Between peace and hate : Framing the 2014 Colombian presidential election on Twitter

RESUMEN Este estudio examina cómo dos grupos en Twitter —los 100 periodistas con más seguidores y el público— encuadraron las elecciones presidenciales colombianas de 2014. Se usó un software con un algoritmo capaz de analizar grandes bases de datos que recolectó y categorizó los tuits publicados por estos grupos durante la campaña electoral (15 de mayo al 20 de junio). Se realizaron dos análisis: los resultados del primero mostraron que los periodistas siguieron el encuadre temático, mientras que el público se interesó en el conflicto. En el segundo, los periodistas adoptaron el odio como encuadre y el público, el de la paz.


INTRODUCTION
On June 16, a day after the second round of the 2014 presidential election in Colombia, mainstream TV news media Caracol News posted a quote on its Twitter account, @NoticiasCaracol, from the winner of the election, incumbent President Juan Manuel Santos: "Yesterday, we turned the page on hate and false accusations", said Santos.Journalists endorsed the quote, vigorously retweeting and replying to Caracol News' post.Santos' remark marked the end of an acrimonious campaign against opposition candidate Óscar Iván Zuluaga, who was supported by Santos' predecessor, former boss, and most formidable adversary, former President Álvaro Uribe (2002Uribe ( -2010)).The fact that Santos included the word "hate" in his tweet was no accident.On several occasions, Santos, his wife and closest allies had used the word hate on Twitter and on other media outlets to describe and frame the opposition party, the Democratic Center, its presidential candidate, Zuluaga, and its top leader, Uribe.
Before the 2014 election, Santos' decision to begin peace talks on November 19, 2012, with the Revolutionary Armed Forces of Colombia (FARC) 1 , the country's largest rebel group at the time, enraged Uribe 2 .Once strong allies, Uribe had backed Santos' candidacy during the 2010 presidential election, hoping Santos would continue his eight-year security policies against the FARC.Santos had served as defense minister under Uribe, and together they dealt major military blows against the FARC (Uribe Vélez, 2012).Yet, once elected president in 2010, Santos distanced himself from his former boss on key issues, including the armed conflict against the FARC and international affairs.During his second term and after four years of peace negotiations in Havana, President Santos and the FARC signed a controversial peace accord on November 24, 2016.As a result of this political split-up, Uribe founded a new political party, the Democratic Center.The new party's main goals were to oppose the Santos administration, block the peace process, mark Uribe's return to the political arena, and defeat Santos in the 2014 presidential election.Uribe's party eventually nominated Zuluaga as its presidential candidate.
Social media, especially Twitter, have been used intensively in Colombia for political debate.Uribe and Santos have been on Twitter since 2009, which has increased online political discussion among journalists and users (García-Perdomo, 2017).Uribe has repeatedly attacked the Santos administration and the peace process with the FARC on several different media outlets, but his favorite platform has been Twitter.Journalists not only follow these political actors very closely on that social media, but they also profusely tweet their opinions on political issues (García-Perdomo, 2017).
This paper examines how two groups on Twitter, the 100 most-followed Colombian journalists and members of the public, framed the 2014 Colombian presidential election on Twitter.Using a social media analysis platform that applies supervised, machinelearning algorithms, this study collected and analyzed all tweets posted by these two groups during the 2014 presidential election campaign, from May 15 to June 20.The report conducts two distinct framing analyzes following previous theory and social network analysis.This research sheds light on how the peace process between the Colombian government and the FARC guerrillas -as well as the country's history with political violence-permeated the political discussions on Twitter.Finally, this paper compares the different frames used by journalists and social media users to interpret a presidential campaign and election.

LITERATURE REVIEW MEDIA FRAMING ON POLITICS
Framing is currently one of the most referred to and growing theories in the field of communication (D'Angelo & Kuypers, 2010;Matthes, 2009).Frames are "organizing principles that are socially shared and persistent over time that work symbolically to meaningfully structure the social world" (Reese, 2001, p. 11).Gitlin (1980) argues that frames make references to principles of selection and emphasis, while others argue that by highlighting specific elements through language choice and repetition, news frames can impact citizens' perceptions (Chong & Druckman, 2007).Other scholars argue that frames are able to define problems, make moral judgment and suggest remedies (Entman, 1993).Hertog and McLeod (2001) underline a broader cultural approach to frames when they describe them as "structures of meaning made up of a number of concepts and the relations among those concepts" (p.140).Reese's and Gitlin's definitions are particularly useful for analyzing long historical processes and ideas that shape and structure societies (e.g., the War on Terror).
