it for campaign asks) compared with only five of eight sites that were studied from 2000. They also found that using Web sites for mobilization had increased significantly (Williams 28-29). Both President George W. Bush and Democratic nominee John Kerry also had blogs on their Web sites (Williams 133). iii. The Year 2006—Gaffes on YouTube The year 2006 marked another crucial step for the Internet and politics because it was the first campaign cycle where YouTube and MySpace were available. The emergence of these social media sites was both positive and negative for campaigns. In some cases, exemplified by Representative George Allen, social media enabled voters to see campaign gaffes that may have not been picked up by the mainstream press. In other cases, such as with Ned Lamont’s Senate campaign in Connecticut, the free communication tool was an extremely important means for a money-strapped campaign to communicate with voters (Cohen). These benefits and weaknesses
  • 18.   18   of social media made the Internet’s use in campaign strategies more vital to campaigns with little funding and close, highly broadcasted races. Although many candidates feared YouTube in 2006, campaign strategists saw its benefits. For example, The Montana Democratic Party taped Senator Conrad Burns of Montana closing his eyes during a hearing for the Montana farm bill (Jalonick). The video was quickly launched on YouTube and even got earned media in newspapers such as USA Today and the Wall Street Journal. The USA Today article from August 22, 2006 quoted Phil Singer, the spokesman for the Democratic Senatorial Campaign Committee, saying that campaigns "have always used video cameras, but the difference this cycle is that YouTube allows information to get uploaded and out to the public a lot faster than it has in the past," (Jalonick). This example shows that campaigns could use YouTube to influence agenda setting in the mainstream media. The same strategy was famously employed against Senator Allen, who was caught calling a young Indian man “Macaca” during a campaign rally. Although the man worked for his opponent, Senator Jim Webb, who was following George Allen and videotaping him at every campaign event, all that mattered was that Alan was caught on camera making a racial slur. After this, these gotcha moments caught on tape became known as “Macaca moments,” (Karpf 144). The year 2006 marked a further expansion of Internet use as the main source of campaign information for Americans. In fact, Pew found that twice as many people reported that they used the Internet as their main source of election information as compared with the 2002 midterm election. Overall, Pew used the name “campaign Internet users” to represent over 60 million people who received and/or shared information about the election online (Election 2006 Online 3). This meant that priming, framing, and agenda setting of the campaigns could be impacted online.
  • 19.   19   Pertinent to this study, the year 2006 marked a drastic increase in the use of the Internet to learn about congressional races. According to Pew, “the number of Americans getting information about Senate races rose fivefold since 2002, and at least doubled in the case of House races, local contests and gubernatorial campaigns,” (Election 2006 Online 20). This may have been influenced by the increased availability of campaign information online and it was becoming more of a common practice to seek campaign information online. With the emergence of social media and the expansion of Internet use for midterm elections, 2006 is an important year to compare with 2010. iv. The Year 2008—“Yes We Can” The Obama campaign in 2008 expanded Internet use drastically from the primary season onward. According to a Pew study from April 2008, during the primary season 40 percent of Americans adults were using the Internet to get campaign information. This indicates a 2.5 times expansion in adult Internet use from the primary season in 2000. Pew also found that the use of social media was gaining in popularity. At this time, 35 percent of adults had watched political videos online and 10 percent had used either MySpace or Facebook to receive campaign information by following candidates or being their “friends” (The Internet and the 2008 election 3). By 2008, the use of YouTube had exploded. A study by Williams and Gulati found that there was a significant increase in Senate campaigns with a YouTube channel between 2006 and 2008. The percent of campaigns with this jumped from about 10 percent of total campaigns to 72.7 percent of Republicans and 70.6 percent of Democrats (Gulati 99). These channels were successfully used to connect campaigns to voters. Pew found that 35 percent of voting-age Americans watched political videos online during the 2008 cycle (Rainie ii).
  • 20.   20   However, campaigns were no longer the only ones using YouTube to post videos. Rather, citizens who felt a stake in the election created independent campaign videos. The two most famous were “Obamagirl”, which featured an attractive female singing about her crush on Obama and Will.i.am’s “Yes We Can” video, which featured a number of famous celebrities singing along with an Obama speech. According to a study done by Dr. Emily Metzgar and Albert Maruggi, “Obamagirl” was viewed 11.6 million times and “Yes We Can” was viewed 13 million times during the campaign (Metzgar and Maruggi 153). Spending money on television airtime no longer was the only way to showcase campaign videos. These two videos exemplify the viral effect of the Internet and highlight that this effect can be utilized by anyone interested in the campaign. YouTube has created a way for citizens to become included in the campaign process by allowing them to make and legally post their own videos for anyone to see. When citizens feel as though they can impact the election, they become more engaged, thus a potential means for increasing voter efficacy is created though social media outlets. Another major advancement by Obama was improving Dean’s meetup.com platform from 2004 through the creation of Obama’s own online network of followers. His online grassroots network was called Obama for America, later known as Organizing for America (OFA). OFA used a “bottom up” campaign approach to get potential supporters involved in the campaign. Like meetup.com, it connected supporters to each other via the formation of local groups, blogging, and campaign events (Organizing for America). OFA targeted minority groups and young voters who are not typically active in elections. It raised an unprecedented amount of money online, totaling a half a billion dollars with small donations (Vargas). Organizing for America thus adhered to the institutional approach focused on involving supporters who may have not had the financial means to become involved before the Internet age. It gave people who
  • 21.   21   did not have the financial means to donate large sums of money to the campaign other ways to get involved, via small donations, different degrees of volunteering, and ways to expand Obama’s reach online. Like via social media, it was a way for people to feel as though they could partake in spreading Obama’s message. 2008 also marked the first year in which Twitter was used by campaigns. Specifically, President Obama used Twitter as dimension of his social media strategy. According to Communicator-in-chief, a book about how President Obama used new media technology to win the election, Fradric I. Solop states that “Obama used Twitter in tandem with other social media tools to fine-tune his message to a technology-savvy population,” (Allen Hendricks 40). Although only 3.5 million people had Twitter accounts in November 2008, Obama managed to obtain 118,000 followers. According to Solop’s study, Obama mainly used Twitter to announce his location and to get people to go to www.barackobama.com. (Allen Hendricks 41). Since Obama used Twitter as a means to drive people back to older campaign technologies, it will be interesting to see if the Nevada candidates for Senate are also using Twitter mainly as a way to communicate their location to voters and gain Web site traffic. IV. What Experts are saying about Twitter and Campaigns As mentioned previously, there is a very small collection of studies on Twitter use by campaigns. Despite the lack of systematically tested data, experts in political communications are actively discussing Twitter and how campaigns should use it. According to Steve Pearson and Ford O’Connell of Campaigns and Elections, “a campaign's tweets need to reflect the candidate's voice.” In other words, they believe tweets will be more effective if they seem as though they are coming from the candidate personally. This is because Twitter’s main function is to give individuals a place to share their thoughts.
