|UFDC Home||myUFDC Home | Help|
This item has the following downloads:
1 POLITICAL ATTITUDES IN 140 CHARACTERS OR LESS By HEATHER HOUSTON A THESIS PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF ARTS IN MASS COMMUNICATION UNIVERSITY OF FLORIDA 2012
2 2012 Heather Houston
3 To Dr. Kaid, a mentor and friend
4 ACKNOWLEDGMENTS This thesis would not have been possible without the support of many people. I wish to express my sincere gratitude to my advisor and chair, Dr. Cory Armstrong, as well as the members of my committee, Dr. Johanna Cleary and Dr. Michael Martinez. I am grateful for the invaluable guidance and encouragement from Dr. Lynda Kaid I would like to thank Jody Hedge, Kimberly Holloway, Sarah Lee and Bridget Grogan for their support throughout my graduate studies program. I would like to convey my appreciation and love to my husband, Hal Houston, for his support in all my endevours, to my brother in law, Gray Houston, for his infinite wisdom and interest in this project, and to my family for their faith and understanding throughout my graduate school career.
5 TABLE OF CONTENTS page ACKNOWLEDGMENTS ................................ ................................ ................................ .. 4 LIST OF TABLES ................................ ................................ ................................ ............ 7 LIST OF FIGURES ................................ ................................ ................................ .......... 8 LIST OF ABBREVIATIONS AND DEFINITIONS ................................ ............................. 9 ABSTRACT ................................ ................................ ................................ ................... 11 CHAPTER 1 INTRODUCTION ................................ ................................ ................................ .... 12 2 LITERATURE REVIEW ................................ ................................ .......................... 16 An Affordance Theory Analysis of Twitter ................................ ............................. 16 Political Identification ................................ ................................ .............................. 18 The I nfluence of Political Identification ................................ ................................ .... 20 Political Attitudes ................................ ................................ ................................ .... 24 Extremity ................................ ................................ ................................ .......... 25 Persuasive Strategies ................................ ................................ ...................... 26 Behavioral Intent ................................ ................................ .............................. 28 The Influence of Social Networking Sites on Polarization ................................ ....... 29 Attitude Strength in Online Political Information ................................ ...................... 31 Issues as Political Cues ................................ ................................ .......................... 34 3 METHODS ................................ ................................ ................................ .............. 35 Sample and Units of Analysis ................................ ................................ ................. 35 Coding Instrument ................................ ................................ ................................ .. 36 Measuring Independent Variables ................................ ................................ .... 37 Measuring Dependent Variables ................................ ................................ ...... 38 Coding Procedure ................................ ................................ ................................ ... 40 Coder Training and Pretest ................................ ................................ .............. 40 Coder Independence ................................ ................................ ........................ 41 Intercoder Reliability ................................ ................................ ......................... 41 Biographies ................................ ................................ ................................ 42 Twitter messages ................................ ................................ ..................... 42 Linked content ................................ ................................ ............................ 43 Data Analysis ................................ ................................ ................................ .......... 44
6 4 RESULTS ................................ ................................ ................................ ............... 46 5 DISCUSSION AND CONCLUSION ................................ ................................ ........ 53 Extremity in Twitter Messages ................................ ................................ .............. 53 Extremity and Persuasion Strategies ................................ ............................... 54 Extremity and Issue Mention ................................ ................................ ............ 56 Attitude Strength in Linked Content ................................ ................................ ........ 57 Limitations and Future Directions ................................ ................................ ........... 59 Conclusions ................................ ................................ ................................ ............ 61 APPENDIX: CODING GUIDE ................................ ................................ ................... 64 REFERENCE LIST ................................ ................................ ................................ ........ 71 BIOGRAPHICAL SKETCH ................................ ................................ ............................ 78
7 LIST OF TABLES Table page 3 1 Intercoder reliability indices for Twitter user biographies ................................ .. 45 3 2 Intercoder reliability indices for Twitter messages ................................ ............ 45 4 1 Frequencies in Twitter user biographies ................................ ........................... 49 4 2 Frequencies for Twitter messages ................................ ................................ .... 49 4 3 Frequencies for attitude strength in linked content ................................ ............. 50 4 4 Comparison of political party and ideology mentions in Tweets ......................... 50 4 5 Comparisons of extremity and persuasive strategies ................................ ......... 50 4 6 Extremity and behavioral intent ................................ ................................ .......... 51 4 7 Persuasion strategies and behavioral intent ................................ ....................... 51 4 8 Attitude strength in linked content found in tweets that express extremity, persuasion strategies and behavioral intent ................................ ....................... 51 4 9 Comparing extremity and issue mention in tweets ................................ ............. 52 4 10 T test attitude strength and issue mention ................................ .......................... 52 5 1 Critical views cross partisan and ideological lines ................................ .............. 63
8 LIST OF FIGU RES Figure page 5 1 Extremity and attitude strength across Twitter ................................ .................. 63
9 LIST OF ABBREVIATION S AND DEFINITIONS A TTITUDE is expressed by evaluating a and Chaiken, 1993 as cited in Clawson & Oxley, 2013, p. 17) A TTITUDE E XTREMITY midpoint of favorabl e Boninger, Chuang, Berent and Carnot, 1993 p. 1132) In the present study the dimensions include: favorable unfavorable, good bad, f oolish wise, harmful beneficial A TTITUDE I NTENSITY he strength of emotional reaction provoked by the attitude object in an individual and is typically measured using self Chuang, Be rent and Carnot, 1993, p. 1132) A TTITUDE S TRENGTH he extent to which attitudes manifest the qualiti es of p. 3). According to Krosnick and Petty (1995), durability refers to the stability of an attitude over time and it s ability to withstand attack B EHAVIORAL I NTENT T t, future plans or encouragement for the cu rrent or future plans of others C ERTAINTY he degree to which an individual is confident that his or her attitude toward an object is correct and is usually gauged by self snick, Boninger, Chuang, Be rent and Carnot, 1993, p. 1132) LIWC2007 Linguistic Inquiry Word Count 2007, an automated textual analysis program based on 80 linguistic categories O PINION E xpression of a preference in regards to a specific object (Clawson & Ox ley, 2013 p. 17) P EOPLE B ROWSR S ocial media analytics platform
10 P ERSUASION S TRATEGIES Persuasion strategies include associating compromise with failure (Campbell et al., 1960; Gusfield, 1962, p.22), denying the legitimacy of a political institution ( Gusfield, 1962, p.23), closing off deliberation to other political viewpoints (Gusfield, 1962, p.23), proposing an end to political institutions (Gusfield, 1962, p.23), attributing one factor for causing current conditions (Brock et al., 2005, p. 102; Smit h, 2003), emphasizing a single value such as mor alism, self reliance, equality (Brock et al., 2005 p. 102; Campbell et. al, 1960; Gusfield, 1962 p. 22; Smith, 2003) interest is at stake (Binder et al., 2009, p. 317) SNS Social n etworking site SPSS Statistical Package for the Social Sciences a statistical software program T WITTER time information network that connects you to the latest stories, ideas, opinions and news about what you find interesting. (Twitter 2012) T WITTERVERSE The environment and events that occur across Twitter
11 Abstract of Thesis Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Maste r of Arts in Mass Commu nication POLITICAL ATTITUDES IN 140 CHARACTERS OR LESS By Heather Houston August 2012 Chair: Cory Armstrong Major: Mass Communication Twitter is a microblogging website where users read and write millions of short messages every day. This study uses the context of the 2011 debt debate to explore attitudes expressed on Twitter and their influence on political polarization. This study is a content analysis using both manual coding and the computer assisted program LIWC2007 to sample 4,500 messages, us er biographies and online information linked in messages disseminated by politically interested and active Twitter users. Independent among persuasive strategies, ment ions of political issues and linked content that expressed attitude strength in messages that expressed attitude extremity. To explain these strong relationships, the researcher speculates that politically motivated individuals seek online content to supp ort their attitudes and participate in politics while persuading others to follow by disseminating information. The present study extends previous research about Twitter practical applications of ef fective online content development by incorporating attitude strength, issue mention, persuasion strategies, and extreme positions to exp and political influence online.
12 CHAPTER 1 INTRODUCTION In the summer of 2011, the political atmosphere was as hot as th e temperature participation by the masses. Concurrently, the rise of Twitter provided a perfect outlet for elites to get out their message, news media to cover the story, and individuals to join the discussion. President Obama gave two speeches in the last week of July as the Now, on Monday night, I asked the American people to make their voice heard in this debate, and the response was overwhelming. So please, to all the American people, keep it up. If you want to see a bipartisan compromise -a bill that can pass both houses of Congress and that I can sign -let your members of Congress know. Make a phone call. Send an email. Tweet. Keep the pressure on Washington, and we can get past this. (President Obama, July 29, 2011, para. 9) The present study analyzed tweets from July 25, 2011 through August 2, 2011 to explore attitudes expressed on Twitt er and their influence on political polarization. The time period of the present study was choosen because of political events that occurred offline that may spur behavior online. During this week, offline events increased pubic awareness about the debt ceiling negotiations. On July 29 th President Obama encouraged citizens to contact their Congressperson, mentioning Twitter as an avenue for communication. Later that week, United States Senator from Florida, Marco Rubio gave a passionate speech on the Senate floor, stating the stalled negotiations between events to behavior during a campaign, t he present study focused on these
13 relationships during national debate about a salient issue and an impending deadline for a solution. The relationships among salient political issues, discussion and participation indicate the health of a democratic system Normative democratic theory requires citizens to participate in politics after obtaining information, evaluating individual and political attitudes (Delli Carpini & Keeter 1996). How the Internet has reshaped this process and its impact on the health of democracy is a topic of debate. Online access to political information offers users a variety of sources. Stroud (2011) argued a broader range of content could spur poli tical interest among new audiences, improve knowledge levels, mobilize citizens and increase political participation (Stroud, 2011). Counterarguments advocate Internet users will select information that aligns with their predispositions to avoid challengi ng their political attitudes (Abramowitz & Saunders, 2008; Abramowitz, 2010; Blum, 2011; Prior, 2006). Some argue this could harm a democracy by dividing partisans, preventing consensus building around salient political issues (Stroud, 2011, p. 176). The Internet has transformed both how people participate in politics and acquire political information. According to the Pew Internet and American Life Project (2011), more than one half of American adults gathered information online during the 2010 mid term elections. In fact, online news, campaign activities and social networks were the sole source of information for one fourth of American adults (Pew Internet and American Life Project, 2011). One available social network is the micro blogging site Twitter the fastest growing social media tool (Hendricks & Denton, 2010, p. 38). Twitter provides users
14 the ability to spread information quickly using messages called tweets. Each tweet is limited to 140 charactes. Despite being140 characters or less, Twitt er has shown to be an effective tool for political participation and offers a platform to express attitudes, engage in political discussions, disseminate information, advocate preferences and mobilize others (Howard, 20111; Lipschultz, 2011; Ostrow, 2011) According to Hendricks & Denton (2010), these features afford users the ability to influence electoral and policy outcomes, some worry at the cost of civil discourse and democratic deliberation. The present study explored how Twitter served as a tool f or political participation through the dissemination of online content during the debt ceiling negotiations that began on July 25, 2011 and ended on August 2, 2011. The researcher selected a purposeful sample of likely interested and aware Twitter users as determined by the social media analytics platform, PeopleBrowsr (www.peoplebrowsr.com). Examining this group offers a better description of the features associated with polarization. Using a quantitative content analysis with both manual and computer assisted coding, this study analyzed how users incorporated political identification in Twitter biographies, expressed political party, ideology and attitudes in tweets and linked content. As noted above, the debt ceiling debate of the summer 2011 was ch osen as the issue for analysis. The debt ceiling debate is an appropriate topic for the study for several reasons. First, events surrounding the negotiations brought greater recognition of Twitter site became infamous for being peppered with partisan sentiment (Howard, 2011; Lipschultz, 2011; Ostrow,
15 time period (Howard, 2011 para. 1). Lastly, White House staffers credited social media, specifically Twitter prominent issue of the day (Howard, 2011; Lipschultz, 2011). There are several potential implications for this research. First, the present stu dy will lead to a more developed theoretical understanding of Twitter American politics by describing the nature and dissemination of political information across the site. For political communication scholars, this study contributes to a deeper results should provide a greater understanding of online information that may motivate a political base, an aware and active group of supporters. Finally, this research may be used to develop online messages that will build support within constituencies, form coalitions and maneuver public opinion more effectively in the digital environment.