Journalists actively participate in the construction of news frames (Reese, 2007), given that news workers are not mere puppets of political leaders and sources, but they include their own perspectives and ideas when reporting stories.Above all, reporters with experience can interpret and include their frames on news events.In this research, I argue that the online political participation of journalists and users in the same social media space could help elucidate why some frames are more relevant than others for journalists and the public.The intersection of professionals and audiences in the process of shaping information may shed light on some principles that organize and structure the social world.In the following section, this study reviews previous research that has identified generic frames present in political coverage.

GAME FRAME
Scholars have been interested in studying how the media have increasingly adopted a reporting style that focuses on the political strategy of the candidates and their possibilities of winning or losing at key turning points of the campaign (Woong, 1997;Patterson, 1993).In the classic book Spiral of Cynicism, Cappella & Jamieson (1997) explain that some features of the strategy frame in political news coverage are, "1) wining and loosing as the central concern, 2) the language of wars, game and competition, 3) a story with performers, critics and audience (voters), 4) centrality of performance, style and perception of candidates, and 5) heavy weighing of polls and the candidates standing in them" (p.33).In this sense, as Patterson (1993) argues, the political game provides the plot of a story that makes it easy for reporters to narrate political processes, and the electorate is seen as a mere spectator of the candidates who are at the center of the political action.
Researchers have paid attention to the implications of this strategic journalistic coverage on the democratic process, as news media may be accountable for improving or reducing civic engagement and political participation (Aalberg, Strömbäck, & De Vreese, 2012).Cappella & Jamieson (1997) argued that this strategic media coverage could lead citizens to completely lose their trust in the political system and its institutions by inviting cynicism (p.31).Cynicism, in turn, seems to promote general mistrust toward politics, regardless of whether there is any evidence of wrongdoing against leaders or institutions (Cappella, 2002).Other studies, however, have argued against the idea of strategic reporting as cynical and the negative effects of news media on the political process (De Vreese, 2005).
Even though the effects of political media coverage are subject to much debate on concepts such as bias (Entman, 2010) and cynicism (Cappella, 2002), there is a growing tendency in journalistic political coverage to emphasize the game aspects of the campaign.Lawrence (2010) has defined the game frame, comparing it with sports: "Game-frame news treats politics much like a sport event, with journalists playing the role of commentators and interpreters of various players' strategic moves" (p.272).Global trends in political news coverage have shown that media have increased the use of the strategic game frame in recent election campaigns, focusing on "winners versus losers rather than substantive discussion of political issues" (Dimitrova & Kostadinova, 2013, p. 84).Aalberg et al. (2012) have not only shown some conceptual differences between the game frame and the strategic frame, but also how these frames are highly present in a number of countries, outnumbering other such as the issue-policy frame and the conflict frame.Lawrence (2000) has also noticed how reporters ensure an aura of objectivity by treating political news as sports commentators.When using the game frame, journalists focus on "the technical aspects of the political game," creating an apparent distance from their sources (Lawrence, 2000, p. 95).Bennett (1996) has argued that the game frame reflects journalists' preference for drama and for the personalization of news given that this frame facilitates presenting news as a sports competition between politicians.Simple narrative structureseasy to produce and consume-are appealing to the media and their audiences.In sum, the game frame privileges winning political strategies while obscuring candidates' proposals.Lawrence (2010) has also developed a compelling argument about the differences between foreign policy news and election campaigns news.The scholar argues that while journalists tend to rely heavily on government officials during times of international conflict, they display more independence from their sources -mainly candidates and campaign staffwhen they cover elections (p.267).Thus, journalists show more independence and are able to introduce their own perspectives and interpret campaign news (Patterson, 1993).Meanwhile, reporters and media are constrained by national security issues when they cover topics related to international affairs.In this paper, I argue that both journalists and audiences promote their frames and show some independence on election coverage by using the network public sphere (Shoemaker & Reese, 2014), particularly social media.

CONFLICT FRAME
The conflict frame has been constantly found in both journalists' and audiences' discourses.Semetko & Valkenburg (2000) illustrate how media capture audience interest by emphasizing "conflict between individual, groups, or institutions" (p.95).In political news, the conflict frame not only includes politicians directly attacking each other, but also strong disagreements and diverging political views expressed in the media (Putnam & Shoemaker, 2007).Distinguishing different types of conflict frames, Bartholomé, Lecheler and De Vreese (2017) identify 1) journalistic interventionism and 2) substantiveness of political debates as key aspects in the conflict frame-building process.In the first case, Bartholomé et al. (2017) argue that journalists use conflict to frame political events when they emphasize news interpretation and the journalists' watchdog role.In the second case, the conflict frame can be used as a substantive element of the political debate (i.e., ideological clash) or a non-substantive device (i.e., controlled personal attacks).Other scholars have confirmed the journalists' tendency to frame stories in terms of conflict as an effective mechanism to narrate news (Semetko & Valkenburg, 2000) even though this particular conflicting approach seems to contribute to different outcomes that range from eroding political trust (Patterson, 1993;Mutz & Reese, 2005) to create awareness of the importance of the political issue at stake (De Vreese, 2004;Schuck et al., 2013) and increasing support for certain policies (Vliegenthart & Schuck, 2008).Moreover, journalists and audiences seem to use similar conflict frames or news values to make sense of events in terms of conflict (Neuman, Just, & Crigler, 1992;García-Perdomo, Salaverría, Kilgo, & Harlow, 2017).