  • 22.   22   Experts believe that social media should be incorporated into other campaign strategies. In an interview with Mindy Finn, co-partner at the political consulting firm Engage DC, she said that “the most effective use of Twitter for close Senate races relies on frequent use that mirrors the most important aspects of the campaign, what I call the three "M"s: message, money and mobilization,” (Finn). In other words, Twitter should be used as another platform to advance campaign messages, increase fundraising totals, and mobilize potential supporters in order to be most effective. Finn believes that Twitter is more often used to promote candidates in a positive manner than attack opponents (Finn). This is because being positive helps raise money and mobilize supporters. Thus, her thinking supports the reinforcement theory. Sticking to campaign messages and using campaign asks is one way to follow this advice. This study will attempt to look into whether campaigns are using Twitter for the “three M’s” by categorizing tweets based on what they are trying to accomplish and by studying how they are being picked up and spread via retweets. V. Current Studies about Twitter Since 2008, a few studies have been conducted to try and determine how candidates, politicians, and the public are using Twitter. Their findings were used to formulate the hypotheses that will be presented in this thesis and to create the proper research design. This study will expand upon techniques used in these previous works and the coding definitions they used to propose an answer to the question of what is the role of Twitter in Senate campaigns for the candidates, the news, and the public. One particularly helpful study was that of Wyth Ruthven, Twitter Trends in Off-Year Elections: Virginia, New Jersey, and Massachusetts. This study analyzed the tweets for a three- month period prior to the 2009 elections for Governor in Virginia and New Jersey and Senator in
  • 23.   23   Massachusetts. Ruthven categorized the tweets on the basis of being calls to action, self- promotion, personal observation, conversation, or passing information. He found that over 50 percent of tweets from the sample of 3,521 tweets observed fell into the category of “self- promotion”, which was defined as “off-message tweets, personal observations not related to the campaign” (Ruthven 2). In Massachusetts, Ruthven found that Senator Scott Brown’s surge in Twitter followers correlated with the surge of donations, interest, and growing popularity in his campaign. He concluded that Twitter can be used as an effective means for calls to action, including fundraising and increasing GOTV efforts. (Ruthven 3-6). These findings suggest that Twitter may be affectively used for self-promotion and helped me formulate the hypothesis that tweets will more often be self-promotional than attacking opposition. Like Finn, Ruthven concluded that candidates should use preserve authenticity in their tweets (Ruthven 23). These findings also helped me decide to code for authenticity, use of external content, and calls to action. Overall, Rythven concluded that having multiple Twitter accounts can be problematic for campaigns. In Nevada during the 2010 Senate race, Senator Reid had two Twitter accounts, one for his position as Majority Leader, and one for his reelection campaign. This research suggests that having the multiple accounts may hurt Reid. Other useful studies have recently been released that rank effectiveness of current elected officials on Twitter. One example is a 2010 Burson-Marsteller study on how House and Senate members are using Twitter to connect with their constituencies. Like Ruthven, they found that more authentic Twitter accounts are more successful than accounts that did not tweet in a personal manner (Burson-Marsteller 23).
  • 24.   24   Another example, the 2010 L2 Senate Digital IQ test by George Washington University Dean Doug Guthrie and NYU professor Scott Galloway, hypothesized that “digital competence provides an opportunity for senators to authentically engage and mobilize voters and constituents,” (Galloway 4). They created a metric to measure digital competence of Senators and ranked all Senators in office. This metric was based on presence number of likes, and growth on Facebook; presence, followers, velocity of Tweets, and follower growth on Twitter; Presence, number of upload,s and number of channel/upload views on YouTube; Velocity of mentions on Blogs and other 2.0 sites and sentiment of blogs; and site traffic to candidate Web sites. According to their rankings, Senator Reid was categorized as a genius and ranked tied in fourth place with Al Franken and John Cornyn (Galloway 5). Reid’s categorization as one of the most effective users of social media supports the choice of analyzing Reid’s Twitter use to gain insight into how Senate candidates are using the site. Both of these works concluded that Republican Senators were using the Internet more effectively than Democrats. The Burson-Marsteller study concluded, “while there are fewer Republicans in Congress, a larger percentage of them are using Twitter to communicate,” (Burson-Marsteller 9). The L2 Senate Digital IQ study added that Senators up for reelection were utilizing social media more effectively than those not up for reelection, and that November challengers in close 2010 Senate races were often using social media more effectively than the incumbents. One example was Carly Fiorina in California who had over 11 times as many Twitter followers as Senator Barbara Boxer (Galloway 9). Despite these numbers, Boxer won the election. Although Facebook and others suggest that social media presence can be a reliable indicator of who will win an election, the California and Nevada Senate races show that the
  • 25.   25   overall number of followers may not be the best indicator of who is successfully using social media to increase their chances of winning (Coyle). On November 2, Sharron Angle had over 2,000 more Twitter followers than Harry Reid’s campaign account, yet she lost the election. According to PBS.org, from September to October 21, Angle increased her Twitter following by 39.5%. Reid’s only increased by 13.8% (Leher). If gaining followers toward the end of an election was the best way to use Twitter, than her increased momentum should have meant that she would win. Thus, this reinforces the need for this study that will look more in depth into Twitter usage and give insight into current strategies. It will seek to figure out how Twitter can best be utilized by campaigns. VI. Hypotheses The main research question I propose is how did candidates in the 2010 Nevada Senate campaign use Twitter? To figure out the answer to this question, it has been divided into three areas of sub-questions. For each area, hypotheses have been formulated. The first area is the interaction between Twitter and the mainstream media. My first research question is was Twitter used to jumpstart the discussion of campaign issues, or continue existing discussions from media coverage? Based on the literature review presented above, I hypothesize that Twitter is being used as a means to continue existing discussions. I came to this conclusion mainly because history shows that it is rare for stories to break on social media or blogs and then get picked up by the mainstream media. Instead, usually the mainstream media sets the agenda and people turn to the Internet to find out more information about the issues that are salient to the campaign. Therefore, I think that stories will show up in both Twitter and the mainstream media around the same time, but will get more coverage on Twitter. Twitter is not able to control the agenda of the mainstream media. Specifically, I believe that when there is
