Online Campaigns in the Social Media Era: A Case Study of Twitter Use During 2010 Elections in Brazil

Transcription

1 Online Campaigns in the Social Media Era: A Case Study of Twitter Use During 2010 Elections in Brazil Patrícia Rossini (PPGCOM/UFJF/BRASIL) patyrossini@gmail.com & Paulo Roberto Figueira Leal (PPGCOM/UFJF/BRASIL) pabeto.figueira@uol.com.br Paper presented at the 22th World Congress of Political Science July 08 to 14, Madrid Abstract: The proposal of this paper is to discuss the relationship between social network sites (SNS) and the decision-making process, based on the assumptions of the sociological theory of voters behavior (Zuckerman et al., 2005). As SNS provide scaffolding for social interactions and information sharing, they lower the maintenance costs of relationships and make it easier for one to participate in larger social networks. Through social ties, like Facebook friends or accounts followed on Twitter, people receive political information even if they're not interested in accessing them. Based on Granovetter's studies (1973; 1983), we argue that people who have more weak ties in their networks have access to different sources of information, and thus are better informed. Contemporary studies on the social logic of politics (Zuckerman et al., 2005) present similar conclusions: the presence of weak ties on one's social network increases one's chances of being informed, recruited and mobilized by activists and political groups, leading to an increase of political participation. This study focuses on the use of Twitter during 2010 presidential race in Brazil, the country's first campaign with the use of social media. Through an empirical analysis based on a survey applied online, we seek to understand how Brazilian voters had access to political information, interacted with candidates on Twitter and felt about the use of social media during the campaign. Keywords: Social Media; Online Campaign; Elections; Voters Behavior. 1 Introduction Contemporary electoral campaigns have been investing in social network sites as an alternative channel for dissemination of political information. This novel approach enables the integration of political figures in dialogical and non-hierarchical environments, modifying the campaign experience and giving visibility to ideas and political opinions that emerge from virtual social interactions. In this sense, the 2010 Brazilian Elections has marked a shift in the country's political history: Dilma Rousseff became the first woman to run the country, as a representative of the Workers' Party and the Web, especially through social media sites, became a central tool for political campaigns. This paper

2 focuses on this second phenomenon. After Obama's 2008 victory with an innovative campaign that utilized the Web as a central resource and relied on social media to engage citizens and reach voters (Gomes, 2009), politicians and parties from many countries began to recognize the power of the Internet for political purposes. Only a year later, in 2009, the Brazilian Congress rushed to approve changes in the electoral law to enable politicians to broadly use social media sites and to rule on online debates, among other minor modifications. As a result, in 2010 candidates were able to join social media sites, such as Twitter, Facebook and Orkut, to expand the reach of their traditional political campaigns. All the candidates running for President in 2010 had at least one official account on the most popular social platforms - Twitter, Facebook, Orkut, YouTube, among others - during the campaign. Social media sites were also highly promoted by candidates on regular media - serving, for smaller parties, as an alternative to the short TV time granted by law 1 - and printed campaign material. They were central to candidates websites, where citizens could find all official campaign-related profiles. The Web was also used to raise funds for campaigns - although it was not the strongest feature, as Brazilians don't have a culture of donating money to politicians and parties - and the law enabled press-related websites and content portals to provide a broader coverage of the political process. As a result, websites were able to promote and transmit debates between candidates, events that enabled voters to participate by sending questions and comments. Among the strategies that politicians used to reach voters and be more visible online, this paper focuses on the use of Twitter as a campaign tool. In previous research, our goal was to understand politicians' activity on Twitter (Rossini, Leal, 2010; 2011; 2012). Now, we shift to the understanding of how Twitter users perceived the political campaign on the microblogging service, concentrating on whether users chose to follow political profiles and how they accessed political information. Our hypothesis is that Twitter is a relevant tool for political decision-making because it enables people to obtain information directly from the sources and it facilitates information sharing, thus expanding its reach. One important consequence of the use of Twitter, among other networks, should be higher incidence of exposure to political difference as they enable users to maintain broader social connections. Based on Granovetter's studies (1973; 1983), we argue that people who have higher presence of weak ties in their networks have access to different sources of information, and thus are more exposed to political information, in general. We believe that online campaigns that explore social media have a potential to contribute to the political processes in many 1 In Brazil, the law obliges TV and radio stations to provide free advertising time for political parties. Specifically regarding electoral campaigns, the time reserved for Presidential candidates is 25 minutes, divided by candidates and calculated as it follows: 1) a third for each candidate; 2) two thirds according to the party representativeness at the Chamber of Deputies (in the case of coalitions, it is the result of the sum of the number of representatives of all parties that comprise it.