There are differences between the game frame and the conflict frame, though they seem to be very close at first glance.Lawrence (2010) has explained some of the dissimilarities between these two frames: "Not all conflict lends itself equally well to the game schema, nor do reporters apply the game schema equally to all types of political conflict" (p.96).Lawrence (2010) conceives the game frame as a "controlled conflict," which is defined by Fallows (1997) as a crucial characteristic present in every sport that revives contest between the parties.In this sense, the game frame can be distinguished from the conflict frame because the former focuses on strategies and the winning and losing aspects of the campaign, whereas the later focuses on open confrontation between candidates.

ISSUE FRAME
Political campaigns offer politicians, institutions and citizens the opportunity to publicly debate important issues.The issue frame has been also called the policy substance frame (Patterson, 1993) or issue coverage (Cappella & Jemieson, 1997).It embraces the fundamental socioeconomic and political concerns that candidates are supposed to address during a thoughtful campaign.According to Woong (1997), the issue-policy frame "highlights proposals for the problems, information about who is advocating which policy alternative, and consequences of problems and proposals" (p.30).Therefore, the issue frame is defined "as news stories focusing on the substance of political problems, issues, or proposals" while other frames such as the game frame focuses more on the "personality contest" and "strategies and tactics for winning" (Aalberg et al., 2012, p. 168).Schnell (2001) has identified the issue frame as being present in a story if the policy debate is thematically structured.In other words, the political issue frame emphasizes the main arguments of the campaign.
Political debates are now mediated as they happen within the context of media or on online social networks.Some scholars are worried about the quality, transparency and truthfulness of online political debates (Wadhwa & Bhatia, 2015;Chau & Xu, 2007), while others celebrate the Internet and online networks as facilitators of political discussion, engagement and participation (Gil de Zúñiga, Jung, & Valenzuela, 2012;Gil de Zúñiga, Molyneux, & Zheng, 2014;Shah et al., 2007).
Media and journalists, in particular, have been accused of being biased in their political coverage (Entman, 2010).Media critics have proposed an alternative reporting method which focuses on issues, problems, and solutions during campaigns as a formula to regain balance and professional authority (Woong, 1997).Framing substantial political issues (and not conflict or just political strategy) seems to be a way to improve campaign coverage and political debate.Recently, studies have documented the increase of more analytical journalism in the news coverage of politics and current affairs (Salgado & Strömbäck, 2012).Bringing socio-political issues to the center of a campaign, however, is not only the journalists' responsibility.Candidates and citizens are key to that process because they also participate in the public media sphere through social networks and other online platforms.
Considering this theoretical background, and given the fascination of journalists with the game frame in previous research, this study poses the following hypotheses and research question to explain the context of Colombia's electoral process: • H1.The game frame (strategy) will be more common than the conflict frame in the journalists' Twitter coverage of the 2014 Colombian presidential election.
• H2.The game frame (strategy) will be more common than the issue frame in the journalists' Twitter coverage of the 2014 Colombian presidential election.
Since there is not enough evidence about the differences between journalists and the public in their coverage of politics on Twitter, this study asks the following question: • RQ1.What are the main differences between users' and journalists' framing activities on Twitter during the 2014 Colombian presidential elections when it comes to the game, conflict, and issue frames?

PEACE AND HATE, THE MAIN CONCEPTS OF THE CAMPAIGN
The main issue that dominated the 2014 Colombian presidential election was the peace process between Santos' government and the FARC guerrilla.A social network-semantic analysis of the data using Crimson Hexagon (CH) -a tool that allows researchers to track and collect all Twitter activity-found that the discussion about peace was not only the most commonly discussed issue, but it also was clustered with the word hate in the Colombian political debate.Based on this evidence, this research sought to find the differences between the 100-most followed journalists and the public when discussing peace and hate on Twitter during the 2014 presidential election.As neither peace and issue frames nor conflict and hate frames were mutually exclusive, it was necessary to separate the analyses of the peace and hate frames from generic political frames, such as the game, conflict and issue frames.