16 CHAPTER 2 LITERATURE REVIEW The impetus for the present study is to bet ter understand Twitter democracy. More optimistic views have argued sites like Twitter spur discussion and mobilize political participation ( Boulianne 2011; Brundidge, 2011; Kim, 2011) From this perspective, social media lives up to Jrge n ideal of the public sphere: a space outside private life and institutions reserved for public discussion and debilberation (McQuail, 2010; Brundidge, 2011; Kim, 2011) However, several perspectives challenge this notion and claim social networ king sites (SNSs) foster a polarized landscape. Polarization occurs when opposing issue positions align with political identities (Abelson, 1995). Previous research has approached this topic by analyzing hashtags, followers, partisan and ideological senti ment (Conover, Ratkiewicz, Francisco, Goncalves, Flammini & Menczer, 2011; Conover, Goncalves, Ratkiewicz, Flammini & Menczer, 2011; King, Orlando and Sparks, 2011; Tumansjan, Sprenger, Sander & Welpe, 2010). However, polarization is shaped by attitudes t hat are influenced by a variety of factors (Zaller, 1992). To better understand this process, specifically whether communication of political information across Twitter encourages polarization, more research is needed about the characteristics and enviro nment surrounding this process. An Affordance Theory Analysis of Twitter In 1977, J.J. Gibson coined the term affordances to describe factors that interact the approac hes to human psychologies that were solely cognitive. He favored a theory that focused on the interaction of environmental attributes, personal characteristics, and
17 considered salient events or patterns (Gaver, 1991 p. 79). Thus, the theory of affordanc es provides a framework that considers Twitter characteristics of its us ers and the political climate. For example, individuals who are compelled to stay informed about a particular topic may perceive Twitter as an avenue for ga thering topic specific information by following other users with similar interests. The theory of affordances has been previously used to analyze the adoption and use of new technologies. In Technology Affordances Gaver (1991) incorporated this concept t o postulate what benefits certain technologies provide to users. He noted benefits from technology exist regardless of whether they are recognized. Additionally, technology users can perceive an affordance that is nonexistent. The user s past experience s, social background and intent determine whether an affordance is perceived. Perception and its strength is the driving force behind behavior. technologies and media in terms of the act Twitter select political information, political messages online and incorporate their own opinions. Despite the 140 character limit, tweets have dra matically altered the flow of political information from elected official to citizen and from citizen to elected official. Political messages in the form of tweets have the potential to influence political outcomes based on their scalability. In 2009, a in was added to President Barack pp. 46 47). When clicked, this and the Twitter username of t
18 button mobilizes constituents by lowering the requirements for political participation. President Obama is using Twitter to maintain engagement after Election Day and build support for his agenda. Constituents utilize his SNSs to demonstrate their preference Twitter offered affordances to President Obama and constituents alike. More recent events support for this claim. Twitter offered seve ral affordances to its users during the debt debate. The United States government was closely approaching deadline for a resolution to the debt dilemma. On July 29 2011 President Barack Obama called on Americans to tweet their congressman. Constituents could reach elected officials while bypassing busy phone lines and avoid drafting emails or letters. While users may be constrained by140 characters, elected officials are focused on tracking how constituents would like them to vote on an upcoming issue. For instance, a fifteen minute phone call explaining a route from constituent to elected official. Thus, 140 characters are more than sufficient for participa tion in the political process under this defination of participation. The present study was designed to describe the characteristics that afford political participation online. It focused on the relationships among attitude extremity, strength, information and participation by politically aware Twitter users. By examining these attributes, the present study offers a different perspective of the features credited with instigati ng polarization on Twitter Each attribute will be discussed below, in turn. Political Identification Self identifications of political party and ideology have been used to analyze how citizens structure attitudes (Converse, 1964; Zaller, 1992), evaluat e information (Downs,
19 1957; Key, 1966; Popkin, 1994; Zaller, 1992) and participate in politics ( Berelson, Lazarsfeld & McPhee, 1954; Campbell, Converse, Miller & Stokes, 1960). Party an enduring sense of psychological attachm ent to a political Goren, Federico & Kittilson, 2009 p. 806). close woven, and far Campbell et al., 1960 p. 192). While distinct, these concepts are related to one another. Generally, both identifications remain the throughout their lifetime (Campbell et al., 1960; Conover & Feldman, 1981). Both identifications are evaluations of salient political issues candidates and events (Campbe ll et al., 1960; Conover & Feldman, 1980; Goren et al., 2009). However, party and ideology influence and structure political attitudes differently (Converse, 1964; Conover & Feldman, 1 981). Previous studies have linked these discrepancies to knowledge, interest and engagement levels among the electorate (Converse, 1964; Conover & Feldman, 1981). Public opinion research has relied on measurement of political sophistication to decipher t he relationships among political attitudes and liberal conservtive orientations (Goren, 2011). Attitudes structured using this contiuum rather than partisanship often reflect a higher level of political sophistication (Delli Carpini & Keeter, 1996; Zaller 1992). Alternatively, Goren (2011) found political sophistication has little influence when it comes to policy prefere n ces based on core beliefs and values about equal opportunity, self reliance and limited government, the military and anticommunism. In his study, citizens incorporated values when evaluating salient issues similarly across levels of political sophistication. However, core principles were incorporated at low
20 rates across all levels of political sophistication. Attitude structure, whet her ideologically or using partisanship, influences poltical participation. There is general agreement among political communications research that party identification in more influencial in shaping attitude s than shaping ideology ( Campbe ll et al., 1960 ;Converse, 1964). The mass public does not frame thought process es using political ideology (Converse, 1964). Rather, ideological identification on & Tedin, 2007 p. 82). Politically aware citizens look to elites for cues indicative of the appropriate ideology that aligns with their party preference. These cues lead to consistent political attitudes and motivate political behavior. Previous rese arch has analyzed political identifications by studying voting behavior, public opinion and media choice ( Campbe ll et al., 1960; Iyengar & Hahn, 2007 ; Zaller, 1992 ). More recently, research has predicted political alignment of Twitter users by analyzing messages, hasgtags and followers ( Conover et al., 2011; Golbeck & Hansen, 2011). However, the present study is the first to examine whether politically interested and engaged Twitter users identiy using ideology or political party. This gap in political communications research informs the following research question: RQ1 a : For Twitter users interested in politics, is partisanship or ideology used to describe the user in Twitter biographies? RQ1b: Do Twitter users mention a dominant political party or ideology in tweets about the debt ceiling debate? The Influence of Political I dentification Various contexts have been used to analyze the formation, influence and behavior associated with political identification. However, vote choice is the most commo n.
21 Voting choice or behavior refers to how individuals form preferences and an electorate reaches political decisions. Classic democratic theory argued election results and partisan attitudes reflect a rational electorat e who desires and i s able to form preferences on issues, candidates, and initiatives (Berelson, 1944). This process modeled a deliberative democracy where participants were interested in public affairs, maintained an accurate perception of political realities, engaged in political discour se, and considered the interests of others, the nation, as well as their own when forming preferences. However, early election studies determined political attitudes and voting behavior did not reflect normative democratic theory. and Voting credited sociological characteristics with & Gaudet, 1944; Berelson, Lazarsfeld & McPhee, 1954). In this model, socioeconomic status, race, age, geographic regio n, gender and religion were listed as accurate predictors of vote choice. Campaigns, salient issues and news coverage had little impact on attitudes or electoral decisions. Social and family groupings explained exposure to political information, attitude s and vote decisions. In the sociological model, political attitudes and voting behavior were reinforced, but rarely changed Previous theories of voting behavior failed to explain the shift in political opinion favoring the Republican Party marked by the 1952 presidential election (Campbell, Converse, Miller and Stokes, 1960 p. 66). In The American Voter authors attributed several characteristics outside a purely soci ological model (Campbell et al., 1960, p. 66). In addition to sociological factors, candidate attributes, political groups, domestic
22 policy, foreign policy, and political parties track record while in office influences political preferences (p. 67). Accor ding to Campbell et al. (1960), partisan and group identifications determine long term voting behavior. Short term factors include candidate characteristics and salient issues, both of which change each election. The formation of political preferences and whether they are based on long or short term factors, reflect how individuals process political information. Campbell et al. decisions (Campbell et al., 1960 p. 42). New information is evaluated and assim identification. Several political scholars have explain ed identification using an alternative approach, often called revisionist models. Revisionist models described partisanship as a comparative evaluation of political parties (Abramson et al., 2010 p. 195). This model is similar to its classic counterpart in two significant ways. First, revisionist models focused on how one processes and evaluates new political information (Green, Palmquist & Schickler, 2002 p. 110). Second, partisan identification played a central role in vote decisions (Abramson et al. 2010 p. 195). Revisionists diverge from earlier models regarding the development and structure of partisan attitudes (Green et al. 2002; Abramson et al., 2010). Revisionist approaches credit both past and present evaluations with determining partisan identification and directing voting behavior (Abramson et al., 2010 p. 196). In An Economic Theory of Democracy (1957), Anthony Downs argued that voters base preferences on the perceived benefits expect ed from one party being in power verses
23 the other. Downs argued that citizens form preferences by incorporating the information they receive into their conception of an ideal society. According to V.O. Key Jr., (1966), if voters fail to encounter information challenging their predispositions or perceive both parties as offering similar benefits, vote decisions are based on partisan identification. Morris P. Fiorina (1981) described partisanship as a summary evaluation of the political past. Similarly, Achen (1992) posited that continuously updated obser vations would eventually lead voters to choose the political party that habitually resonated with their worldview. Revisionist theories challenged previous theories in several ways. Earlier models credited social identification and psychological processes with determining partisan affiliation (Berelson, Lazarsfeld, & McPhee, 1954; Campbell et al., 1960). According to Green, Palmquist, and Schickler (2002), revisionist studies focused on how individuals obtain and assimilate new political information. Thes e approaches highlight the dynamic relationships among partisanship and perceptions, evaluations and ultimately candidate and policy preferences (Abramson et al., 2010 p. 207). t reflect a true ideological position or identification For those who identify with a particular ideology, Conover and Feldam (1981) contend these attitudes reflect evaluations of symbols associated with groups, mainly political parties Therefore, pari stanship remains the core component influencing political attitudes and behavior. As previously mentioned, only a small fraction of individuals use ideology to structure their political attitudes.