  • 26.   26   negative news coverage of one candidate, the opposing candidate will use Twitter to lengthen the discussion of the issue. I have come to these conclusions based on my understanding of priming, agenda setting, and understanding history of the web. Since the web has been used previously for candidates to share their points of view, I believe Twitter is the same. The second set of questions is related to how Twitter is being used by the campaigns. Are Senate candidates using Twitter as a means to attack their opponent or to relay positive information about themselves? And if attacks are present, are they direct or indirect? Indirect attacks will use links to sources and may reference news articles whereas direct attacks will come from the campaign. Based on the 2009 study’s results, I hypothesize that candidates are using Twitter mainly as an outlet for releasing positive information about themselves. However, I believe that I will find that when attacks are made using Twitter, they are presented in an indirect fashion, via the usage of links to external sites and news stories. This is based on my understanding of the current state of the Internet and social media. Candidate Web sites, Candidate Facebook pages, blogs and even online news stories often use links to the original sources, so I expect Twitter use to do the same. Also, political attack ads often use someone other than the candidate running for office to make the attack. It is not common practice in campaigning to have the candidate directly attack their opponent via any outlet and the literature review does not present reasons to think it will be different with Twitter. Another aspect of this question refers to the types of information campaigns tweet about. Are candidates using Twitter more frequently to inform people about campaign events, post news, discuss issues, reinforce campaign messages, or for campaign asks (i.e. donations, GOTV, signing up for events)? Does this change over the course of a campaign? I hypothesize that the most common thing candidates are tweeting about are ways for voters to get involved. This
  • 27.   27   includes GOTV efforts, as well as donations, event information, and random campaign asks. I came to this hypothesis based on what candidates have used other online outlets for in the past, such as was done by Barack Obama and Howard Dean who fundraised and created online, grassroots networks. I also hypothesize that the main type of information campaigns put out on Twitter changes throughout the campaigns. I think at the beginning campaigns use Twitter to discuss issues whereas close to Election Day they use Twitter for GOTV. This is because overall campaign strategy shifts to a GOTV focus close to Election Day. The third area of questions is related to the viral nature of Twitter. To examine this concept, I will look at retweets and hashtags. Specific questions include whether inclusive language makes people more likely to retweet the candidates? Does using consistent hashtags lead to more retweets? I hypothesize that tweets with inclusive language are more likely to be retweeted. This is because personalized messages are more appealing because they make people feel as though they can influence the results and are a part of the political process. This relates to the concept of external efficacy. I think that using consistent hashtags will also lead to more retweets because it makes the tweets easier to find. I chose to look at retweets because number of followers proved to not be sufficient to show who is reaching the broadest audience on Twitter and activating potential supporters. I decided retweets was better because depending on who retweets the candidates, their messages potentially can reach a much larger and more diversified audience. Since the instrumental approach suggests that as cost of participation decreases, people will become more apt to participate, and retweeting is free, I thought it was a good way to look at audience engagement.
  • 28.   28   VII. Research Design This study used a content analysis of all of the tweets posted by Sharron Angle on her campaign Twitter account, @SharronAngle, and Senator Reid on his campaign Twitter account, @HarryReid, from October 1 2010 through Election Day (November 2, 2010) to understand how Twitter was used in this campaign. This includes 127 tweets from Sharron Angle and 458 tweets from Harry Reid.2 It will also compile data on the retweeting of these tweets. Lastly, it includes a content analysis of news stories related to the campaign in the Las Vegas Review-Journal during this time frame. Analyzing all of this data will pave the way for an analysis of how candidate Twitter usage relates to the news coverage of the campaign, how it relates to campaign strategies, and how it relate to the viral nature of Twitter. The tweets were numbered and their date and time was recorded. They were then be coded for usage of common hashtags. I chose to code for the three most frequently used hashtags per candidate. For both, this includes #nvsen. For Reid, the other two include #p2, which means progressives 2.0, and #GOTV. For Angle, the other two include #tcot, which means top conservatives on Twitter, and #dumpreid. I coded for these hashtags since hashtags are used to categorize tweets and make them more accessible to people on Twitter. I compared the hashtag use to the amount of retweets. Next, I coded based on the messages the tweets attempt to relay. Since it is possible to relay more than one type of message per tweet, this is not mutually exclusive. Categories include                                                                                                                 2  I thought a census of all the tweets from the last 33 days of the election would allow me to have a better understanding of how campaigns were using Twitter. I came to this conclusion after using the Web site www.tweetstats.com to get data on total tweets from each candidate. The data provided showed a sharp increase in twitter usage as Election Day neared. Therefore, I thought analyzing the tweets from the time period where candidates were tweeting the most would be more useful. It also allows me to see if specific uses of Twitter change as Election Day nears.  
  • 29.   29   Get out the vote (GOTV), other calls to action, campaign events, news stories, and issue mention. GOTV messages include tweets that ask people to vote, ask people to get out the vote, or relay information about voting. Included in this category are candidate retweets of Twitter users. Examples of voting information include information about polling locations and when people can vote. Other calls to action focus on something other than getting people to simply vote for the candidate or push others to do the same. Examples include fundraising asks, event sign-ups, and email sign-ups. Tweets asking people to show their support on Twitter, such as through changing your Twitter picture to a campaign badge, are also included in this category. Campaign events are tweets that relay information about campaign events, including dates, time, locations, topics discussed, type of events, and guest speakers. If images from events are posted, they will be counted in this category. News story messages are tweets that reference campaign news stories. They can include links to actual stories or just reference the story. References to polling data released will be included in this category. For example, a tweet that says ABC exit poll shows Angle up by 2.7 would be counted here. Issue mention messages include tweets about a substantive issue. For example, mentions of healthcare, the economy, immigration, jobs, and education initiatives would be included. Tweets that do not give specific issues, but relay campaign promises such as “will fight for Nevada” will also be counted here. I specifically coded for the national issues of healthcare, immigration, the economy, government spending, jobs, big government, and education.3 I                                                                                                                 3  Please note: I combined jobs with economy and government spending with big government when analyzing the data.  