3 ways, such as changing the voters' attitudes towards the political sphere, enabling more participation and access to information, as well as a higher chance of exposure to political difference. In order to properly develop our argument, we will begin with a brief review of the rational (Downs, 1957; Popkin, 1994) and sociological (cf. Zuckerman et al., 2005) theories of voter behavior, concentrating on the importance of social relations and information for political decision-making. Subsequently, we will review contemporary contributions on the use of Internet for campaigns and access to political information (Bimber & Davis, 2003; Brundidge, 2010; Garrett, 2009a, 2009b). On the third section, we will present the analysis of the results of an online survey that we conducted to measure the users' perception of Twitter use during 2010 elections. 2 What motivates political decision-making? Many political scientists have concentrated their efforts on the understanding of citizens' motivations and reasoning regarding political matters. During the 20th century, two of these behavior theories received more attention: the sociological theory, developed at Columbia University during the 1940's, and the rational theory of voting behavior, also known as the economic theory, proposed by Anthony Downs at the University of Michigan in These perspectives provided the basis of Popkin s reasoning theory (1994), which combines assumptions and discoveries of both schools with cognitive psychology to provide an understanding of how people acquire political knowledge to vote. Scholars aligned with the economical point of view of the decision-making process tend to believe that citizens are more influenced by their personal motives (regarding well-being and economic situation) than by the social environment in which they participate. The low interest in political information leads to the reduction of costs for gathering knowledge to make political decisions, therefore, as stated by Downs, "voters will rely on informational shortcuts because they do not have much incentive to gather information about politics solely in order to improve their voting choices" (Popkin, 1994: 13). The Columbia studies present a rather different approach, suggesting that the individual vote will reflect one s social context. According to this perspective, voters decisions are related, among other variables, to their social interactions and the nature and intensity of their political and social relationships, as well as the power of identity with the group. Social interactions, regarding political affairs, have three outcomes: the two individuals involved leave the conversation with different opinions; one of them changes his opinion, aligning with the other's; both involved change their opinions to a third alternative (Figueiredo, 1991: 52).

4 According to Zuckerman (2005), the most important classical study that arose from this perspective was The People's Choice, by Paul Lazarsfeld, Bernard Berelson and Hazel Gaudet. These authors introduced questions about citizens' social relations into electoral polls to understand how voters' social environment and connections can influence their political behavior. Their research demonstrates how personal contacts affect the electoral choices of undecided citizens, whose decision-making processes are influenced by several factors: "flow of communications, in which information flows through opinion leaders; personal contacts need have no particular purpose; flexibility when encountering resistance; rewards of compliance; greater level of trust in the source and persuasion without conviction" (Zuckerman, 2005: 6) 2. The main point of disagreement between Down s economic theory and Lazarsfeld, Berelson, Gauldet s point of view was that the first one described the decision making process at the individual level and considered that the citizen s personal context and perceptions in political and economic terms provided shortcuts for a rational decision. The Columbia proposal was more focused on the social level of voting, describing the decision-making process as a combination of the voter s perceptions and his or her social context and suggesting that the decision at the individual level reflected the group s decision. The social context is rather complex, as it combines relationships among individuals, primary groups and the broader society (Zuckerman, 2005: 8). The sociological perspective is sometimes neglected because political scientists in the US broadly use data from the American National Elections Survey, thus focusing more on party identification, issues and perceptions of the electorate rather than on social context (Zuckerman, 2005). Nevertheless, the social logic of voters behavior still offer powerful theoretical assumptions to explain how inter-personal relationships affect political decision-making processes, especially regarding the use of social media (i.e. services based on social network structures) and a variety of web-based platforms for political purposes that vary from campaigning to social movements and protests. As we intend to show, the use of social media in online campaigns has important implications, such as broader access to sources of information and extended reach especially regarding reaching the less interested part of the electorate. To support our analysis, in this section we will briefly review Popkin s theory on how voters learn what they need to know to vote and some contemporary contributions on the social logic of politics focused on the influence of weak ties (Zuckerman, 2005). 2.1 Voters' rationality: the cognitive psychology approach Samuel Popkin argues that the economic perspective and the sociologic theory are complementary and elaborate his thesis based on contributions 2 Page number provided by Kindle. As the version cited is a digital copy, a more reliable citation would be the "location" (277/5428). To avoid redundance, we will cite the Kindle version according to the page number provided by the software, not the locations.