24 The confluence of identification and political sophistic ation influences an indivdual in several ways. First, identification and political sophistication impact an responsive to political polarization among political elites than lesser informed individuals vote choice. Lesser informed individuals are more likely to make voting decisions inconsistent with their interests (Niemi et al., 2011). Las tly, inde n tification and will be addressed later in this review. More recent research has examined how attitude structure may influence the flow of political inform ation on Twitter Wu, Mason, Hofman and Watts (2011) determined the differences in attitude structure led to differences in content produced by users onTwitter Political elites are a small and focused group, however Twitter affords them the ability to expand the reach of their messages through politically aware intermediaries ( Epstein and Kraft 2010; Wu et al. 2011). Wu et al. (2011) advocated that future research analyze how this information flow influences political attitudes and dialogue across this channel on Twitter In an effort to fill this gap in research, the present study examined attiudes and attributes expressed in tweets generated by politically aware users. Political Attitudes a general evaluative Often political a ttitudes focus on an toward candidates, issues, institutions or groups (Clawson & Oxley, 2013). Through political socialization, individuals acqui re information early in life that establishes political attitudes, ideological values and party preferences (Downs,
25 1958). During both their formation and evolution, attitudes and their attributes are shaped by information (Erickson & Tedin, 2007). Gener ally, political attitudes endure over time, shifting mainly with transformations in socio economic status, education, events o r other life altering changes. The influence of political attitudes on the behavior, preferences and information processing is att itude strength (Abelson, 1995). Public opinion research has evaluated attitude strength by analyzing attitude attributes individually and in relation to one another (Krosnick, Boninger, Chuang, Berent & Carnot, 1993). Some of these attributes are studied directly using self reports, while others rely on inferences derived from interviews or expressions (Abelson, 1995). Both approaches have analyzed attitude direction and importance (Abelson, 1995; Clawson & Oxley, 2013 ). Direction refers to whether an i ndividual supports or opposes an attitude object, while importance indicates the saliency of an attitude ( Abelson, 1995; Clawson & Oxley, 2013 ; Krosnick et al., 1993). The absolute value of an attitude position, regardless of the direction, indicates how f ar an attitude deviates from neutrality. The present study conceptualized extreme positions as the extent to which an attitude deviates from neutrality. To examine whether these positions were significant and their presentation in tweets, the present stu dy analyzed the use of persuasive strategies and expressions of behavioral intent in Twitter messages. This section details these conceptions and their relationship to one another and their manifestation in Twitter messages. Extremity Attitude extremi from the midpoint of favorable un 1993 p. 1132).
26 attitude toward an object is positioned at the midpoint, positively or negatively along an attitude scale (Binder et al., 2009; Smith, 2003). According to Smith (2003) s imple solutions and ideological consistency are the core criteria for possessing pol i tically extreme attitudes. Extreme attitudes can emerge anywhere along the ideological continuum (Smith, 2003). However, his study found that conservatives were more likely to hold ideologically consistent dispositions, thus, are more inclined to hold e xtreme attitudes than moderates or liberals. To better understand the role of political communication on attitude extremity, Binder, Dalrymple, Brossard, & Scheufele (2009) questioned if participation through engaging in political discussions led to or was the resu lt of political extremity. Their study found that interpersonal discussions solidified preferences and led to political extremism (Binder et al., 2009 p. 333). The micro blogging site Twitter mirrors the interpersonal discussion setting. Thes e findings support earlier research that uncovered correlations between political discussion and extremity (Krosnick et al., 1993). Zaller (1992) contended individuals update political attitudes when called on to do so through political participation, in cluding discussion. At this time, political elites p. 895). When activated, individuals rely on values when shaping their attitude position. The saliency of these va persuasive affect on politically attentive and aware individuals. Persuasive Strategies To describe how political information, particularly persuasion, is expressed in tweets the present study examine d rhetorical strategies. Twitter messages that use affective
27 appeals indicate persuasion, possibly persuasion that incites political extremity. The partisan source cues and ethos a ppeals in linked content are likely predictors of this type of opinion expression (Mondak, 1993). According to Chilton (2004), candidates create relationships with their opponents through this discourse (p. 201). Chilton (2004) argued that messages are b e structured to symbolize group identities, promote certain representations, and define political associations. Similarly, Gamson (1992) claimed (p. 24). These messages can affect the nature and structure of attitudes. Political persuasion that reinforces political attitudes may be the most powerful because it has the potential to mobilize vested individuals. According to Campbell et al. (1960), attitude attributes include evaluative expressions and expressions associating a political policy with enhancing or denying a value or values. Similarly, emphasizing one political value (Brock et al., 2005; Campbell et al., 1960; Gusfield, 1962; Smith, 20 03) and linking self interest to the value and the associated political issue are manifestations of attitude intensity. Intensity is also expressed by associating compromise or negotiation with dishonor (Gusfield, 1962, p. 22), denying the legitimacy of a political institution (Gusfield, 1962 p. 23), closing off the deliberation process to individuals with other political views (Gusfield, 1962, p. 23) and proposing an end to political institutions, processes and procedures (Gusfield, 1962, p. 23). Other a ttributes include the oversimplification of complex or ambiguous issues including a single, narrow explanation for the current political climate (Brock, Huglen, Klumpp and Howell, 2005; Smith 2003). Additionally, Campbell et al. (1960) and Binder et al. ( 2009), a perceived stake in an issue or the political values salient to that issue, is
28 likely to spur attitude expressions. These statements indicate the dialogue, strategy and content pattern related to political extremism. On Twitter users create rel ationships among their attitudes and opposing viewpoints through communications, incorporating other media content and following other users ( Binder et al., 2009 ). For politically attentive users, these relationships may have a persuasive effect based on communicated message reflects or references. The present study focused on the persuasive strategies expressed in Twitter messages, regardless of whether the ocate on behalf of the expressed position Behavioral Intent Several studies have analyzed how the confluence of political identity, attitude attributes and political information influences and guides behavior (Binder et al., 2009; Campb ell et al., 1960; G oren 2005). Since its creations, Twitter has afforded users an additional avenue for political behavior. Political, communications and technology research has analyzed user behavior and its relation to poltical attitudes ( Bode, Hanna, Sayre, Yang & Shah 2011; Tumasjan, Sprenger, Sander and Welpe, 2010). On Twitter citizens can engage with elected officials, disseminate information and molibize groups for a cause (Bode et al., 2011 ; Hendricks & Denton, 2010 ). Previous research has found that analyzi ng attitudes, persuasion and behavior identifies how coalitions build support within constituencies, steps that are crucial to maneuvering public opinion to achieve a political outcome (Bode et al., 2011; King, Orlando & Sparks, 2012; Tumasjan et al., 2010 ).
29 Tumasjan et al. (2010) used a content analysis to determine whether attitude expressions in tweets reflected the political climate offline. This study found that Twitter time in ( Tumasjan et al., 2010 p. p.184 ). Hendricks and Denton word about offline activities such as speeches, rallies and events. The authors wrote the book, Communicator in Chief has used online media, including Twitter to mobilize political behavior outside of a campaign context. The present study has conceptualized behavior as future action or encouragement for others to act. These conceptions stem from Twitter strategic mobilization tool during natual or manmade disasters, political revolutions and participation in offline events (Hendricks & Denton, 2010). To better understand whether Twitter users are truly political elites and engage in participatory activities, the present study analyzed behavior outside these specific contexts. The present study will code for mentions of political action, intended action or encouragement for others to take action in tweets. The Influence of Social Networking Sites on Polarization According to Zaller (1992), it is when individuals apply political information that they update political attitudes. This application takes shape in the form o f participation (Zaller, 1992). whether an attitude is altered (Zaller, 1992). Zaller (1992) contends if political elites disagree, a polarization effect ensues and politically aw are individuals will update their original attitudes along partisan and ideological lines. Fiorina Abrams and Pope (2006)
30 described a similar process when he referenced sorting. Sorting is a phenomenon that has occurred over time making it more likely t the associated ideology (Fiorina et al. 2006). However, these measures indicate al elites across the ele ctorate (Fiorina et al., 2006, p. 69). Popular press claimed passionate floor speeches and appeals targeting the public that capitalize on fear are an indication of the wides pread polarization of America. Several studies have examin ed how Social Networking Sites (SNS) engage users in both sociological and psychological processes in a manner that fosters political polarization Conover et al. (2011) analyzed the 2010 midterm elections to understand whether Twitter contributed to pol arization online. The authors determined Twitter comprised of tweets collected six weeks prior to Election Day, a time when political cues are highly visible. However, these r esults may reflect the context of the political study The present study offers a unique context to study political attitudes and their influence on polarization. Unli ke previous research, the present study analyzed whether Twitter spurs polarization by examing content during national debate about a salient political issue and coded for attitude attributes that contribute to polarization RQ2: Are extremity, persuas ive strategies or behavioral intent expressed in Twitter messages?
31 Attitude Strength in Online Political Information On Twitter users can create relationships by linking content through messages. The present study examined content linked in Twitter m essages. Linked content is the hyperlinks incorporated in Twitter messages refering to online media. An example of a Republican stubbornness on debt negotiations will damage economy | The Progressive: http://t.co/BloQrpq (PeopleBrowsr, 2011, July 30). This content However, its presence forms a picture that may influence the attitudes of other users by incorpora ting linked content to support their attitudes, enhance credibility, appeal to emotions or invoke fear in others. This process is what Moir (2011) identifies as the m (Moir, 2011 p. 246). This process involves political realities constructed by the media and shared with others (Mo ir, 2011, p. 246). Iyengar and Hahn (2009) claimed m edia choice reflects collective attitudes, particularly attitudes of politically aware and engaged groups. Previous studies have revealed partisanship influences news choice (Iyengar & Hahn, 2009; Stroud, 2010) Several scholars assert the opportunity fo r selective exposure will encourage Internet users to access information, political or otherwise, that conforms with previously held beliefs and values (Abramowitz & Saunders, 2008; Abramowitz, 2010; Blum, 2011). This type of exposure may lead to the rein forcement of pre existing attitudes, avoidance of information challenging opposite viewpoints, and an increase in political polarization (Abramowitz & Saunders, 2008; Abramowitz, 2010; Blum, 2011). These approaches conceptualized and operationalized polit ical
32 partisanship and polarization by choice of media outlet, evaluations of political leaders, and vote choice (Iyengar & Hahn, 2009; Prior, 2006; Stroud, 2010). Stroud (2010) analyzed selective exposure among partisans and determined their media choice l ed to polarization. In this study polarization was defined as reactions to President Bush and John Kerry in the 2004 National Annenberg Elec tion Survey. Several scholars challenged this conception of polarization because measure issue positions across th e conservative/liberal spectrum, but affective attitudes towards leaders (Fiorina & Abrams 2007; Fiornia & Abrams, 2010). Iyengar and Hahn (2009) argue the opportunity and accessibility to choose among several news sources will lead to further polarizatio n. Cognitive consistency theory provides the theoretical underpinning for selective exposure research (Iyengar & Hahn, 2009). The authors conducted an experiment that prompted participants to indicate which news story they would like to read from the fol lowing sources: Fox News, NPR, CNN, the B According to Iyengar and Hahn (2009), the fact that a Republicans chose Fox News indicates that th e mass electorate is polarized. The Internet offers its users the ability to select conten t from a myriad of media choices, but this does not guarantee a ll choices will reflect political preferences. According to Brundidge (2011 ), Internet users are exposed to information that differs from their preferences in small amounts and often by accide nt. Brundidge (2010) which the structural boundaries of the contemporary public sphere combine with
33 selective exposure processes and individual differences to facilitate opportunities for inadvertent exposure to expressed in politic al information is not the only factor influencing attitudes. According to Edwards (1990), it is not position, but affective appeals that are likely to foster attitude change. Affective content is designed to induce an emotional response, an attribute mor e directly related to attitude than cognitive reason ing (Edwards, 1990). Attitude strength is extent to which these attitudes endure and encompasses both emotional and intellectual engagement (Krosnick & Petty, 1995, p. 3). The present study analyzed emo tional and intellectual engagement of attitudes in linked content disseminated in Twitter messages. By analyzing attitude strength in linked content, the present study examined information that appealed to attitudes emotionally and cognitively. The pre sent study conceptualized attitude strength as using previous research on intensity and certainty ence in their attitude being correct is certainty (Krosnick et al., 1993). Gross, Holtz, and Miller (1995) argued, study described whether linked content structure d using attitude strength was attractive to politically aware and interested Twitter users during the debt ceiling negotiations. RQ3 : What is the role of attitude strength in linked content of tweets in the spread of political information across Twitte r ?