  • 30.   30   also coded for campaign-specific issues that arose over the course of the campaign. This includes Angle’s ignoring of the mainstream media, the Hispanic vote, dairy farming, nuclear waste in Yucca, and Ashjian recording his conversation with Angle. See the appendix A for definitions and specifics about how these were coded. Next, tweets were coded based on their “voice”. Categories looked at include personal observation, retweeted, and inclusive language. Personal observation tweets are tweets that sound genuine and like they are coming from the candidate. For example, a tweet like “great turnout at the event” would count as personal observation. Another example could be “thank you for helping spread the message.” User Retweets are tweets from other Twitter users that the candidate reposts, with or without adding their own message. Inclusive language includes tweets that use language to include the reader of the tweets. The words that count in this category are us, we, you yours, and our. Next, I marked whether or not links are present. Analyzing the types of links will help answer the overarching question about the role of Twitter and provide insight into answers for my sub-questions. Types of links were separated into whether they link to an outside source, a site that is not owned or managed by the campaign, link to the campaign, via a campaign Web site, link to party, via a party-owned site, or video link, via YouTube. I also marked whether or not there is a photo link. Lastly, I looked at the general purpose of the tweet. Tweets will be categorized as either self-promotion or attack and must be put into the category they fit best into. Self-promotion tweets include all tweets that promote the campaign and put the candidate in a positive light. Tweets that are categorized as GOTV, other calls to action, and campaign events were included
  • 31.   31   in this category. Attack tweets attack the opponent in some manner. Attack tweets could be found in issue mention tweets. One way that I looked for attacks is by coding separately for opponent mention, tweets that mention the opponent by name in the 140 characters. If links are present, the attack will be considered indirect, whereas if they are not mentioned, the attack will be considered direct. The number of times each tweet is retweeted was then recorded. This helped me formulate conclusions about what types of tweets and messages used were more successful in getting followers involved and reaching a wider audience. Specifically, this will help me determine if the use of inclusive language leads to more retweets.4 Campaign news stories in the print version of the Las Vegas Review-Journal, the largest newspaper in Nevada, during this time frame was also coded. The stories were selected through a Lexis Nexus search of stories in the Las Vegas Review-Journal that mentioned “Nevada Senate race”, “Reid”, or “Angle”. In total, this includes 225 articles. However, the only stories in the search that were coded were those that reference the Nevada Senate race in at least two paragraphs. Therefore, any stories that are not focused on this specific race were discarded. In total, 131 stories were coded. The stories were categorized based on their type of story, the issues mentioned, and also whether articles as a whole are positive, negative, or neutral toward each candidate. I coded for both type of story and issues so that I could better compare the types of stories and the issues they mention with candidate tweets at that time.                                                                                                                 4  It is important to note some issues that arise with using the Twitter function to look at retweets. The retweets recorded by Twitter only occur when people hit the retweet button. If people copied the message and added their own words it would not be known.Therefore, there is no way to know how many people did not hit this button, but still retweeted the message by manually typing the candidate’s message into their own tweets.  
  • 32.   32   Types of story include process, horse race coverage, event stories, personal information, endorsements, and issue pieces. Process includes information about the electorate, predicted turnout, important dates, and campaign finance stories. Horse race coverage includes stories about who is up or down in the polls. Event stories are stories about specific campaign events, such as the Reid rally that Michelle Obama went to. Personal information stories are stories about the candidates. These stories may allude to the morals of the candidates, personality traits, and information about their lives. Endorsements are stories about whom the paper or other figures endorse. Issue pieces are stories about specific campaign issues, like healthcare. Releases of news stories compared to releases of stories on Twitter will then be compared to determine whether stories most often appear first on Twitter or in the mainstream press. Next, I coded for the issues. I used the same set of national and local issues that I used to code the Tweets listed above so I can compare issue mention on Twitter and in articles. I coded for whether the article is positive, neutral, or negative toward Angle and toward Reid. To determine this, I looked at the words used to describe the candidates and their positions. For example, if Reid is called a socialist or his policies are referred to as reckless, I would code the story as negative. If Angle were described as a Washington outsider, I would code the story as positive toward her. I chose this specific research design for a variety of reasons. Since I wanted to figure out how candidates were using Twitter, a content analysis would allow me to do a much-needed in- depth analysis of a specific campaign. I also compared Twitter use to the mainstream media environment. Because there is so little known about Twitter use in campaigns, looking at just one race in depth allows me to look at more variables and have a deeper analysis than would have been possible if I was looking at a variety of races more generally.