5 from both. The focus of Popkin's reasoning model is on cognitive psychology, which he believes can fill the gaps left by the two classical studies. The author builds on the idea that people gather political information in their everyday activities, relying on informational shortcuts and cues to aid political reasoning (1994: 15-17). The idea is that decision-making processes are driven by low-information rationality or gut rationality, which basically means that voters economically incorporate what they learn from political campaigns, media coverage, daily life, past experiences and social relationships to reason about politics. The term gut rationality refers to the process of evaluating and framing new information to facilitate choice based on own experiences and beliefs (Popkin, 1994: 212). Similarly to what Downs suggested, Popkin also considers that the costs the investment of time and energy - of being well informed are elevated compared to the expected return of political decisions to the average citizen, especially when compared to other activities directly related to daily life. As a result, the information that people usually have to make political decisions is a byproduct of the data and knowledge that they acquire in their daily lives. Therefore, much of what people use to evaluate the political context is based on everyday observations of the economy of their community and on whether or not they feel safe (1994: 22-24). The media is also a source of information that affects voters directly, when they voluntarily access the news, and indirectly, when they learn about facts and discuss information with their social networks. Daily life information and media information interact because, although we are able to observe the issues that affect our lives in our everyday activities, we need the media to tell us what the government is doing about them. Therefore, much of the importance of media coverage is connecting issues to the government to allow people to evaluate government performance (1994: 27-28). In summary, daily-life and media information are mediated by reasoning and expectations. A vote is more than a direct expression of a voter s social group, pocketbook, or personal problems; voters take account of national as well as personal conditions, and they discriminate among their own problems and problems for the government to address which are relevant to them (Popkin, 1994: 33-34). In the electoral context, political campaigns are important to reinforce the differences between candidates and to provide information about who the candidate is and what he or she stands for. They are also important because people don t usually pay attention to politics, therefore they do not know much about what the government is doing. People make projections based on what they are able to absorb from campaigns they seek for identification and tend to choose the candidate who seems aligned to their perspectives (Popkin, 1994: 38-43). In this sense, contemporary campaigns are shifting from the ideology-based program to the candidate-centered model that focuses on the candidate or his or her personal credentials and background to convince voters. The idea is to provide shortcuts for voters to project their expectations on candidates who they perceive are aligned with their beliefs (Leal, 2002).

6 As people have to make political decisions based on incomplete information, because the costs of being well informed are high, they rely on information shortcuts to improve reasoning. The notion of information shortcuts refers to cues from daily life, media, past experiences and political campaigns that people incorporate to reason. People use shortcuts to evaluate, obtain and store information, delegating the effort of being well informed to others (media, friends, family, and so on). The notion of low-information reasoning does not mean that people devoid substantive content to make decisions (1994: 44). The opinion of other people is an example of a shortcut for evaluating information: given that the average voter seeks to reduce the costs of political information, he trusts others whom he believes are more informed and interested in politics to access analysis of the political context. In this sense, the individual is influenced by the perspectives of his social ties (strong and weak) and also by personalities or opinion leaders who comment on political environment in the media. Therefore, friends, family and co-workers mediate one s perception of the political facts. As Popkin (1994) argues, there are many factors that influence the average voters decision-making processes. According to this theoretical framework, people learn what they need to know to make reasonable political decisions in their daily activities, talking with others, following the news and campaigns. The importance of the social network is to provide evaluation of political content on daily conversations about the news, the community and topics of mutual interest, such as security and the economy. In general, people know little about the government s affairs and competences, and therefore delegate to others and the media the effort of connecting the political sphere to perceived issues and evaluating competences. 2.2 The effect of weak ties on political behavior The social logic of voters behavior relies on the idea that the vote represents the individuals social context and his or her belonging to groups or communities. As our research focuses on the users perception of the online campaign on Twitter an information network that relies on social connections -, we believe that the sociological approach provides valuable cues to understand how the use of web-based social platforms affects the voters access to political information. Contemporary studies focused on the influence of weak ties social relations characterized by low intimacy and affection, such as colleagues, co-workers and friends (Granovetter, 1973) on political behavior show that this type of relationship affects the voter in different ways. We will briefly review some of these contemporary contributions. Considering that weak ties are predominant on social media sites, where one is able to maintain broader social networks, the sociological approach enables us to understand the importance and the function of those ties maintained through social media on the construction of political opinion during Brazilian s 2010 presidential race.