34 Issues as Political Cues The present study examined the use of political issues in Twitter messages. According to Fiorina et al. (2006), issues encouraged political participation, particularly discussion. To understand the dynamics of political disc ussion, Gamson (1992) argued discourse should be studied with in a specific context. public discourse says about an issue, since it is a central part of the reality in which (Gamson, 1992 p. 27). Additionally, by analyzing elements of political discussions, research identifies political cues derived from predispositions or designed to motivate behavior. As previously mentioned, attitudes about issues are shaped by political p redispositions, particularly political party and ideology. This structure creates a foundation for transforming policy issues into political cues. Benoit and Wells (1996) observed this connection is so strong that issue positions influenced voter percept ions e ffect of the campaign communication setting. Issues develop a personal tone in an effort to communicate these cues w ith potential voters (Kaid, 2004 ). According to Goren, Federico and K ittilson (2009), the impact of these and other cues on attitudes may result in a rise in political extremism. According to Goren et al. (2009), these factors prime individuals to express ideological values that align with political identification s often a process that leads to extremity. This study analyzed the relationship between issues mentioned in tweets, attitude attributes and linked content that expressed attitude strength to understand how policy issues may influence the polarization process onlin e. RQ 4 : What is the relationship between issues and extremity in tweets and between issues and attitude strength in content linked in Twitter messages?
35 CHAPTER 3 METHODS The present study employed a quantitative content analysis, that is, the systemat ic measurement of specified characteristics (Berelson, 1952). It also systematically summarized the partisanship, ideology, extremity, persuasion strategies, behavioral intent and attitude strength in linked content disseminated across Twitter by the use rs of the social networking site. The present study was completed using manual and computer assisted coding instruments. Sample and Units of Analysis The researcher collected 270 Twitter messages with working hyperlinks from a nine day period starting J uly 25, 2011, and ending August 2, 2011. The units of analyses were the Twitter through each tweet. The sample was collected from PeopleBrowsr a social analytics platform with access to the Twitter firehose, a datafeed of all Twitter messages (www.peoplebrowsr.com). Using the PeopleBrowsr grid platform, the researcher collected a purposeful sample of 500 Twitter messages from each day that contained Twitter messages were published by users in the United States identified to members of the PeopleBrowsr Twitter users based on self identified interests, hashtags, previous tweets and hy perlinks (Personal Communicati on with Jonathan Craft, 2012). To understand if these users could be considered political elites, the present study examined how Twitter users identified their political preferences in biographies. To better understand the salience of
36 identifications, the present study examined twee ts for mentions of a political party or an ideology. The researcher collected a purposeful sample of tweets that contained a working hyperlink to online content that contained text. As the sample was collected, tweets from news media organizations or without hyperlinks were eliminated leaving an average of 134 tweets per day. A random sample of Twitter messages was selected for each day during the nine day period using a randomization feature on Microsoft Excel. After randomization, two tweets were eliminated because the linked content contained videos without text; eight tweets were eliminated because they were news organizations or media personalities; six tweets were eliminated because they w ere duplicate tweets, and 31 tweets were removed that contained non working links were removed from the sample. Removed tweets were replaced with the next random tweet that met the criteria for that day. The total sample size was 270. Each selection inc luded a tweet, biography and the content linked through Twitter messages. Coding Instrument Independent variables included a Twitter content linked through his or her tweet. The dependent variables include measures of extremity, behavioral intent expressed in Twitter messages, persuasion strategies expressed in Twitter messages and attitude strength in content linked through tweets. As previously mentioned, the present study relied on both human coders and a compute r assisted content analysis program. Therefore, two distinct measurement instruments were required for this project. For manual coding, the researcher conceptualized and operationalized variables in a codebook and provided coders with a coding form.
37 The second coding instrument was the textual analysis program, Linguistic Inquiry Word Count 2007 (LIWC2007). LIWC2007 relies on 80 language categories. These or emotiona l state from written or spoken language (Tausczik & Pennebaker, 2010). The construction of these categories is based on of the evaluations of three different groups of judges chosen from LIWC2007 who evaluated categories derived from dictionaries, thesauru ses, questionnaires and lists consisting of research assistants (Tausczik & Pennebaker, 2010, p. 27). To be included in a category, two out of three judges had to agree on its inclusion or exclusions from the list. These trials were completed independent ly and repeated by a separate group of judges. The second rating phase produced between 93% and 100% agreement on all categories. Measuring Independent Variables Twitter users identifi ed themselves by incorporating political party or ideology. The categories for partisan identifications were Democrat, Republican, Tea Party and Independent. The content categories for ideology were degrees of progressive, liberal, conservative and moder ate self intercoder reliability for these variables is summarized in Table 3 1. Twitter messages were coded manually to determine the presence or absence of a dominant political party, dominant ideology and issue mentions. The dominant political party and ideology mentioned in each tweet were defined as the first mention in each tweet and manually coded. The issue categories were determined by previous political communications researc the national economy and federal government. Issues categories included
38 unemployment, the cost of living, taxes, recession/depression, social security, welfare, Medicare, dissatisfaction with governm ent, healthcare and poverty. During intercoder reliability tests, the welfare category was not observed and was removed. The intercoder reliability for these variables is summarized in Table 3 2. Measuring Dependent Variables The dependent variables code d using manual coders were expressions of extremity persuasion strategies and behavioral intent in Twitter messages. These variables were coded manually to ensure the stated op inion, not its presentation, and were accurately measured according to the de finition put forth by Krosnick et al., (1993). According to Krosnick et. al (1993) includes the following continuums: fa vorable unfavorable, good bad, foolish wise, and harmful beneficial (p. 1132). This measure determined whether or not tweets expressed an extreme position in regard to the debt ceiling debate. Persuasion strategies were operationalized using previous stud lysis of language and extremity. Persuasion strategies were coded manually due to their specifications designed by the researcher and the nature of the content. Persuasion strategies included associating compromise with failure (Campbell et al., 1960; Gusfield, 1962, p. 22), denying the legitimacy of a political institution (Gusfield, 1962, p.23), closing off deliberation to other political viewpoints (Gusfield, 1962, p.23), proposing an end to political institutions (Gusfield, 1962, p.23), attrib uting one factor for causing current conditions (Brock et al., 2005, p. 102; Smith, 2003), emphasizing a single value such as mor alism, self reliance, equality (Brock et al., 2005 p. 102; Campbell et. al, 1960; Gusfield, 1962 p. 22; Smith, 2003) and expr
39 all at stake (Binder et al., 2009, p. 317). Derailing democratic deliberation, overly expressing sentiment and symbolism and advocating for the overthrow of government entities are also predictors of political extr emity (Brock et al., 2005; Gusfield, 1962). Due to the difficulty in identification in a tweet, this was operationalized by inclusion either in closing off deliberation to other viewpoints or proposing an end to political institutions. However, whether t hese indicators appear in t weets has not been examined. The present study fills this gap in politi cal communication scholarship. rrent or future plans. Behavioral viewing a press conference or engaging in some oth er form of political participation Political participation in both digital and offline spaces was analyzed in the present study. identify and measure this concept easily and effec tively. Table 3 2 summarizes the intercoder reliability for categories in Twitter messages. The present study questioned whether Twitter messages and content linked through tweets expressed attitude strength. The lack of theoretical and practical clari ty distinguishing affect and certainty makes it difficult for human coders to capture without inducing political bias. The computer assisted content analysis program LIWC2007, was designed to capture both of these variables and allows for t hese variables to be combined. Furthermore, it was developed to study emotional, cognitive and structural components in verbal and written language (LIWC2007 Manual). This program offers
40 an attractive solution to defining and measuring attitude strength, specifically as it relates to affect and certainty. To minimize the bias of measuring attitude strength, this was done using LIWC2007. gth of emotional reaction provoked by the present study, LIWC2007 assessed attitude strength by measuring certainty and affective processes in content linked through a tweet. The variable certainty consists of Pennebaker, 2010). In LIWC2007, certainty is a part of the larger category of cognitive processes. The category of affective processes combines positive emotion and negative emotion. Negative emotion consists of the smaller variables anxiety, anger and sadness. The statistical softwa re program, Statistical Package for the Social Sciences (SPSS), was used to examine the relationship between affect and certainty in the present study. These two variables were found to have a strong correction ( r = .74). Coding Procedure Coder Training a nd Pretest Twitter biographies and messages were manually coded at separate times from one another. Therefore, the present study required two separate training sessions and intercoder reliability tests. For each session, two coders were trained in a 45 minute session. During these sessions, the researcher explained the coding scheme and operationalized each variable. After coders became familiar with the schemes, a pretest was conducted. Using a coding sheet, coders analyzed Twitter biographies
41 and m essages for clarification of the coding procedure and agreement. For biographies, the process required 5 sample units and 10 units for messages. Following the training, coders received a randomly selected sample set for intercoder reliability testing. Du which were coded during reliability testing, were incorporated into the full sample. According to Lombard, Synder Duch and Bracken (2010), this sample can be incorporated into utilized to solve disagreements and allow the reliability sample to be incorporated into the full sample set. The following section provides a more detailed description of the coding process for each unit of analysis: Twitter biographies, messages and linked content. Coder Independence To maintain objectivity, coders completed their analysis separately and coded separately. All Twitter biographies were coded at the same time and independent of one another as were all Twitter messages. Intercoder Reliability Coders analyze d 10% or 27 units that were randomly selected from the present levels usin intercoder reliability, coder disagreements were assessed and resolved. As necessary,
42 the researcher made alterations to the coding guide. Subsequently, coders independently and s eparately coded an additional 10% of the sample. At this point, adequate intercoder reliability levels were achieved for variables measured in each unit of analysis. Biographies Initial intercoder reliability was weak due to inaccurate measurement levels and coding categories, which were not mutually exclusive. The initial scheme required coders to account for the presence or absence of party and ideological identification and presence or absence measurement levels in a covariant set. For example, if coders seldom identification manifests itself, intercoder reliability will be lower when usin alpha. very liberal, libertarian, social conservative and extrem ely conservative. These identifications were pared down to conservative, moderate, liberal and progressive. Following these amendments to the coding guide, intercoder reliability was strong f or gender, partisanship and ideology. Intercoder rel iability is reported in Table 3 1. Twitter m essages After the initial 10% of tweets were coded, intercoder reliability was strong for a majority of the variables. However, after coding an additional 10 % of the sample size, weaker intercoder reliability demonstrated several limitations with the coding scheme for
43 Twitter regarding extremism, issues and valence toward issues mult iplied coding errors. Additionally, the focus of each tweet is the debt debate. Therefore, requiring coders to indicate a separate focus or secondary topic was superfluous, particularly due to the lack of specificity inherent in Twitter ch aracter limitation. After secondary topics were eliminated, intercoder reliability improved using the adjusted coding scheme. Reliability indices are reported in Table 3 2. Linked c ontent Computer assisted programs cannot approach reliability measurement s in a similar manner as other assisted content analysis methods can. For the textual analysis program LIWC2007, developers use a psychometrically validated internal dictionary program consisting of 80 different word categories. The LIWC2007 determines t he rate at which words belong to psychological, affective, cognitive or structural categories ( Tausczik, Y. & Pennebaker, J., 2009) The statistical software program SPPS was used to calculate the internal consistency of the two variables affect and certa inty. emotion includes anxiety, anger and sadness. These variables, combined with these six items were .57. While the reliability is lower than ideal, Tumasjan, A., Sprenger, T., Sander, P., and Welpe, I. (2010) used LIWC2007 to successfully predict an election outcome. Tumasjan et al. (2010) analyzed political sentiment online using LIWC2007 Twitter can be seen a s a valid real time indicator of political s they should be examined together, albeit with caution.