  • 33.   33   Nevada was a good state for examining how Twitter was used in the 2010 campaign for a variety of reasons. First, it was labeled a toss-up from April through Election Day. Second, Sharron Angle was a Tea Party-endorsed candidate who represented the disgruntled, conservative wing the of electorate, while Harry Reid was the Senate Majority Leader, largely responsible for decisions made by the Democratic Congress. Harry Reid represented a Washington insider, while Angle was a fresh, new figure. Thus, this gave added media attention to the race. I also thought this race would be interesting to study because Angle ignored the mainstream media throughout the campaign. VIII. Limitations of this Study Using a content analysis does have some limitations that must be addressed. Through a content analysis, there is no way of knowing if the candidates themselves are tweeting or if someone on their campaign staff is tweeting for them. Therefore, there is no way to tell how genuine the tweets are. This is relevant since I am coding for inclusive language and personal observation. I also do not know what campaigns did with the information they had on their retweeters or Twitter followers or what their strategies were. This limitation could be addressed by talking to the campaigns directly and asking them who did the tweets and what goals the campaigns had for Twitter use. There are also limits in just studying the Nevada Senate race. Since this race was so unique, it also presents limitations for my study. It may not be reflective of the use of Twitter for other tight Senate races that did not have the same factors at play, thus limiting the scope of conclusions I can make. Also, since Twitter is used more with younger demographics and Nevada has a lot of retirees, the audience Twitter was able to reach may have been smaller than
  • 34.   34   in other states. Lastly, a majority of the research discussed in my literature review is about presidential races, not congressional races. Therefore, uses may differ. Another limitation is that I only looked at one form of mainstream media. The print version of the Las Vegas Review-Journal, which only comes out once a day, makes it difficult to know if Twitter jumpstarts the discussion, or if the discussion is already going on via online news outlets such as blogs and stories that are just posted online. However, it is extremely hard to compile all of the online sources that discuss the campaign and determine whether or not those sources have a following. The readership of the newspaper is clear and the articles are well documented and accessible. VIII. Findings The findings of this study deviated somewhat from what was expected. In regards to some of the research questions, data provided substantial evidence in support of the hypothesis. In regards to other questions, the sample size was too small to obtain high confidence level from the results of this study. Finally, in some other cases, the data proved the hypotheses were incorrect. To check for reliability, a second person also coded 11 percent of the tweets and news articles used in this study. The coding of tweet variables between both coders varied between 83% and 100%. It should be noted that the news stories and issue mentions had an 83 percent agreement while every other variable had a 94% or higher agreement. For the news articles, the coding of the issues varied between 77% and 100%. The three categories that only had 77% agreement were Ashjian’s recorded conversation with Angle, jobs, and government spending. This may have been because jobs were so similar to economy and big government was so similar to government spending. For the exact breakdown, see Appendix C.
  • 35.   35   A. Findings on the Relationship Between Twitter and the Mainstream Media In reference to the first subset of questions, the interaction between Twitter and the mainstream media, there was not enough data to provide a high level of confidence in the findings from this study. This was mainly because tweets were infrequently about campaign issues for long periods of time. Overall, 201 of the total tweets (34 percent) referenced an issue of some sort. However, the issues mentioned varied across eleven different topics. The topic that got the most mentions was the economy with 44 total tweets. Within the timeframe of 33 days, this is not enough data to be able to make definitive conclusions about the timing on Twitter compared to the newspaper. Although the dataset’s size was limiting, some patterns were evident when comparing the way issues were covered in the Las Vegas Review-Journal and by the candidates on Twitter. For overarching campaign issues that defined the 2010 Midterm election nationwide such as the economy, healthcare, and big government, coverage in Twitter and the newspaper remained fairly consistent over the observed time period. In all three cases, the issues were mentioned more frequently on Twitter than in the Las Vegas Review-Journal. However, no clear pattern emerged showing mention in one source preceded mention in the other source. At best, this data supports the hypothesis that candidates use Twitter to continue issue discussions in order to weigh into the discussion and try and get their point of view across. For example, both candidates used Twitter to talk about the economy in about 7 percent of their tweets. See figure one for a graph of economic mentions in both sources.
  • 36.   36   5 For localized issues, a different pattern emerged than was seen for national issues. Most of these issues were covered more frequently in the Las Vegas Review-Journal than they were by Harry Reid and Sharron Angle on Twitter. This was the clearly the case for the Hispanic vote, immigration, nuclear waste in Yucca and Ashjian’s recorded conversation with Angle. Thus, Twitter was not effectively used to lengthen discussion of these issues. As an example, see figure 2, which shows the mentions of Angle ignoring the mainstream media in tweets and articles. Lastly, an interesting pattern emerged with issues that are bad for the opposition. As I predicted, in these instances, Twitter was used by the opposing candidate to continue issue discussion. For example, Reid tweeted about Ashjian leaking his conversation to the press six times, tweeted about how Angle ignores the press twenty-one times, and mentioned the Hispanic vote seventeen times, while Angle never mentioned any of these issues on Twitter. This study still could not conclusively say what came first. However, figure two below indicates that Angle ignoring the mainstream media was being discussed in the month of October the Las Vegas Review-Journal before by Harry Reid on Twitter.                                                                                                                 5  see appendix B for the rest of the issue graphs  
  • 37.   37   B. Findings on How Campaigns are Using Twitter In regards to the second set of questions about how the campaigns used Twitter, the data showed that tweets were slightly more likely to be promotional than attack. Combined, 52 percent of the total tweets were promotional while 48 percent of tweets were attack tweets. However, when differentiated by candidate, a clear difference in purpose arose. Angle’s tweets were used as a means of promotion 66 percent of the time while only 49 percent of Reid’s tweets were used as a means of promotion. In fact, slightly more of Reid’s tweets were attack tweets than promotional tweets. Since I thought attacks would be indirect via the mention news stories and/or use links, I also compared whether or not news stories were present in them. I found that 52.5 percent of attack tweets referenced news stories while only 40.8 percent of promotional tweets referenced news stories. Thus, these findings support my hypothesis that attacks via Twitter are more likely to be indirect since a majority of the time they reference a news story.