In his book Media and Political Conflict, Gadi Wolfsfeld (1997) explains the notion of peace framing by looking at how journalists cover peace negotiations and conflict.Wolfsfeld's main conclusion after looking at different peace talks in the Middle East was that the "role of news media in political conflict varies over time and circumstance," depending on the political resources available, the amount of power of the antagonists, the perceived state of public opinion, the ability of journalists to gain access to the conflict events, and what is going on in the field (p.4).In this vein, Wolfsfeld's political contest model can help us understand not only how media portray conflicts between authorities and challengers, but also the circumstances in which conflicting factions compete for media access and social meaning to impose their frames.According to Woldsfeld, the media contribute to determining what frames and issues prevail over others because antagonists have the opportunity to promote their own interpretations.Therefore, authorities' level of control over the political environment determined the role of media in political conflict and peace coverage (Wolfsfeld, 1997).This argument, of course, conditions media's and journalists' participation in framing conflict because media seem to articulate only ideas and actions presented by authorities and political actors.Galtung (1986Galtung ( , 2002) ) was the first to propose war and peace journalism as two competing frames in the news coverage of conflicts (Cozma, 2015).The war frame adopts conflict as a central news value and, taking a violent perspective, shows with sensationalistic language the horror of casualties to catch the audience's attention (Lee & Maslog, 2005).In contrast, peace journalism "concentrates on areas of agreement, tones down political and ideological differences, (…) thus promoting conflict resolution and reconciliation" (Cozma, 2015, p. 436).Galtung (2002) highlights how journalists particularly hold the power to promote news perspectives that avoid violent-orientation and foster conflict resolution, reconstruction, and reconciliation (Galtung, 1986(Galtung, , 2002;;Lee & Maslog, 2005).
In this struggle for imposing meaning in the Colombian context, peace and hate (not war) appeared as the most mentioned issues during the 2014 political campaign.Below, this research expands the context in which these overarching concepts were used.
Peace frame.Santos -who was aiming for reelection in 2014 after four years in power-decided to present himself as the peace candidate in order to obtain popular support for the peace talks between his government and the FARC guerrilla in Havana, Cuba.The peace process became one of the main issues of the campaign, with the two major candidates espousing different visions.Peace was an issue that defined the election.Journalists, politicians and citizens engaged in long discussions about the political convenience of continuing or ending the negotiations.
Hate frame.One of the most common words during the Colombian presidential campaign, hate was identified in both the public's and the 100-mostfollowed journalists' discourses.Hate and peace usually appear related, very close to each other, and connected in word clouds and clusters in the CH analysis.Hate and violence have a special historical connotation in Colombia.The country even has a period of history called La Violencia [The Violence] (1948-1953), in which political hatred had devastating consequences.During the 2014 presidential election, the opposition candidate was portrayed as the leader of the hateful group.Because opposition candidate Zuluaga was against the peace process between the Colombian government and the FARC guerilla in Havana, he and his followers were depicted on social media as vindictive people who were in favor of war and hate.
After providing this context, this research poses the following question: • RQ2.What are the main differences in the way that the 100-most-followed journalists and the public understand peace and hate on Twitter during the 2014 Colombian presidential campaign?

METHOD
This research conducts two separate analyses using the software Crimson Hexagon (CH), a system that accesses and analyzes the full "firehose" of Twitter.The Pew Research Center (2012), which used the tool for its 2012 U.S. presidential election coverage, says that the "software is able to analyze the textual content from millions of tweets on social media platforms.Crimson Hexagon (CH) classifies online content by identifying statistical patterns in words" (p.16).CH allows researchers to create data sets and analyze them in independent spaces called monitors.There are different types of monitors in CH.The most basic of them, called a Buzz Monitor, is used to understand word clusters and the general sentiment -positive, negative or neutral-surrounding an issue, person or organization.The most complex feature of CH is the Opinion Monitor, which contains an algorithm that can be trained to analyze the content at the assertion level.The Opinion Monitor tracks, recognizes, codes and compares specific ideas, topics or frames inside large data sets.The Opinion Monitor of CH has settings that allow researchers to choose the language and target specific nations and regions.For this study, Spanish and Colombia were used as settings.