44 Data Analysis After intercod er reliability was achieved, attributes were collapsed into 12 variables for data analysis. Coding variables include d user party identification, user ideology identification, user gender identification, type of post, issue mention, dominant party mention, dominant ideology mention, extremity, behavioral intent, persuasion strategy, hashtag and attitude strength in linked content. Data analysis was conducted using SPSS. Chi Square and independent sample t tests were used to examine the present earch questions.
45 Table 3 1. Intercoder reliability indices for Twitter user b iographies Variable Reliability Political party Political ideology Gender 0.88 0.87 0.76 address, were not observed in Twitter messages. It is important to note the absence of these variables despi te their exclusion in the present study. Table 3 2 Intercoder r eliability i ndices for Twitter m essages Variable Reliabili ty Tweets: Type of p ost: Retweet 1 Tweets: Type of post: Reply or m ention 0.85 Tweets: Issues: Unemployment/ l ack of j obs 1 Tweets: Issues: Cost of l iving 1 Tweets: Issues: Taxes 1 Tweets: Issues: Recession/ d epression 1 Tweets: Issues: Social Securit y 1 Tweets: Issues: Medicare 1 Tweets: Issues: Dissatisfaction with g overnment 0.82 Tweets: Issues: Healthcare 1 Tweets: Issues: Poverty 1 Tweets: Dominant p arty 0.91 Tweets: Valence toward p arty Tweets: Ideology 0.95 1 Tweets: Valence toward i ndivi dual 1 Tweets: Valence toward e mphasized i ssue 0.95 Tweets: Extremity 0.86 Tweets: Call to a ction 1 Tweets: Associates negotiation or compromise with failure 0.84 Tweets: Denies the legitimacy of political institution 0.8 Tweets: Closes off deliberat ion process to other political viewpoints 0.85 Tweets: Proposes end to political institutions, processes and procedures 0.79 Tweets: Attributes one factor for causing current conditions 1 Tweets: Emphasis on single value such as moralism, self reliance equality, etc. 0.79 Tweets: Expresses self interest is at stake 1 Tweets: Hashtags : Does the story mention another user or subject using #? 0.90 Tweets: Top progressives : #topprogs 1 Tweets: P rogressive : #p2/#p2b 1 Tweets: T op conservative : #tcot 1 Tweets: Organized Conservative Resistance Alliance: #ocra 1
46 CHAPTER 4 RESULTS The present study analyzed t hree separate units of analysis: user biographies user tweets and c ontent linked in tweets that consisted of 12 different variables. Table 4 1 summarizes users sel f identifications of political party, ideology and gender i n Twitter biographies. Table 4 2 summarizes the findings of six categories analyzed in tweets: issues, political party, ideology, extremity, behavioral intent and persuasion strategie s. For link ed content, Table 4 3 summarizes findings for the remaining category, attitude strength. The following sections will address each research question by outlining variable construction based on will also report the descriptive statistics of the study sample. The first research question analyzed whether Twitter users interested in politics describe themselves us ing partisan or ideolog ical terms in Twitter biographies. Coders found that 8.5% ( n = descriptions contained a political party. Political party was defined as biographies that identifi ed the user as a member of the Democratic, Republican or Tea P arty as well as indicating the absence of par tisanship by identifying as an I ndependent. Ideol ogy categories included progressive, liberal, moderate or conservative. Coders found that 24.1% ( n = 65) of Twitter users identified the mselves incorporating ideology. To compare the relationship between party identification and ideology among Twitter users, a Chi square analysis was used. The results were not significant ( 2 = .557, p =.456) Users interested in politics are not more likely to describe themselves using one identity instead of another in Twitter biographies
47 Additionally, t he second part of this research question examined whether tweets mention a dominant political party or ideology. A dominant political party was mentioned in 51.9% ( n = 140) and a dominant ideology was mentioned in 4.8% ( n = 13) of the 270 tweet sample s. Due to the low cell counts within the ideology category, no analysis could be conducted. Table 4 4 summarizes these results. The second research question explored expressions of extremity, behavioral intent or persuasive strategies in tweets Extremity was expresse d in 30.7% ( n = 83) Twitter Twitter 4.8% ( n = 13) of tweets and enc ouraged other users to watc h a p residential address, contact their U.S. congressional representatives or engage in a political activity. Persuasion strategies included were observed in 22.2% ( n = 60) of tweets. Three chi square analyses were conducted comparing extremity versus per suasion strategies, extremity versus behavioral intent and behavioral intent versus persuasion strategies. The results for these tes ts are summarized in Tables 4 5, Table 4 6 and Table 4 7. For behavioral intent, no significant relationships occurred bet ween extr emity and persuasive strategies. For tweets that express ed extremism, these messages were significantly more likely to use persua sive strategies than tweets, which d id not express extremity ( 2 = 38.489, df = 1, p < .001). The third question an alyzed the role of attitude strength in linked content in the spread of political information across Twitter An independent sample t test was conducted to compare attitude strength in linked content in tweets that expressed extremity, persuasive strateg ies and behavioral intent. These findings are presented in
48 Table 4 8. Significant differences emerged for two of these conditions: Tweets that expressed extremity ( M = 5.81, SD = 3.15) were significantly more likely than tweets that did not express extr emism ( M = 4.81, SD = 2.18) to link to content that expressed attitude strength, t (268) = 3.04, p = .003 (equal variances assumed). Twitter messages which incorporated persuasion strategies ( M = 5.8, SD = 3.44), were significantly more likely to link to content that expressed attitude strength than tweets that did not use persuasion strategies ( M = 4.91, SD = 2.21); t ( 73) = 2.005, p =.049 (equal variances not assumed). The fourt h research question examined issue mentions in tweets. Coders catalogued the mention of 10 issues that were identified as salient to the debt debate and prevalent in previous political communication studies (Banwart & McKinney, 2003). Unemployment/lack of jobs, cost of living, taxes, recession/depression, social security, Medicar e, dissatisfaction with government, healthcare and poverty were mentioned in 27.8% ( n = 75) tweets. For tweets that expressed extremism, it is significantly more likely that these tweets mentioned an issue ( 2 = 72.645, df = 1, p < .001) than those withou t extremism. Also, Twitter messages, which mentioned issues ( M = 5.83, SD = 3.26) were significantly more likely to link to content that expresses attitude strength than tweets that did not mention issues ( M = 2.83, SD = 2.18) in tweets; t (268) = 2.901, p =.004. See Tables 4 9 and 4 10 for a summary of these results.
49 Tabl e 4 2. Frequencies for Twitter m essages Table 4 1. Frequencies in Twitter user b iographies Variable Count Percentage Political party 23 8.5 Ideology 65 24.1 Gender 120 44.4 Male 87 32.2 Female 33 12.2 Variable Count Percentage Issues 75 27.8 Unemployment/lack of j obs 1 0.4 Cost of l iving 3 1.1 Taxes 9 3.3 Recession/d epression 5 1.9 Social Security 2 0.7 Medicare 1 0.4 Dissatisfaction with g overnment 57 21.1 Healthcare 1 0.4 Poverty 4 1.5 Political party 140 51.9 Ideology 13 4.8 Extremity 83 30.7 Behavioral intent 13 4.8 Persuasion strategies 60 22.2 Associates compromise with failure 9 3.3 Denies the legitimacy of a political institution 24 8.9 Closes off deliberation process to other viewpoints 13 4.8 Proposes an end to a political institution 2 0.7 Attributes one factor to causing current conditions 16 5.9 Emphasis on one single value 2 0.7 Expresses self interest is at stake 4 1.5
50 Table 4 3. Frequencies for attitude strength in linked c ontent Variable Mean Standard d eviation Attitude s trength 5.12 2.56 Affect 4.18 2.36 Positive emo tion 2.3 1.32 Negative emotion 1.82 1.74 Anxiety 0 .30 0 .40 Anger 0 .69 0 .84 Sadness 0 .46 0 .94 Certainty .93 .71 of positive emotion, negative emotion, anxiety, anger and sadness. Table 4 4 Comparison of p olitical p arty and i deology m entions in Tweets Ideology m ention No ideology m ention Total Political p arty Yes Count 11 129 140 % 7.9% 92.1% 100% No Count 2 128 130 % 1.5% 98.5% 100% Total Count 13 257 270 % 4.8% 95.2% 100% Table 4 5 Comparisons of e xtremity and p ersuasive s trategies Persuasion s trategies No persuasion s trategies Total Extremity Yes Count 38 45 83 % 45.8% 54.2% 100% No Count 22 165 187 % 11.8% 88.2% 100% Total Count 60 210 270 % 22.2% 71.8% 100% 2 (1, N = 270) = 38.489 (p < .001)
51 Table 4 6. Extremity and behavioral i ntent Behavioral i ntent No b ehavioral i ntent Total Extremi ty Yes Count 6 77 83 % 7.2% 92.8% 100% No Count 7 180 187 % 3.7% 96.3% 100% Total Count 13 257 270 % 4.8% 95.2% 100% Table 4 7 Persua sion strategies and behavioral i ntent Behavioral i ntent No b ehavi oral i ntent Total Persuasion s trategies Yes Count 4 56 83 % 6.7% 93.3% 100% No Count 9 201 187 % 4.3% 95.7% 100% Total Count 13 257 270 % 4.8% 95.2% 100% Table 4 8 Attitude s trength in l inked c ontent found in t weets that express e xtrem i ty, persuasion strategies and b ehavioral i ntent Attitude s trength N Mean Standard d eviation t df p Extremity Yes 83 5.81 3.15 3.04 268 .003* No 187 4.81 2.18 Persuasion s trategies Yes 60 5.85 3.44 2.005 73.373 .049* No 210 4.91 2.2 1 Behavioral i ntent Yes 13 5.79 2.31 .972 268 .332 No 257 5.08 2.57 t(73.373) = 2.005, p =.049; t(268) = 3.039, p = .003
52 Table 4 9. Comparing e xtremity and i ssue m ention in t weets Issue m ention No i ssue m ention Total Extremity Yes Cou nt 52 31 83 % 62.7% 37.3% 100% No Count 23 164 187 % 12.3% 87.7% 100% Total Count 75 195 270 % 27.8% 72.2% 100% 2 (1, N = 270) = 72.645 (p < .001). Table 4 10 T t est a ttitude s trength and i ssue m ention Attitude s trength N Mean Standard d eviation t df p Issue m ention Yes 75 5.83 3.26 2.901 268 .004* No 195 4.84 2.18 t(268) = 2.901, p =.004
53 CHAPTE R 5 DISCUSSION AND CONCL USION The present study found a strong nexus between extremity, persuasion strategies, iss ue mentions and attitude strength. Tweets that expressed extreme positions were more likely to mention issues and incorporate persuasion strategies than non extreme tweets. Also, tweets that linked to online content, which expressed attitude strength we re more likely to incorporate persuasion strategies and issues than tweets that linked to content not expressing attitude strength. Finally, tweets with extreme political attitudes were linked to content that expressed attitude strength more often than Tw itter messages that did not express extremity. This chapter describes these relationships and discusses how they impact the polarization process across Twitter Extremity in Twitter Messages One sed extreme attitudes in regards to the ongoing negotiations. This is slightly higher than the estimated number of Americans who hold extreme viewpoints offline (Fiorina, et al., 2006). This slight increase is not surprising because extremity is linked to political discussion and participation in addition to polarization. These results extend previous studies findings that suggest Twitter blog with the ability to spur polarization (Conover et al., 2011). The present study observed tweets tha t expressed extreme attitudes were significantly more likely to incorporate strategies, language and content to support their position. For the polarization process to take effect, extreme attitudes must be maintained or enhanced over time (Abelson, 1995)
54 The present study determined extreme attitudes were at the very least maintained, if not enhanced on Twitter from July 25 through August 2, 2011, regarding the debt ceiling negotiations since there was a slightly higher percentage of extreme tweeters in the sample than the estimated number of extreme Americans generally. According to Binder et al. (2009), deliberation and expression of political preference influence attitude strength, while discussion solidifies preferences and leads to extremism. App lying Twitter messages that express attitudes should encourage extremity. The encouragement of extreme positions fosters and leads to polarization (Abelson, 1995). While the present study was not des igned to determine whether or not expression leads to extremity, it does describe how incorporating political information and persuasion strategies may encourage extreme attitudes on Twitter Extremity and Persuasion Strategies The present study observe d that tweets expressing extreme attitudes advanced their position using persuasive strategies. The present study found that Twitter users who express ed extreme viewpoints were more likely to incorporate a persuasive strategy than users that did not expr ess extremism. These strategies are persuasive because they serve as sound bites, buzzwords or markers that reflect positions along an ideological continuum. Particularly when coupled with extreme positions, these strategies persuade the political middl e to move in either direction, thus organizing the political landscape (Brock, 2005, p.110). According to Brock et al. (2005) analyzing these strategies sheds light on attitude change and the motivations that link political participation to ideology.