  • 38.   38   To better answer this question, I also looked at whether links were present in attacks. I predicted that attack tweets would use links to outside sources as a way to curb negativity from being attributed to the candidate. If they were directly linking to their site or not linking at all, the attack would more directly come from the candidate, making him/her seem more negative. I found that over three-fourths of attack tweets included links to either the campaign site or an outside site. Interestingly, the vast majority of these links were to the campaign site, which appears to show that attacks are direct. I found that 59 percent of all attack tweets went directly to a campaign site while only 19 percent went to an outside source. While this would seem to undermine the claim that attacks are indirect, this can be partially attributed to the fact that Harry Reid had a news section on his Web site that displayed articles from outside sources. This section included positive coverage about Reid and negative coverage about Angle. This page had been counted as links to the campaign Web site. Since I had not specifically coded the links to a news page on a campaign site, I cross- referenced the tweets that included news stories against the tweets that linked to the campaigns’ web sites and found that there were 171 of such tweets. Then, I compared those tweets with the tweets whose general purpose was to attack and found that 34 percent of the attack tweets included a news reference and linked to the campaign. Therefore, candidates were frequently tweeting news articles that they had posted on their campaign Web site to attack their opponent indirectly and also drive potential voters to their site. This was noteworthy since Obama had used Twitter in 2008 to drive people to his Web site, therefore suggesting a similar strategy. To figure out how candidates were using Twitter, I wanted to see how they were integrating their Twitter strategy with other campaign strategies. Links to their campaign site was one obvious way to do this. In total, the campaigns linked to their own site 51 percent of the
  • 39.   39   time. Angle linked to her site only 25 percent of the time while Reid linked to his site 58 percent of the time. Angle’s infrequent use of linking to her own site differentiates her strategy from Reid and Obama in 2008. Since this did not account for all the campaign related links, I also recorded the number of times tweets linked to a video or to a photo. Both candidates used links to YouTube videos in about 14 percent of their total tweets. The vast majority of these video links were to campaign advertisements posted on the candidate’s YouTube page. Thus, Twitter was also being used to drive people to the candidates’ YouTube pages. Photo links were only used in about five percent of the total tweets. This made sense since the photos linked were taken at campaign events and only fifteen percent of the tweets referenced campaign events. See figure three to understand the total distribution of links. C. Findings on Types of Information Candidate Tweet About
  • 40.   40   In regards to the second set of questions and the types of information campaigns tweet about, this study found that the campaigns most frequently tweeted news stories. See figure four for the complete breakdown. In total, 46 percent of tweets references news stories. The next most frequent use of Twitter was to discuss an issue, which was over a third of the tweets. Therefore, my hypothesis that campaigns were most frequently using Twitter to make calls to action was incorrect. Only about a quarter of total tweets were GOTV efforts or other campaign asks. Although less than a fifth of the tweets had a GOTV message, I was correct that toward the end of the campaign cycle, GOTV efforts were more prevalent. In fact, over sixty percent of GOTV tweets occurred in the last week of the campaign cycle. On Election Day, the GOTV message took over. Out of the 24 Election Day tweets, 21 of them (88 percent) had a GOTV message. This shows that candidates were integrating Twitter into their overall campaign strategy to promote potential supporters to vote. See figure five to understand how GOTV tweets increased drastically in popularity toward the end of the campaign.
  • 41.   41   *Please note: figure shows number of GOTV tweets on a give day In order to fully understand how candidates were using Twitter, I also recorded the number of their tweets that were retweets, the number of tweets that were personal observations, and the use of inclusive language (we, your[s], our[s]). Retweets were recorded to help measure how candidates were engaging with their followers. I found that 12 percent of the total candidate tweets from October 1-November 2 were user retweets (13.5 percent of Reid’s tweets and 8.4 percent of Angle’s tweets). Thus, Angle and Reid were using Twitter to engage with followers by retweeting them. Since all of the current studies on Twitter suggest that using an authentic voice enables politicians to most effectively reach followers, I expected that both Reid and Angle would use an authentic voice. I coded for authenticity based on whether tweets were categorized as personal Figure  5:  GOTV  Tweets  Across  Time  
  • 42.   42   observation or if they used inclusive language, or did both. Surprisingly, only 10 percent of the total tweets were coded as personal observation and only 24 percent used inclusive language. This overall percentage was greatly impacted by Harry Reid, who only used personal observation four percent of the time. This may be because the Reid campaign used an outside firm, Well & Lighthouse LLC, to do all of the campaign tweeting. Instead of making the tweets sound like they were coming from Reid, they chose to make the Twitter account sound like it was coming from the campaign office. Therefore, Reid’s tweets often referenced him in the third person. Sharron Angle on the other hand used personal observation 29 percent of the time. This may have been because Sharon Angle was not accessible to supporters through the mainstream media, and therefore needed to communicate directly with voters through alternative venues, like Twitter. Angle’s tweets never referred to her in the third person. In sum, tweets that failed to be seen as personal observation simply sounded generic for both candidates. D. Findings on Viral Nature The third area of questions relates to the viral nature of Twitter. To examine this concept, I looked at the number times each candidate tweet was retweeted and the presence of hashtags in the tweets. Candidate tweets averaged 14 retweets for the entire data set. However, it is important to note that Angle’s average number of retweets was three times higher than Reid’s average number of retweets. Angle averaged about 30 retweets while Reid averaged about 10 retweets. This may have to do with the frequency each candidate was tweeting. Reid tweeted an average of fourteen times a day while Angle tweeted and average of five times a day. Reid also would post the exact same tweet multiple times in a single day or two-day span. Therefore, his followers may not have been as willing to retweet information since it was repetitive. Another
  • 43.   43   possibility is that people were less willing to retweet Reid’s tweets because they did not sound like they were coming from him. This content analysis suggests that there is a relationship between use of inclusive language and number of retweets. Looking at both candidates, the mean number of retweets is 18 when inclusive language is present and 13 when it is not present. The difference in the mean number of tweets is significant (t=3.82, p<.05). However, this relationship was highly driven by Angle. Each candidate was individually analyzed based on number of retweets and presence of inclusive language. For Reid, the difference between average number of retweets when inclusive language was and was not used was less than one. On the other hand, for Angle the average number of retweets was five more when inclusive language was used, (33 retweets) than when it was not used (28 retweets). Another factor at play may have been use of hashtags. Angle used #nvsen in 97 percent of her tweets, thus making them easily accessible on Twitter, while Reid used #nvsen 88 percent of the time. More noticeable was the difference in Angle’s inclusion of hashtags to be used by likely supporters. Angle also used the hashtag tcot, “top conservative on twitter”, 91 percent of the time, while Reid only used the hashtag p2, “progressives 2.0”, 6 percent of the time. She also used the hashtag dumpreid 86 percent of the time. Reid failed to consistently use a similar slogan hashtag. I then looked at the tweets that used #tcot and those that did not use #tcot and a difference emerged. The mean number of retweets was 31 when she used it and dropped to 27 when she did not use it. However, this was not statistically significant because the sample size for tweets that did not have #tcot is only 12. A similar pattern did not emerge with #dumpreid. In this case, the difference in number of retweets was less than one.