Before doing the framing analysis, two exploratory Buzz Monitors were opened to look at Twitter activity as a whole.The first exploration of the Colombian presidential election gathered tweets from May 1 to June 30, 2014, using a Buzz Monitor that targeted key words in Spanish, such as "election," "vote," "democracy," "parties," "candidates," etc.The first Buzz Monitors collected the general public's activity on Twitter between those dates.In this exploration, the software displayed auto sentiment analysis, relevant topic waves of discussion and word clusters.Peak user participation occurred on May 25 and June 15.This exploration was useful because it not only provided the auto sentiment analysis of the campaign-35% negativity, only 14% positivity, in the case of the public-but also major players, clusters, links, word clouds, topic waves and general activity on Twitter.The second Buzz Monitor explored the Twitter accounts of the 100 most-followed Colombian journalists.Peak journalist participation also occurred on May 25 and June 15.This exploratory stage was useful for creating a codebook, explaining the frames and training the algorithm with exemplars.
To do the actual framing analysis, this study opened and trained four Opinion Monitors in CH, collecting tweets from May 15 to June 20, 2014.The final timespan selected includes the first round of the presidential election and the runoff between Santos and Zuluaga.Opinion Monitors allow researchers to command the algorithm to look for specific information in the data by providing exemplars of tweets for each frame or category.A codebook was used as a guide to select exemplars of tweets and train the supervised-learning algorithm.Once trained, the software classified vast amounts of tweets automatically and visualized vast amounts of data.To ensure the reliability and validity, this research incorporated traditional content-analysis methods with computational analysis.
The first and second Opinion Monitors tracked the activity of general users and journalists to analyze generic frames already defined by journalism scholars, such as the game, conflict and issue frames.In the third and fourth Opinion Monitors, the algorithm was trained to identify the peace and hate frames from the general public's and journalists' activities.These monitors were created in a different analytical set to avoid conflict with the more traditional frames (game, conflict, and issue frames).To ensure the exclusion of journalists from the public's Twitter activity, a black list with the accounts of the 500 most active Colombian journalists-including the 100 most-followed reporters considered in this study-was created in order to block journalists' activity from monitors that tracked the users' tweets.Each of the four Opinion Monitors was trained individually.

RESEARCH DESIGN
This study analyzes two different groups: The 100 most-followed Colombian journalists on Twitter and members of the public who use Twitter.The 100 mostfollowed journalists were chosen using the website Twitter-Colombia.com,one of the most comprehensive lists of Colombian journalists who are on Twitter.The list was compared and enhanced using other lists, including TwitterCounter.comand Twittercolombia.net.Though having a large number of followers was a condition for selecting the journalists' Twitter accounts, this study also considered the professional relevance of the journalists.Some sports and entertainment personalities were excluded from the list because they seldom express political positions or get involved in the electoral process.Twitter users from Colombia who tweeted in Spanish were part of the second group.
Coding.The unit of analysis of this study is the tweet.Previously, a codebook was written to serve as a guide to analyze and select the tweets on which to train the algorithm.Three trained human coders, who were native Spanish speakers with a graduate education and knowledge of Latin America's historical and political context, helped make the study reliable.The first coder and the second coder agreed upon the relevance of the tweets that were pulled from the data as exemplars to train the four monitors' algorithm.These two coders validated the tweets that represented the frames and issues and trained the social media analysis platform.CH applies a supervised, machine-learning algorithm to analyze large amount of data.Sixty exemplar tweets were carefully chosen for each one of the five frames in two separate data sets: one for the 100-most-followed journalists and one for members of the public on Twitter.A total of 600 tweets served as a semantic guide to instruct the algorithm during this study.To ensure intercoder reliability, a random 10% sample from those 600 tweets was pulled out in an Excel spreadsheet.A third human coder, who also was a native Spanish speaker with a graduate education and with knowledge of the Latin America's historical and political context, read the codebook and then coded a sample in an Excel spreadsheet outside the CH tool.Two pre-tests were performed on the sample until the first and the third coders reached an agreement of 90% or higher on all variables (Poindexter & McCombs, 2000).

MEASURES
To choose tweets that truly contained the frames and issues, coders adhered to the following operationalization of measures: Conflict frame.Following Semetko & Valkenburg (2000), this measure was operationalized using the following questions to decide whether tweets contained the conflict frame or not: "Does the story [tweet] reflect disagreement between parties-individuals-groupscountries?Does one party-individual-group-country reproach another?Does the story [tweet] refer to two sides or to more than two sides of the problem or issue in open confrontation?" (p.98).An exemplar tweet that represents the conflict frame for this analysis is the following: On June 11, 2014, user @andressep0 posted on Twitter, "Santos and Zuluaga are attacking each other really hard using propaganda and media, [Why don't they] hit each other in their faces, and finish this [campaign]".