55 Pe past and future political behavior. For example, tweets, which deny the legitimacy of a political institution, undermine the ability of pluralist politics to mediate conflicts and favor political participation through extreme channels (Gusfield, 1962, p. 24). According to the blame attribution theory, individuals who attribute one factor for causing a condition are likely to perceive a benefit from acting to change that condit ion (Zuckerman, 1979; Kelley 1980). For users who expressed these tactics, the cost of voting will be a worthwhile investment because perceived different leadership will change the present conditions (Downs, 1958). While the current study does not liken persuasive strategies to voting, it provides a framework for understanding how messages encourage and foster online participation. These persuasion strategies are linked to extremity showing current political attitudes and may serve as indicators of futur e political behavior. Persuasion strategies emerge independently of the current issues of the day and are useful for projecting the future political environment (Brock, 2005, p. 109). If the strategies expressed in the present study are valid indicators, users who expressed persuasion strategies are likely to engage in future political behavior. The current link to information that expressed attitude strength. If a campa ign is looking to mobilize its base online, political messages should target extreme users and those incorporat ing persuasive strategies with content that express es attitude strength. These extreme, persuasive users will further the reach of these message s by disseminating this information across Twitter
56 Extremity and Issue Mention mentions and extreme attitudes in tweets. The following tweets exemplify this relationship : madness was always about killing social s ecurity scam, means more t US debt gridlock roils markets as stocks fall and gold touches record high These messages reveal considerations connected to the current salient issue, the deb t ceiling debate. Whether media generated or a true with that Twitter user and potentially with followers as well. Alternatively, Twitter users interested in politics may perceive the debt ceiling debate as an opportunity for online political participation, particularly discussing an issue of their choice. The affordance theory, which considers how the environment, ehavior, provides the framework for analyzing this situation. Twitter users perceived the debt debate as an opportunity to express their position or increase its saliency among the mass public. For one business, the debt debate climate afforded social m edia marketing opportunities. risi s impacts homeowners and buyers For political communications research, these findings show that Twitter users interested in politics often combine extreme attitudes with the use of cur rent issues to make their point. For communications practitioners, the present study supports a targeting strategies. Campaigns have the opportunity to s issue preference mining
57 technology continues to emerge, campaigns can effectively target Twitter users who express extreme attitudes and cite current issues to appeal to th ese users. Attitude Strength in Linked Content The present study confirmed a strong relationship among linked content that expressed attitude strength and tweets that expressed extreme attitudes, persuasion strategies and mentioned issues. Findings were minimally significant for persuasion strategies. Content that expressed attitude strength asserted positions using emotion laden words. These results indicated that the dynamics of the affective online content likely shape the flow of information across Twitter By analyzing political information linked through tweets, the present study provides insight into levels of attitude strength that resonate with users. and intensi ty as related concepts. Attitude strength indicates how intensely the stated opinion is held. According to Edwards (1990), affective attitude dimensions, including intensity, provoke biased information processing and are related to certain attributes of an attitude object. Applying, this theory to the present study, Twitter users with extreme attitudes may have been attracted to political information that expressed intensity because it complemented their position and confirmed their predispositions. Fo r political communications scholars, this relationship may indicate how Twitter users attitudes are formed and the likelihood they may change in the future. For users who linked content that expressed attitude strength, its affective nature indicates use r attitudes were formed under similar conditions (Edwards, 1990). Political strategists could use this knowledge to develop affective rhetoric and target content to Twitter users who express extreme attitudes. These users could further disseminate this
58 information across Twitter change. By incorporating content generated by a third party, users have the ability to connect online content to a particular issue. This connection may serve to enhan ce the credibility of a position, to encourage engagement in an issue discussion or to update attitudes about an issue. For political communications scholars, issues reveal Twitter on is processed. For campaigns looking to change issue dialogue, these results can provide insight into what type of content will gain traction online. Partisanship and ideology were not a statistically significant subset of this study sample to determi ne whether these identifications were related to or influenced political information on Twitter The lack of connection between partisanship and information roughly t he same period of this study, the Pew Research Center found that 72% of survey respondents reacted negatively toward the budget deal and these views crossed both party and ideological lines (Pew Research Center, August 1, 2011). During a political campaig n, when voters must choose to vote for a Democratic or Republican candidate, partisan cues and symbols may be more readily apparent. Individuals are better primed to invoke partisanship when updating their political opinions. Campaign strategists may wan t to further study the connection between extremity, issue mention and partisanship within a campaign context. The majority of the mass public is not extreme, however most individuals disagree with their party on at least one issue and one third disagree on a salient issue (Hillygus & Shields, 2008, p. 79). As previously
59 mentioned, partisans who feel strongly about an issue, but who disagree with their policy preferences. Limitations and Future Directions It is important to address several limitations of this descriptive research. In order for PeopleBrowsr to comply with Twitter policies, data extraction was limited to the last 500 tweets within a given search criteria. Therefore, the generalizability of this However, the information flow across Twitter is repetitive and the sample was purposefully chosen to obtain the last informatio n of that day. Secondly, the reliability the results relating to that variable need to be interpreted with caution. Additionally, tweets may contain or reproduce th e headline of linked content when Twitter feed as opposed to solely user generated content. A media generated headline or other forms of automated content do not reflect or attitudes. However, the focus of this study was to analyze the flow of political information, as related to attitude strength and extremity, and whether or not that political information incorporated persuasion strategies, party, ideology and issues in tweets. As a result, this limitation should have had a minimal effect on this analysis. Finally, this study looked at only one topic, the debt debate negotiations. Future studies should consider a number of issues, time per iods and political contexts. Despite these limitations, this study provided both political communication scholars and practitioners with a method for analyzing the flow of political information across
60 Twitter ht into how tweeting linked content may influence the polarization process online. Future research should work to develop and test methods for measuring extremity and attitude strength online, accounting for Twitter character limit. Twitter is a medium that is growing exponentially in its political mobilization potential, so more research is needed to learn about how users employ these key concepts. Additionally, future research should discern differences in the spread of political information b y politicians, the media and other Twitter users. These findings show that more research needs to be done to understand how information influences online political discussions. Extending this study to other topics, such as following an article tweeted b y the media, politicians and interest groups, or analyzing how users incorporate linked content when messaging elected officials, or studying hashtags, as they are developed or discussed in online content, would help flesh out these ideas. Further research into the segmentation and targeting of different publics on Twitter may prove very useful for both the political and commercial arenas. Similar to strategies used in targeted mailing and door to door campaigns, voters or potential customers can be ident ified using multi modal strategies. For example, new methods of opinon mining can segment Twitter users based on followers, interests, location and lingustics, which LIWC2007 analyzes to understand a users cognitive and physiological processes. For busi ness application, this data identifies audiences for target specified advertisements or purchase specific articles and materials. In terms of practical applications, this specialized search identifies potential voters and can determine wedge issues within a group or political party. These data can be used to reach out to
61 individual Twitter users and their followers. Often, Twitter users look to those they follow for information they find relevant, information that may be used to update their own politic al attitudes. The strength of this application as opposed to other similar research is the ability to analyze political attitudes and their attributes using Twitter messages. Unlike survey research, Tweets are organic self identified statements whethe r user generated or retweets unaffected by the atmosphere of a polling environment. Twitter users are operating under their natural conditions, a situation that provides a better indication of thod stems from the limited number of characters, often leading to difficulty for coding and interpretation. With the 140 character limitation, it is difficult to differentiate agreement, ambivalence, preferences and attitudes. This has generally been a concern for research in the public opinion and political communications fields (Zaller, 1992). However, by analyzing tweets, research is capturing how citizens use information to form preferences and attitudes as this process unfolds and in real time. Conclusions Rather than focusing on political polarization, political communication scholars and strategists should examine attitudes expressed across Twitter Various approaches have been made to analyze political opinions and attitudes online. Hashta gs, followers, and language have been analyzed to gauge extremity and polarization of political opinions or attitudes across Twitter Little research has considered how Twitter users impact the flow of political information, which is a key component in understanding political attitudes and the potential for a polarization effect.
62 This study explored attitude strength in linked content as well as expressions of extremity, persuasion strategies, issues and behavioral intent in tweets. This study has dev eloped a straightforward method for identifying and segmenting Twitter users, based on the information that resonates with them and encourages political action. As previous research has demonstrated, the most likely participators are those who intensely hold extreme opinions. Persuasion strategies and issue mentions relating to extreme positions were observed with both extremity and attitude intensity. A campaign can use this knowledge to introduce issues that will spark political dialogue and invoke en gagement, thus motivating political behavior. Campaigns can also mobilize online engagement by identifying which information incites extreme points, persuasion strategies and attitude strength. This can be used for online campaign strategies in the develo pment of communication materials that engage political participation online.