  • 44.   44   IX. Conclusions This study found that Sharron Angle and Harry Reid had different approaches to using Twitter. Angle tweeted less frequently, sounded more genuine, and perhaps in return was retweeted more and had more followers. Reid used Twitter to amplify the same campaign messages he was using elsewhere and to drive people to his Web site and YouTube page. In regards to answering the question,” was Twitter used to jumpstart the discussion of campaign issues, or continue existing discussions”, the small dataset in this study gave limited results, which limited my ability to speak to this question. The study suggests that for national issues, Twitter is used to continue discussion about the issues being talked about in the mainstream media. This study could not tell which came first because they were often on the same day and also because these issues were already present in the campaign before October 1. However, for smaller, more localized issues, it suggests this is not the case. The issues are often discussed more in the paper. If they were discussed on Twitter, it was by the opposing candidate. In regards to the question,” are candidates using Twitter as a means to attack their opponent or to relay positive information about themselves”, this study found that candidates are using Twitter for both. Overall, Twitter was used slightly more for promotional purposes. However, the amount of tweets used for promotion and attack varied greatly across candidates. Angle only used a third of her tweets to attack Reid while Reid used over half of his tweets to attack Angle. In regards to the question, “if attacks are present, are they direct or indirect”, this study found that attacks were mostly indirect. In other words, the majority of the attacks referenced news stories and/or had links present in them. More interestingly was the fact that over a third of attack tweets mentioned news stories and linked back to the candidate’s own Web site. Thus,
  • 45.   45   Twitter can be used to drive traffic to the candidate Web sites by posting news stories. This may be because referencing an outside source increased the credibility of the attack and because it is a common practice in other aspects of campaigning. In regards to the question, “are candidates using Twitter more frequently to inform people about campaign events, post news, discuss issues, reinforce campaign messages, or for campaign asks (i.e. donations, GOTV, signing up for events)”, this study found that candidates most frequently use Twitter to post news and discuss the issues. Since Twitter is so instantaneous, it is not surprising that it is being used to release campaign news. In regards to the question, “does the purpose change over the course of a campaign,” this study found that the answer is yes. At the beginning, Twitter was used more to discuss campaign issues and news regarding issues. However, as Election Day neared, GOTV messages flooded the Twitter accounts of both candidates. This suggests that like with other forms of social media, Twitter was used at the end of the campaign to drive supporters to the polls. This finding along with the finding that 62 percent of the tweets link back to a candidate’s Web site or YouTube page suggests that Twitter use in campaigns may align better with the reinforcement theory than the mobilization theory. In regards to the question, “does inclusive language makes people more likely to retweet the candidates,” this study’s findings suggest that it does. The difference in the mean number of tweets is significant (t=3.82, p<.05) when inclusive language was used. However, this was driven by Sharron Angle whose mean number of retweets rose by 5 when she used inclusive language. On the other hand, for Harry Reid, who only used inclusive language 4 percent of the time, there was no difference in mean retweets. Since Harry Reid averaged much fewer retweets than Sharon Angle (10 vs. 30) and rarely used inclusive language, this supports the belief among
  • 46.   46   experts that sounding genuine on social media matters. Inclusive language is one way to sound more genuine, and Sharron Angle shows that this also helps expand reach throughout the “twittersphere”. This also supports the reinforcement theory because as President Obama exemplified in the 2008 election cycle, using inclusive language is a way to activate your base. In regards to the last question,” does using consistent hashtags lead to more retweets,” this study suggests that the answer is yes, it does. Sharron Angle who consistently used hashtags got retweeted an average of three times more than did Reid. Angle’s mean number of retweets rose by four when she included #tcot, which she used over 90 percent of the time, than when she did not. This was not statistically significant because of the small sample size in tweets without #tcot. Unlike Angle, Reid only used #nvsen 88 percent of the time and #p2 6 percent of the time. Sharon Angle’s consistent use of the hashtags nvsen and tcot are only one way her Twitter use was different than Harry Reid. Therefore, it is important to take into account the other differences in Twitter style between the candidates and realize hashtags may only be one contributing factor to her ability to get retweeted. As the Internet becomes more integrated into campaigns, it is important to continue studying how new innovations online contribute to campaign strategies. As history shows, strategies online are constantly evolving and studies must stay up to date. For example, YouTube was not around prior to the 2006 campaign cycle and candidates did not use Twitter before the 2008 cycle. Since innovations online are happening so quickly, it is important that each election cycle, questions are raised on how candidates are integrating the new techniques into their online strategy and overall campaign strategy so that there is no gap in the literature on this subject. The increasing trend in people turning to the Internet for campaign information proves that no online source of information distribution can be forgotten. Before this study was
  • 47.   47   conducted, there were very few studies available on Twitter use in campaigns. This is partially because before 2010, it had only been used in four major races. Unlike with older online outlets like campaign Web sites and YouTube, there is no uniformity within Twitter strategy. Therefore, this study shows that candidates still have a great deal to learn about effective Twitter use. This study is important because it increases the available knowledge on how Twitter can be most effectively used. It shows that Twitter can and should be incorporated into other online strategies through use of links. This could be links to the candidate’s Web site, other social media sites, or even outside news articles. Since posting on Twitter is so immediate, it can be a useful outlet to release campaign information on or for posting breaking news stories that are favorable to the candidate. Based on this study, I conclude that even if the candidate is not the one tweeting, those receiving the messages must believe that the tweets are coming from the candidate. If voters do not think this is the case, they may be less susceptible to become activate supporters on Twitter or elsewhere. This is mainly because there was a significant difference in number of retweets when inclusive language was used in this study. I believe that the best use of Twitter in campaigns is twofold. Candidates should use Twitter to reach a wide audience that may be missed in other forms of campaign strategy and to energize their base as Election Day nears. This audience specifically includes the younger voters who may not actively seek out campaign information. If candidates can use Twitter to get retweeted by people with large Twitter followings, their messages will be broadcasted to a large audience of people who are not choosing to seek out political information. If they make their tweets interesting and include links to their Web site or to their other social media pages, they can drive potential voters to sites where they have complete control over the message. If they can