Game frame.The game frame privileges strategic and tactical aspects of the campaign over other important issues.Winning and losing is the most important aspect here.According to Lawrence (2010), "gameframe news treats politics much like a sport event, with journalists playing the role of commentators and interpreters of various players' strategic moves" (p.272).For this reason, this measure was operationalized with the following questions to analyze whether tweets contain this frame or not: Does the tweet mention any winning or losing aspect of the campaign?Does the tweet mention strategic alliances that allow one of the candidates to advance in the race?Does the tweet mention any strategic move or response from candidates?Does the tweet mention poll results?Does the tweet mention any tactical aspect of the campaign?An example of a tweet that contained the game frame and that was used to train the algorithm was the following: "This is the political strategy with two days left until the election: Uribe is in charge of all objections while Zuluaga appears as a just and balanced man".Issue frame.Coders followed this question as a guide: Does the tweet mention any important issue for the campaign such as the peace process, land reform, environment, legislation, employment, economic reform or any fundamental issue that could be relevant for the country?A good example of a tweet containing the issue frame is the following: On May16, Twitter user @darioacevedoc posted, "Zuluaga has serious political proposal in many fields: [He wants a] peace negotiation but not at any cost, and to increase the budget for education via@ElColombiano".
Peace frame.In the process of training the CH algorithm, coders looked for tweets that made references to the peace process between the government and the FARC guerrilla that valued nonviolent responses to the conflict, creating peaceful opportunities for society at large (Lynch & McGoldrick, 2005, p. 5).The algorithm was trained to identify those out-of-context references and exclude them from the analysis.Good examples of the peace frame during the presidential election are the following: Journalist @GiraldoSantiago posted on June 10, "I want peace with justice!@OIZuluaga even much more than the negotiated peace agreement without justice of @JuanManSantos".Journalist @Carloaruizr wrote on June 9: "Santos shows he has control over the topic of peace.Zuluaga doesn't respond to the great amount of criticism around the paramilitary law".
Hate frame.To train this CH opinion monitor, coders selected tweets that contained the word hate in the political context of the Colombian presidential election, as well as emotional, negative words, phrases or meanings that serve to intensify conflict and the cycle of hatred and violence (Wolfsfeld, 2003).Usually, the frame came from government allies who wanted to portray the opposition party as a hateful group.For example, on June 4, 2014, first lady María Clemencia de Santos used the hate concept to lament the tone of the opposition during the presidential campaign.She posted on Twitter that the election was "full of lies and hate, and not of political proposals".Likewise, journalists @Eccetina tweeted on May 23: "Mrs.@ OIZuluaga and @JuanManSantos are failing in this political [TV] debate because their only thesis and proposal is hate".Journalist @Carlosaruizr also wrote on May 26: "I had nightmares.I dreamt that the election had been won by those [politicians] who prefer lies, illegal phone intersections, intolerance, hate and war".

RESULTS
Hypothesis 1 predicted that journalists would use the game frame more frequently than the conflict frame in their tweets.The results confirm this hypothesis during the 2014 Colombian presidential election.From a total of 8,387 relevant tweets gathered between May 15 and June 20, 35% were coded as belonging to the game frame, whereas only 21% were categorized as using the conflict frame (χ² = 293.545,df = 1, p < .001).The results are consistent with the theory in the sense that journalists prefer the game frame over the conflict frame when they cover elections.
According to the data, the game frame obtained the highest volume of tweets on May 18 (304 tweets), May 23 (380 tweets) and June 10 (387 tweets).A closer analysis reveals that most of the journalists' activity occurred around two political debates broadcasted by national TV channels and a breaking news event.The 100 most-followed journalists reported about who was winning or losing the debates when political candidates answered questions and argued among themselves.For example, on May 23, @solano posted during the first presidential debate the following message: "Intelligent strategy from @ClaraLopezObre @mluciaramirez and @enriquepenalosa to show the polarization between Santos and Zuluaga".Before the runoff election, on June 10, @YolandaRCN, from RCN Radio, posted on her personal Twitter account during a televised debate: "Candidates Santos and Zuluaga bare their teeth again during the [presidential] debate," and @diegosenior, from Caracol Radio, posted: "Zuluaga [looks] more assertive, Santos [looks] more sentimental #ColombiaDebate".Likewise, a major breaking news event occurred on May 18 that sparked most of the game frame.Semana Magazine, the main agenda-setter for the Colombian press, published an investigative piece about the arrest of a political advisor from Zuluaga's campaign.The magazine framed the news in terms of an unethical strategy from the opposition to obtain classified information about the peace process between the government and the FARC.Journalists followed Semana's strategic frame of this event in their coverage on Twitter.For instance, @SalcedoRamos retweeted: "@RevistaSemana EXCLUSIVE | Zuluaga and the hacker defined strategy against Santos and the #peaceprocess".