63 Figure 5 1. Extremity and attitude strength across Twitter This figure compares the relationships between extremity and issue mentions, extremity and persuasion strategies attitude strength and issue mentions, attitude strength and persuasion strategies, and attitude strength and extremity. The dark blue persuasion strategies, and extremity) i n tweets also with the presence of either extremity or linked content with attitude strength as referenced on the middle categories in either non extreme tweets or those linked to content lacking attitude strength. The length of the line and percentage indicate the corresponding presence of the middle categories. For example, the first row illustrates the statistically significant finding that issue mentions were present in 62 .7% of extreme tweets while only present in 12.3% of non extreme tweets. Table 5 1. Critical views cross partisan and ideological lines Tone of one word response (%) Negative Neutral Positive No Answer Total 72 11 2 15 Republican 75 9 2 14 Democra t 72 13 4 12 Independent 72 11 1 16 Among Rep/lean R Tea Party 83 10 2 6 Not Tea 69 10 2 19 Pew Research Center/Washington Post July 28 31, 2011 58.7% 58.4% 58.6% 45.8% 62.7% 44.3% 46.2% 44.9% 11.8% 12.3% 10% 20% 30% 40% 50% 60% 70% Extremity Persuausion strategies Issue mention Persuausion strategies Issue mention Attitude Strength Extremity No Yes
64 APPENDIX CODING GUIDE Var Definitions Coding Example Case n umber : U nique number identifying eac h tweet 101, 102, 103, etc. 1 Coder ID: N ame 2 Biography: User p arty i dentification : Whether him or herself by incorporating a political party (Choose 1. If more than one listed, choose the first one list ed in each tweet.) 1. Democrat 2. Republican 3. Tea Party 4. Independent 99. N/A I'm a writer that enjoys blogging, politics, hard news, technology, and writing for other folks. I like to blog and consider 3 Biography: User i deological i dentification : describes him or herself by inco rporating ideological positions (Choose 1. If more than one listed, choose the first one listed in each tweet.) 1. Progressive 2. Liberal (left, left le aning) 3. Moderate (center, middle) 4. Conservative (right, right leaning) 99. N/A 1. Progressive 2. Liberal 3. Moderate 4. Conservative 99. N/A science, and liberal politics -ciple of the arts, social media, progressive politics and pop culture. Based in Park Slope, Brooklyn with his wife and 4 B iography: User gender: biography (Choose 1) 1. Male 2. Female 3. Cannot determine
65 Var Definitions Coding Example 5 Tweet: Type of p ost : Retweet or reply Indicate either 0:absent or 1:present Retweet: RT@username; Republishin g Twitter message Reply: @username ; Commenting on another Twit ter message or replying to Retweet: 0 absence, 1 presence Reply: 0 absence, 1 presence Retweets: RT@username Reply: @username 6 Tweets: Issues: C ode for the presence or absence of issues Indicate either 0:absent or 1:present for ea ch issue Unemployment/lack of j obs : M entions unemployment rate, lack of jobs Cost of living : M entions inflation, compares living costs to previous times Taxes : M entions federal income taxes on individuals and/or corporations Recession/depression : S peci fic mention of recession or depression Social Security : M ention of Social S ecurity program 1. Unemployment/lack of j obs 2. Cost of living 3. Taxes 4. Recession/d epression 5. Social Security 6. Medicare 7. Dissatisfaction with g overnment 8. Healthcare 9. Poverty
66 Var Definitions Coding Example 6, cont. Medicare : M ention of Medicare program Dissatisfaction with g overnment : P roblematic issues with government structure, institutions or elected officials Healthcare : In general; cost of healthcare, insura nce, access to healthcare Poverty : Homelessness, hun ger 7 Tweets: Emphasized Issue: If multiple issues are discussed, is one emphasized more than others? Please specify 8 Tweets: Dominant Party: T he political party that is the predo minant focus of th e coded tweet Consider hashtags as well. (Choose 1) 1. Democrat 2. Republican 3. Tea Party 4. Independent 99. N/A 9 Tweets: Valence toward Party: For each, code whether feelings toward attitude object are negative, neutral, positive or N/A 1. Negative 2. Neutral 3. Positive 9 9. N/A 10 Tweets: Political Ideology: Among the political (Choose one; If more than one listed, choose first mentioned.) 1. Progressive 2. Liberal 3. Moderate 4. Conservative 99. N/A 11 Twe ets: People: List all individuals in the
67 order that they appear in the message. Var Definitions Coding Example 12 Tweets: People: Valence toward first individual mentioned: Code whether feelings are negative, neutral or positive toward this person 1. Negative 2. Neutral 3. Positive 99. N/A 13 Tweets: Extremity: Whether one refers to the debt debate overall as good, bad, wise, foolish, beneficial, harmful or whether one favors or opposes the focus (Krosnik et al., 1993) The object is the dominant fo cus of each tweet. For each continuum, select the appropriate code or N/A. 1: Good, wise, beneficial or f avorable 2: Bad, foolish, harmful, o pposed 99: N/A Good b ad: The dominant topic of each tweet ( i ssue/ p olicy, p olitician, p olitical institution, p olitical p arty, s elf i nterest, p olitical e vent or v alues) is viewed as extremely good, extremely bad, or neutral Wise f oolish : The dominant topic of each tweet (i ssue/ p olicy, p olitician, p olitical institu tion, p olitical p arty, s elf i nterest, p olitical e vent or v alues) is viewed as extremely wise, extremely foolish or neutral 1: Good, wise, beneficial or favorable 2: Bad, f oolish, h armful, o pposed 99: N/A
68 Var D efinitions Coding Example 13, cont. Beneficial h armful : The dominant topic of each tweet (i ssue/ p olicy, p olitician, p olitical institution, p olitical p arty, s elf i nterest, p olitical e vent or v alues) is viewed as extremely beneficial, harmful or neutral Favor o ppose : The dominant topic of each tweet (i ssue/ p olicy, p olitician, p olitical institution, p olitical p arty, s elf i nterest, p olitical e vent or v alues) is viewed as extremely favorable, faces extreme opposition or neutral 14 Tweet: Call to action : Indicate whether there is a call to action in each tweet. Choose one answer. The emphasis is on the author stating s 1. Call to action 2. No call for action 15 Tweets: Persuasion s trategies : Indicate either the presence or absence for each strategy listed below Indicate either 0:absent or 1:present for each issue Associates negotiation or compromise wit h dishonor or failure (Campbell et. al, 1960; Gusfield, 1962 p. 22) Denies legitimacy of political institution 1. Associates negotiation or compr omise with dishonor or failure 2. Denies legitimacy of political institution 1. The communist libs tactics are to wear us down,screw them !!!!cut spending!!!!We have not money to raise the debt limit,were are the cuts??? 2. Debt ceiling, just stash it the debt basement under machines that prints money. Come on congress, Arjun would turn over in his grave.
69 (Gusfield, 1962 p. 23) Var Definitions Coding Example 15, cont. Closes off deliberation process to other political viewpoints (Gusfield, 1962 p. 23) Proposes end to political institutions, processes, and/or procedures (Gusfield, 1962 p. 23) Attributes one factor for causing current conditions (Brock et al., 2005 p. 102; Smith, 2003) Emphasis on single value such as moralism, self reliance, equality (Brock et al., 2005 p. 102; Campbell et. al, 1960; Gusfield, 1962 p. 22; Smith, 2003) Expresses self interest is at stake (Binder et al. 2009 p. 317) 3. Closes off deliberation process to other political viewpoints 4. Proposes end to political institutions, processes, and/or procedures 5. Attributes one factor for causing current conditions 6. Emphas is on sing le value such as moralism, self reliance, equality 7. Expresses self interest is at stake 3. @count_01 The debt ceiling law should be repealed. Period. allowing this whole issue to arise again. 4. RT @TheNewDeal: #Retweet to Tell @BarackObama to Grow a Pair and Use the 14th Amendment Option to Raise the Debt Ceiling #TaxtheRich #p2 ... 5. @DaneilSperling massive portions of this debt can be attributed to the Bush tax cuts, our defense spending increases 6. It is IRRESPONSIBLE for anyone to think that the United States of America should default on its debt. #p2 7. RT @Rep_Southerland: President ran up $3.7 trillion in #debt over 30 months in office. I refuse to hand him a blank check to spend even ... 16 Tweets : Hashtags: Does the story mention another user or subject, using #? 1. Yes (please specify)_____ 2. No
70 Var Definitions Coding Example 17 Tweets: Hashtags : Code for the presence or absence of each hashtag (Conover, 2011) Indicate either 0:absent or 1:present for each issue 1. Top progressives : #topprogs 2. P rogressive : #p2/#p2b 3. T op conservative : #tcot 4. Organized Conservative Resista nce Allicance : #ocra 5. Compromise: #compromise #topprogs #p2/#p2b #tcot #ocra #compromise
71 REFERENCE LIS T Abelson, R. (1995). Attitude extremity. In R. Petty & J. Krosnick (Eds.), Attitude Strength: Antecedents and consequences Ohio State University series on attitudes and persuasion, Vol. 4., (pp. 25 41). Hillsdale, NJ, England: Lawrence Erlbaum Associate s, Inc. Abramowitz, A. & Saunders, K. (2008). Is polarization a myth? Journal of Politics 70 (2) 542 555. Retrieved from http://www.csupomona.edu/~smemerson/ business318/AbramCulWarMythVSFIORINA.pdf Abramowitz, A. (2010). The 2008 election: Polarization continues. In R. Niemi, H. Weisberg, & D. Kimball (Eds.), Controversies in voting behavior (pp. 266 296). Washington, D.C.: CQ Press. Abramson, P., Aldrich, J & Rhode, D. (2010). Change and continuity in the 2008 elections Washington: CQ Press. Achen, C. (1992). Social psychology, demographic variables, and linear regression: Breaking the iron triangle in voting research Political Behavior, 14 (3), 195 211. doi: 10.1007/BF00991978 Banwart, M. & McKinney (2003). Verbal debate codebook. [Content Analysis Codebook]. Copy provided by Dr. L.L. Kaid. Benoit, W. L., & Wells, W. T. (1996). Candidates in conflict: Persuasive attack and defense in the 1992 presidentia l debates. Tuscaloosa: University of Alabama Press. Berelson, B. (1952). Content analysis in communication research. Glencoe, IL: Free Press. Berelson, B Lazarsfeld, P., & McPhee, W. (1954). Voting: A study of opinion formation in a presidential campai gn. Chicago: University of Chicago Press. Binder, A., Dalrymple, K., Brossard, D. & Scheufele, D. (2009). The soul of a polarized democracy: Testing theoretical linkages between talk and attitude extremity during the 2004 presidential election. Communicat ion Research 6 (36), 315 340. doi:10.1177/0093650209333023 Blum, J. (2011, March 29). Polarized political speech discussed at LSU. The Advocate Retrieved on March 30, 2011 from http://www.2theadvo cate.com/ blogs/politicsblog/118888309.html Bosker, B. (2011, March 29). As Internet grows, is it polarizing political views? The Huffington Post. Retrieved from http://www.huffingtonpost.com/2011/03/ 29/internet polarizing politics_n_842263.html
72 Boulianne, S. (2011). Stimulating or Reinforcing Political Interest: Using Panel Data to Examine Reciprocal Effects Between News Media and Political Interest. Pol itical Communication 28 (2), 147 162. doi:10.1080/10584609.2010.540305 Brock, B., Huglen, M., Klumpp, J. & Howell, S. (2005). Making sense of political ideology: The power of language in democracy. Lanham, MD: Rowman & Littlefield Publishers. Brundidge, J contribution of the Internet to the heterogeneity of political discussion networks. Journal of Communication, 60 (4) 680 700. doi:10.1111/j.1460 2466.2010.01509.x Campbell, A., Conve rse, P., Miller, W. & Stokes, D. (1960). The American voter. New York: Wiley. Chilton, P. (2004). Analysing political discourse: Theory and practice New York, NY: Routeledge. Clawson, R. & Oxley, Z. (2013). Public opinion: Democratic Ideals, democratic p ractice (2 nd ed.). Thousand Oaks, CA: CQ Press, an imprint of SAGE Publications. Conover, M., Ratki ewi cz, J., Francisco, M., Goncalves, B., Flammini, A., & Menczer, F. (2011) Political polarization on Twitter Association for the Advancement of Artifici al Intelligence. Retrieved from http://truthy.indiana.edu/ site_media/pdfs/conover_icwsm2011_polarization.pdf Conover, P. & Feldman, S. (1981). The origins and m eaning of liberal/conservative self identifications. American Journal of Political Science, 25 (4), 617 645. Converse, P. (1964). The nature of belief systems in mass publics. In D. E. Apter (Eds.), Ideology and Discontent (pp. 206 261). New York, NY: Free Press. Delli, C. M. X., & Keeter, S. (1996). What Americans know about politics and why it matters. New Haven, CT: Yale University Press. Downs, A. (1957). An economic theory of democracy New York: Harper. Edwards, K. (1990). The interplay of affect and cognition in attitude formation and change. Journal of Personality and Social Psychology 59(2), 202 216. doi:10.1037/0022 35126.96.36.199 Epstein, B., & Kraft, R. (2010, January). Why less is doing more: The political uses, influence and potential of Twit ter Paper presented at the Southern Political Science Annual Conference. Retrieved from http://citation.allacademic.com/ meta /p_mla_apa_research_citation/3/9/5/9/8/pages395985/p395985 1.php Erikson, R., & Tedin, K. (2007). American public opinion: Its origins, content and impact (7 th Ed.). United States: Pearson Education, Inc.