  • 48.   48   consistently use hashtags and inclusive language, they may engage more people and help their messages go viral. Secondly, Twitter should be used like other online forms of social media to energize and activate the candidate base. This can be done by using hashtags that your base will be likely to see, linking to your site, including campaign asks, and using inclusive language. For example, if you are a conservative candidate, you can use the hashtag tcot to specifically reach conservative voters. This can be used to reach conservatives in Nevada and elsewhere. Since campaign asks, like fundraising, are not limited to state residents, Twitter may be used to target out of state contributors. This study creates a baseline to which Twitter usage in the future can be compared. It should be noted that the number of people following these candidates was low and that the number of retweets was also relatively low. However, since other forms of social media, like YouTube and Facebook, did not catch on instantly, but grew over time, the same trend is likely to occur with Twitter. For example, today, candidates are expected to invest in a Web site, but just 10 years ago, less than a majority of candidates for Congress had Web sites. I expect that as time passes, more candidates and more citizens will use Twitter. This study paves the way for further analysis on how candidate use of Twitter is different than candidate use of other forms of social media. Candidates must understand Twitter as a medium to use it to its maximum potential. This study could be expanded upon in a few different ways to give a more complete picture of how Twitter use influenced the campaigns. First, this study did not look at actual fundraising data and how it compared to Twitter use. Looking at this data would have helped determine if Twitter can help candidates fundraise as other online technologies have done and
  • 49.   49   see support the notion that Twitter can be used to reach potential contributors nationwide. It also does not compare increase in number of followers to fundraising patterns. Secondly, this study was unable to compile date on the users that retweeted the candidate such as how many followers they had and what their ties to the election were. Getting this information would help further answer the question of how large of an audience the tweets are reaching and help campaigns understand who they should target on Twitter. If celebrities are retweeting candidates, they may have a much larger Twitter following than the candidates themselves and be able to reach people who would not typically be motivated to search for campaign information. Also, if those retweeting and/or following the candidates are journalists, the candidates’ messages may have a wider reach because of the amount of influence journalists have on agenda setting and because of the network of people those journalists can reach. Thirdly, this study only looked at the Las Vegas Review Journal in print. Perhaps looking at other mainstream news outlets in both print and their online publications would help determine who discusses the issues first. To further address that question, a longer time period could also be observed. The Nevada Senate race was one of the most monumental races in the 2010 Midterm elections because it captured the mood of the electorate during that election cycle. The fact that a Tea Party candidate who had radical beliefs including abolishing the Department of Education could be in a predicted dead heat with the Senate Majority Leader showed how disgruntled voters were with the Democratic Congress and Obama administration. Studying how these two extremely different candidates used the newest online outlet, Twitter, provides insight into how candidates in general can use this outlet to reach voters and win their race. In close races, like this one, every piece of the overall campaign strategy matters. Therefore, political
  • 50.   50   communications experts must continue studying new innovations as they begin to be integrated into campaigns to figure out best practices and help candidates convince more citizens to vote. Appendix A Coding Sheet: Tweets 1. Tweet number a. Write the ID of the tweet. 2. Date a. Write the date of the tweet. 3. Time a. Write down the time the tweet was recorded. 4. Hashtags (For Reid: #nvsen, #gotv, #p2) (For Angle: #nvsen, #dumpreid, #tcot) a. See if each of the assigned hashtags are present or not present b. If present, write a 1, if not present, write a 0 5. Message of the Tweet Since it is possible to relay more than one type of message per tweet, this will not be mutually exclusive, but rather each type will be marked with “yes” or “no”. Categories include GOTV, other calls to action, campaign events, news stories, and issue mention a. GOTV i. Definition: GOTV messages include tweets that ask people to vote, ask people to get out the vote, or relay information about voting. Included this category are candidate retweets of Twitter users. Examples of voting information include information about polling locations and when people can vote. ii. Is there a GOTV message in the tweet? 1. If yes, mark 1 2. If no, mark 0 b. Other Calls to Action i. Definition: Other calls to action focus on something other than getting people to simply vote for the candidate or push others to do the same. Examples include fundraising asks, event sign-ups, email sign-ups, and asking to take action on twitter. Tweets asking people to show their support on Twitter, such as through changing your Twitter picture to a campaign badge, are also included in this category. ii. Is there an other call to action message in the tweet? 1. If yes, mark 1 2. If no, mark 0 c. Campaign events i. Definition: campaign events are tweets that relay information about campaign events, including dates, time, locations, topics discussed, type of events, and guest speakers. If images from events are posted, they will be counted in this category. ii. Is there a campaign event message in the tweet? 1. If yes, mark 1.
  • 51.   51   2. If no, mark 0 d. News stories i. Definition: News stories are tweets that reference campaign news stories. They can include links to actual stories or just reference the story. References to polling data released will be included in this category. For example, a tweet that says ABC exit poll shows Angle up by 2.7 would be counted here. When there are links to news stories, the links will also be coded by the same schema used to code news stories from the Las Vegas Review-Journal. ii. Is there a news story mentioned in the tweet? 1. If yes, mark 1 2. If no, mark 0 e. Issue Mention i. Definition: include tweets about a substantive issue. General campaign rhetoric does not count here. Mark each specific issue as follows: 1. 1=healthcare; 2=immigration; 3=economy; 4=government spending (includes stimulus and social security); 5=jobs (including unemployment); 6=big government (includes taxes); 7=education; 8=Angle’s ignoring mainstream media; 9=the Hispanic vote; 10=farming; 11=nuclear waste in Yucca; 12=Ashjian recording his conversation with Angle; 13=Pork barrel spending; 14=other 2. If no issue is mentioned, mark 0. 6. Personal Observation a. Definition: Personal observation tweets are tweets that sound genuine and like they are coming from the candidate. For example, a tweet like “great turnout at the event” would count as personal observation. Another example could be “thank you for helping spread the message.” b. If present, mark 1; if not, mark 0. 7. User Retweets a. Definition: User retweets are tweets from other Twitter users that the candidate posts, with or without adding their own message. b. If present, mark 1; if not, mark 0. 8. Inclusive language a. Definition: Inclusive Language includes tweets that use language to include the reader of the tweets. Specific words that count in this category are us, we, you, your, yours, and our. b. If present, mark 1; if not, mark 0. 9. Links a. I will code for three mutually exclusive types of links. Types of links include: i. Link to the campaign, via a campaign Web site or official social media site (i.e. YouTube, facebook). ii. Link to party site, a site that is owned by a particular party. If a tweet links to the DNC, it would count here. iii. Link to an outside source, a site that is not owned or managed by the campaign. iv. Photo link, a link to some sort of photo sharing site, like twitpic.