Hypothesis 2 predicted that the game frame (strategic) would be more common than the issue frame in the journalists' Twitter coverage of the 2014 Colombian presidential election.The hypothesis was not supported.The issue frame was found in 44% of tweets, while 35% of the Twitter activity was classified within the game frame (χ² = 85.8008, df = 1, p < .001).The relevance of the peace process during the 2014 campaign probably amplified the issue frame.
In particular, journalists with historical knowledge of the devastation of the war highlighted the social significance of continuing with the peace talks.For instance, journalist @solano posted the following message: "Of the 5.5 million victims of this armed conflict, 1.16 million of them are children under 12 [years old].How many more [victims] do we need before seeking peace?".Despite the significance of the issue frame in the journalists' conversation, the sum of the game frame and the conflict frame in this analysis still contains the majority of the data (56%).This means that journalists continue to use belligerent frames when reporting on social media.
The first research question asked about the main differences between the way journalists and users adopted the game, conflict and issue frames.Conflict was the most common frame in the general public's Twitter activity during the 2014 Colombian presidential election, whereas journalists were more inclined to use the game frame.From a total of 4,153,740 relevant tweets from users gathered by CH from May 15 to June 20, 2014, the conflict frame was present in 53% of the data, the issue frame in 42% and the game frame in just 5% of the public's activity.Results showed a significant difference between journalists and the public on Twitter when it came to the adoption of the game and the conflict frame: Journalists used the game frame (35%) more often than the public (5%), whereas the users implemented the conflict frame (55%) more than reporters (21%) (χ² = 16681.11,df = 2, p < .001)(see table 1).
RQ2 looked into what frames (peace or hate) proved more relevant for the 100 most-followed journalists on Twitter during the 2014 campaign.A network analysis of this group of journalists offers a total of 6,327 relevant tweets related to the analyzed issues from May 15 to June 20, 2014.Results show that journalists were more likely to choose the hate frame (66%) than the peace frame (34%) (χ² = 648.1152,df = 1, p < .001).Peaks of journalistic coverage occurred on May 23 and June 10.Looking at clusters and word clouds associated with hate, most of this activity referred to a political consultant of the opposition campaign who was arrested by authorities and charged with illegally intercepting communications from the peace negotiators in Havana who were participating in the peace process between the Colombian government and the FARC guerrilla.This consultant, known as "the hacker" by the press and authorities, prompted a great deal of hateful Twitter activity with direct accusations against the Zuluaga campaign, which appeared to have the intention of ending the peace talks and to contribute to hate and war.Also, most hate-frame tweets were present on June 10 during the last televised presidential debate due to a political ad by Zuluaga's campaign that portrayed a furious woman complaining about the high cost of living.Part of the conversation used the hashtag #LaLocaDeLasNaranajas [#CrazyOrangesWoman].For example, @andrewholes posted: "The ad #LaLocaDeLasNaranajas is a farce from Zuluaga's campaign to scare undecided voters".
Also, the second question inquired which frame, peace or hate, proved more relevant to Twitter users during the 2014 presidential campaign.The same analysis offers a total of 2,675,232 relevant tweets from the public from May 15 to June 20, 2014, that were specifically linked to the frames being studied.Results showed that the public preferred the peace frame (57%) over the hate frame (43%) (χ² = 52434.41,df = 1, p < .001).Most users on Twitter embraced the idea of continuing with the peace process and created hashtags to support the initiative of the Santos' government.The public's Twitter activity had its peak on June 10, during a televised presidential debate, and on June 15.A comparison between the two groups showed a significant difference between journalists and the public when it came to the peace and hate frames: Journalists adopted the hate frame (66%) more often than the public (43%), whereas the public used the peace frame (57%) more frequently than reporters (34%) (χ² = 1360.978,df = 1, p < .001)(see table 2).

DISCUSSION
When an important social problem such as Colombia's peace process is present in campaigns, the issue frame becomes relevant for journalists within the political debate.Given the appropriate historical context, Colombian journalists have acquired through years of peace talk fiascos a certain level of sophistication when debating the negotiation process with the FARC.Topics such as the agrarian reform, the roots of the historical violence, victims' claims, child solders, land mines, redistribution of land, inequality and pacification are present in the discussion surrounding the electoral campaign and give relevance to the issue frame.Journalists have learned from their own mistakes in the past.During the 2014 presidential election, reporters seemed to be more cautious with the negotiation process, so they used the issue-policy frame to explain the campaign.
Some interesting conclusions can be inferred when comparing the general public's and journalists' framing activity on Twitter.Even though the 100 mostfollowed journalists adopted the issue frame (44%), the combination of the game frame and the conflict frame in the deductive analysis still contains 56% of the data.The fact that the peace process between Santos and the FARC guerilla, one of the most important issues for the Colombian society, was taking place at the same time of the election transformed this event in the most relevant factor of the campaign.Peace appears at the center of any political debate during the entire campaign.