73 Fiorina, M., & Abrams, S. (2007). Political polarization in the Amer ican public. Forthcoming in Annual Review of Political Science, 11, 1 54. Retrieved from http://dev.wcfia.harvard.edu/sites/default/files/Fiorina_0.pdf Fiorina, M., & Abrams, S D. Kimball (Eds.), Controversies in voting behavior (pp. 309 318), (5th ed.) Washington, DC: CQ press. Fiorina, M. (1981). Retrospective voting in American national elections. New Haven: Yale University Press. Fiorina, M., Abrams, S. & Pope, J. (2006). Culture war?: The myth of a polarized America. United States: Pearson Education, Inc. Freelon, D. (2008). ReCal: Reliability calculation for the masses. [Online Utility] Retrieved from http://dfreelon.org/utils/recalfront/ Gamson, W. A. (1992). Talking politics. Cambridge, UK: Cambridge University Press. Gaver, W. (1991). Technology affordances. Association for Computing Machinery. ACM 0 89 791 383 3/91/0004/0079 Gibson, J. (1977). The Theory of Affordances. In R. Shaw & J. Bransford (Eds.), In Perceiving, Acting, and Knowing: Toward an Ecological Psychology (pp. 67 83). Hillsdale, NJ: Erlbaum Golbeck, J. & Hansen, D. (2011). Computing politi cal preference among Twitter followers. Tech Report HCIL 2010 20 Retrieved from http://cgis.cs.umd.edu/localphp/hcil/tech reports search.php?number=2010 20 Gold e, S., & Nie, N. (2010). The effects of onl ine news on political behavior. Retrieved on June 22, 2011 from http://www.stanford.edu/~sgolde/Papers/ the_effects_of_onl ine_news_on_political_behavior.pdf Goren, P. (2005). Party identification and core political values. American Journal of Political Science, 49 (4), 881 896. doi:10.1111/j.1540 5907.2005.00161.x Goren, P., Federico, C., Kittilson, M. (2009) Source cues, pa rtisan identities, and political value expression. American Journal of Political Science. 53 (4), 805 820. doi: 10.1111/j.1540 5907.2009.00402.x Graber, D. (2005). Political communication faces the 21 st century. Journal of Communication, 55 (3), 479 507. do i: 10.1111/j.1460 2466.2005.tb02682.x Green, D. P., Palmquist, B., & Schickler, E. (2002). Partisan hearts and minds: Political parties and the social identities of voters. New Haven, CT: Yale University Press.
74 Gross, S., Holtz, R. & Miller, N. (1995). Att itude certainty. In R. Petty & J. Krosnick (Eds.), Attitude Strength: Antecedents and consequences Ohio State University series on attitudes and persuasion (Vol. 4). Hillsdale, NJ, England: Lawrence Erlbaum Associates, Inc. Gurevitch, M., Coleman, S., & Blumler, J. (2009). Political communication: Old and new media relationships. The Annals of the American Academy of Political and Social Science 625 (1), 164 181. doi: 10.1177/0002716209339345 Gusfield, J. (1962). Mass Society and Extremist Politics. Ameri can Sociological Review 27 (1), 19 30. Retrieved from http://www.jstor.org/stable/2089715 Hart, R. P. (2000). Campaign talk: Why elections are good for us. Princeton, NJ: Princeton University Press. Hen dricks, J. A., & Denton, R. E. (2010). Communicator in chief: How Barack Obama used new media technology to win the White House. Lanham, MD: Lexington Books. Hillygus, D. S., & Shields, T. G. (2008). The persuadable voter: Wedge issues in presidential camp aigns. Princeton, NJ: Princeton University Press. Howard, A. (2011, August 2). Debt Twitter talk shows influence of political communication over social networks. CBS News Retrieved on August 15, 20 11 from http://www.cbsnews.com/8301 504943_162 20087135 10391715.html Iyengar, S. & Hahn, K. (2009). Red media, blue media: Evidence of ideological selectivity in media use. Journal of Communication 59( 1), 19 39. doi:10.1111/j.1460 2466.2008.01402.x Ja mieson, K. H., & Birdsell, D. S. (1988). Presidential debates: The challenge of creating an informed electorate. New York, NY: Oxford University Press. Kaid, L. L. (2004). Handbook of political communication research. Mahwah, N.J.: Lawrence Erlbaum Associ ates. Kelley, H. & Michela, J. (1980). Attribution theory and research. Annual Review of Psychology, 31 ,457 501. doi: 10.1146/annurev.ps.31.020180.002325. Key, V. O., & Cummings, M. (1966). The responsible electorate: Rationality in presidential voting, 1936 1960 Cambridge: Belknap Press of Harvard University Press. Kim, Y. (2011). The contribution of social network sites to exposure to political difference: The relationships among SNSs, online political messaging, and exposure to cross cutting perspec tives. Computers in Human Behavior, 27 (2), 971 977. doi=10.1016/j.chb.2010.12.001
75 Krosnick, J., Boninger, D., Chuang, Y., Berent, M. and Carnot, C. ( 1993). Journal of Personality and Social Psychology, 65 (6), 1132 1151. doi: 10.1037/0022 35188.8.131.522 Lazarsfeld, P. F., Berelson, B., & Gaudet, H. (1968). The people's choice: How the voter makes up his mind in a presidential campaign. New York: Columbi a University Press. Lipschultz, K. (2011, August 1). White House: Twitter helped debt agreement along: Virtual pressure pushed deal. Adweek Retrieved from http://www.adweek.com/news/technology/white house twitter helped debt agreement along 133793 Lombard, M., Synder Duch, J., & Bracken, C. (2010, June 1). Practical resources for assessing and reporting intercoder reliability in content analysis r esearch projects. [Online resource]. Retrieved from http://matthewlombard.com/reliability/ McQuail, D. (2010). Thousa nd Oaks, CA: SAGE Publications. Moe, H. (2009) Online media participation and the transformation of the public sphere: Moving beyond the fragmentation debate. Paper for the AoIR conference Internet Research 10.0 Internet: Critical, 7 10 October 2009, Milwaukee, WI. Retrieved from www.hm.uib.no Moir J. (2010). The language of political opinion: Discourse, rhetoric and voting behavior. In U. Okulska, U. & P. Cap (Eds.), Perspectives in politics and discourse (Vol. 36). (pp. 237 254). Philadelphia, PA: John Benjamins Publishing Co. Niemi, R., Weisberg H., & Kimball, D. (2010) Is the American electorate polarized? In R. Niemi, H. Weisberg H, & D. Kimball (Eds.), Controversies in voting behavior (5th ed.) Washington D.C.: CQ press. O'Shaughnessy, N. J. (2004). Politics and propaganda: Weapons of mass s eduction. Ann Arbor, MI: University of Michigan Press. Ostrow, A. (2011, July 30). Obama loses 36,000+ Twitter followers in #Compromise Campaign [STATS]. Mashable. Retrieved from http://mashable.com/2011/ 07/29/obama compromise campaign stats/ Pennebaker, J., Booth, R. and Francis, M. (2007). Linguistic Inquiry and Word Count 2007 (LIWC2007). [Text analysis software]. Retrieved from http://www.liwc.net/index.php Petty, R. & Krosnick, J. (1995), Attitude Strength: Antecedents and consequences Ohio State University series on attitudes and persuasion (Vol. 4). Hillsdale, NJ, England: Lawrence Erlbaum Associates.
76 Pew Internet and Am erican Life Project. (2011). The Internet and the campaign 2010 [Data file]. Retrieved from http://www.pewinternet.org/Reports/2011/The Internet and Campaign 2010.as px Pew Research Center for People and the Press. (2011). Public Sees Budget [Data file]. Retrieved from http://www.people press.org/2011/08/01/public sees budget negotiations as ridiculous disgusting stupid/ Popkin, S. L. (1994). The reasoning voter: Communication and persuasion in presidential campaigns Chicago: University of Chicago Press. Prior, M. (2011). News vs. entertainment In R. Niemi, H. Weisberg, & D. Kimball (Eds.) Controversies in Voting Behavior. Washington, D.C.: CQ Press. Smith, R. (2003). Political extremism: Left, center, and right. The American Sociologist,34 (1), 70 80. Retrieved from http://www.springerlink.com/ content/4carceep72mce408/ Stroud, N. (2010). Polarization and partisan selective exposure. Journal of Communication 60 (3), 556 576. doi:10.1111/j.1460 2466.2010.01497.x Stroud, N. J. (2011). Niche news : The politics of news choice. New York: Oxford University Press. Tausczik, Y. and Pennebaker, J. (2009). The psychological meaning o f words: LIWC and computerized text analysis methods. Journal of Language and Social Psychology 29 (24), 24 54. doi: 10.1177/0261927X09351676 Tumasjan, A., Sprenger, T., Sander, P. & Welpe, I. (2010). Predicting elections with Twitter : What 140 characters reveal about political sentiment. Association for the Advancement of Artificial Intelligence. Retrieved from www.strategie.wi.tum.de/fileadmin/tuwib17/www/ICWSM10_Predicting_Elections_ with_Twitter.pdf Tedesco, J. (2004). Changing the channel: Use of the I nternet for communicating about politics, In L. L. Kaid (Ed.), Handbook of political communication research Mahwah, NJ: Lawrence Erlbaum Associates. Thelwall, M., Buckley, K., & Paltoglou,G. (2011). Sentiment in Twitter events. Journal of the American So ciety for Information Science and Technology. 62 (2), 406 418. doi:10.1002/asi.21462 Timeline of events in debt ceiling debate. (2011, August 1). CSPAN. Retrieved from http://www.c span.org/Events/Timeline of Events in Debt Ceiling Debate/10737423199/ Twitter (n.d.). Retrived July 26, 2012 from Twitter : https://twitter.com/about#about
77 Wu, S., Mason, W., Hofman, J., & Watts, D. (2011). Who says what to whom on Twitter Retrieved from: http://research.yahoo.com/pub/3386 Young, L. & Soroka, S. (2011). Affective news: The automated coding of sentiment in political text. Forthcoming in Political Communication, 1 32. Retrieved on July 27, 2011 from www.snsoroka.com/files/2011YoungSoroka(PolComm).pdf Zaller, J. R. (1992). The nature and origins of mass opinion. New York: Cambridge. Zappen, James, P. (2005). Digital rhetoric: Toward an integrated theory. Technical Communication Q uarterly 14 (3), 319 325. Zuckerman, M. (1979). Attribution of success and failure revisited, or: The motivational bias is alive and well in attribution theory. Journal of Personality, 47 (2), 245 287. doi: 10.1111/j.1467 6494.1979.tb00202.x
78 BIOGRAPHICAL SKETCH In 2007, Heather Houston graduated from the Univers ity of Florida with a Bachelor of Sc ience degree from the College o f Journalism and Communications in Telecommunication with a specialization in News. Heather then attended graduate school at the University of Florida to study political communication and campaigns. During her graduate prog ram, Heather was a member of class VII in the Florida Gubernatorial Fellowship Progr am under Governor Rick Scott. Heather Houston received the degree of Master of Ar ts in Mass Communication at the College of Journalism and Communications in August 2012.