In contrast, the results of the second analysis confirmed that journalists were more likely to cover or react to the hate frame during the campaign.Most of their tweets that related to hate referred to an online political consultant who was arrested during the campaign for allegedly "hacking" the peace process and providing classified information on the peace talks to the opposition party.For the most-followed journalists, "the hacker" somehow embodied a hateful machination of the opposition to destroy the peace process.Media offered an extensive coverage of this event.The hacker's arrest caused a frenzied reaction from journalists on social networks, in part because the news was a scoop published by one of the main media agenda-setters in Colombia, Semana Magazine.Santos used the incident as an example of the hateful strategies of the opposition.The second event that grabbed a lot of attention from both the public and journalists was a political ad aired by the Zuluaga campaign on June 10, during a political debate.In the ad, an upset woman complained about high prices of basic needs.Yelling and throwing groceries at a supermarket during the ad, the woman's attitude was a good representation of hate for Twitter users.The hashtag #lalocadelasnaranjas [#CrazyOrangesWoman] was a trending topic at the end of the campaign.
On the other hand, the public showed little attachment for the game frame.It seems that the game frame is not a reasoning scheme or structure that online users tend to implement in order to understand or comment on the political debate.In contrast, users are fascinated by the conflict frame as one of the easiest ways to understand the Colombian political race.characteristic for understanding elections and political campaigns.Results also confirm that journalists continue framing elections extensively as a sport game in which strategy, winning and losing provide meaning to politics.This research is important for the future of framing as a theory, because it looks at the intersection of journalists and the public on social media to show how news frames emerge due to the violent historical conditions of Colombia and Latin America.Also, this study shows -as previous research has found in other contexts and with traditional media such as print and broadcasting (Neuman et al., 1992)-that journalists and the public frame political campaigns and elections differently using these social networks.Also, it demonstrates that although people can construct meaning about public issues with the help of media and journalists, the public does not interpret political events in the way that journalists intended particularly on Twitter.The difference between the two groups may be explained by the fact that social media has made more evident and reinforced the disconnection between the way the public and journalists frame politics and discuss their perspectives on the web.One of the most intriguing aspects of the results is the contrast between the public inclinations toward the conflict frame, which reinforces direct confrontation between politicians, but at the same time its support for the peace frame during the 2014 Colombian presidential election.
This network analysis of the most popular frames during the Twitter coverage of the 2014 Colombian presidential election looked at both the 100 mostfollowed journalists' and the public's Twitter activity to understand the overarching ideas in which the political debate was articulated on social media.This study contributes to existing literature by shedding some light on frames such as peace and hate, which can help to understand Colombia's historical and political processes.This study is limited since it uses a third-party algorithm that automates the analysis of large amounts of data, raising questions about the reliability of machines to evaluate content.For instance, tweets mentioning the word "hate" or "peace" can at times overlap because most of the hateful ideas and campaigns on Twitter occurred while the Colombian peace process was the main issue of the 2014 election campaign.Consequently, it is possible to interpret peace and hate as issues instead of frames, as they seem to be the two sides of the same coin.However, the two concepts appear to be independent and relevant organizing principles of the social world in Colombia, given the violent history of the country.Nevertheless, this study is valuable for offering insight into the social networked political discussion outside the English-U.S.-European realm.Further research is needed in order to identify peace and hate as frames present in other political campaigns and to understand the dynamic between journalists, politicians and the public on social media in Latin America.NOTES 1.The Revolutionary Armed Forces of Colombia (FARC) guerrillas proclaimed themselves to be both a Marxist-Leninist organization and a peasant army.It has a long history of violence and aborted peace processes (Bergquist, Peñaranda, & Sánchez, 2001).Drug trafficking and illegal mining have partially funded the FARC's military campaign.
2. FARC guerrilla leaders and the government decided to hold peace talks outside of Colombia.They selected Cuba as a neutral venue for these conversations.Mainstream media echoed former President Álvaro Uribe's reaction against the negotiations.See, for example: "Uribe: 'Es un error pensar que el único camino para la paz es el diálogo" [Uribe: It is a Mistake to Think Dialog is the Only Way to Reach Peace] (Semana, 2012).Or "Ex presidente Uribe criticó inicios de diálogos de paz," [Former President Uribe Criticized the Peace Talks] (El Tiempo, 2012).

Table 2 .
Results from this research confirm the general public's fascination for online messages with open confrontation between candidates as the main Public's and journalists' peace and hate frames.