7 Before presenting the review, it is important to explain that the terms weak ties and strong ties, broadly used by political scientists, refer to Mark Granovetter s 1973 study of the importance of social connections on daily life. The strength of an interpersonal tie is defined by Granovetter as a combination of the amount of time, the emotional intensity, the intimacy (mutual confiding), and the reciprocal services which characterize the tie (1973: 1361). Strong ties are those between close friends and family, while weak ties refer to casual relationships, such as co-workers, acquaintances, friends of friends and so on. Regarding one s social network 3, a strong tie refers to people with whom one interacts frequently and intimately. Strong ties tend to form a highly connected group where people know one another, although they have few contacts not tied to ego (the center of the network). Weak ties form bridges between closeknit groups because they represent one s contacts that are not tied to one another, but to individuals with whom one is not connected. It is through weak ties that one can reach indirect contacts (i.e.: people who one does not know). (Granovetter, 1973: ). Therefore, it is through weak ties that ego, the center of the network, will have more opportunities to access information and knowledge of the world beyond his friendship circle. For political purposes, strong ties usually reinforce one s beliefs and perspectives, as people tend to relate with those they share socio-economic situation, geographical proximity, interests and beliefs. Weak ties, in this sense, provide broader access to information that is not directly related to one s beliefs and therefore provide broader knowledge of the political environment beyond one s social context and preferences. 2.3 Contemporary findings on the social logic of voters behavior In his 2005, Jeffrey Levine presents an analysis of the influence of strong and weak ties on political behavior and concludes that social network members exert a direct and consistent influence on key political decisions that is not restricted to closer relationships (strong ties). In his findings, Levine shows that, although individuals are more susceptible to the influence of their relatives, people with whom they have weak ties (colleagues, acquaintances and co-workers, for instance) also exert influence on the decision-making process. While intimates can exert influence because they are trusted, frequently encountered and often have knowledge of politics, several studies previously conducted by Huckfeldt and colleagues discovered that close friends could have a weaker effect on voting than less intimate friends and regular contact. They also found that citizens frequently engage in political discussions with regular contacts that are not intimate (Levine, 2005: 139). 3 Generally called ego-centered network because the analysis considers an individual (the ego ) as the center of the network and classifies strong and weak ties according to their relationship with ego.

8 Levine s analysis shows that, compared to just friends and regular contacts, close friends are less frequently interacted with and less likely to be viewed as politically knowledgeable (2005: 140). The results suggest that nonrelated intimate relationships do not posses the characteristics 4 considered to be important for social influence. Therefore, social network members exert a direct, powerful, and consistent impact on the choices of modern citizens (ibid. 146). The author believes that there is evidence that modern American citizens engage in real debates, where opinions are expressed, challenged and sometimes altered. Choice is likely to be influenced and altered in discussions with network members who carry novel and often contradictory information. Weak ties are more likely to carry this type of information than strong ties. Laurence Kotler-Berkowitz presents a different approach, showing how diverse friendship networks can affect political participation. He applies findings of the sociological literature, such as Granovetter s 1973 cited above, that supports the thesis that larger social networks provide benefits such as broader access to information and opportunities in politics, suggesting that social networks that are more diverse, especially relative to one s total network size and one s own social characteristics produce higher levels of political participation (2005: 152). Kotler-Berkowitz states that the participatory influence of social ties works through three factors: information, recruitment and mobilization. The effects of diverse social networks also occur because many opportunities for political participation are structured around social groups defined by traits like race, ethnicity, class, religion and sexual orientation (2005: 152). In his empirical analysis, the author proves that greater diversity of friendship across multiple group boundaries and diversity related to one s social characteristics and network size are positively related to greater levels of participation. Broader social networks have more weak ties, therefore being more susceptible to accessing different information (regarding social context and preferences). They also provide higher chances of recruitment into political participation, as a diverse set of friends exposes people to various opportunities and requests for political participation in many different contexts, such as group membership and community leadership circles (2005: ). Friendship diversity also increases engagement in non-electoral political acts. Kotler-Berkowitz s findings are consistent with previous research, and lead to the conclusion that, In politics, as in job hunting, community organizing, corporate management and healthcare, social networks that incorporate a diverse set of people provide greater access to information, opportunities for social action and personal advantages. Political behavior, like behavior in other fields of human endeavor, responds 4 Levine considers four characteristics that can be assessed in large opinion surveys and that have been shown to encourage social influence: correct perception, frequent political discussion, absence of political disagreement and perceived political competence (2005: 140).

9 fundamentally to the composition of our social contexts and networks (Kotler-Berkowitz, 2005:167). Levine s and Kotler-Berkowitz s findings, along with many others (see Zuckerman, 2005), provide support for our hypothesis that social media use can influence decision-making processes by providing greater access to political information, spontaneously or not, during campaigns. As we will show, virtual platforms structured by social connections allow people to create, maintain and interact with larger networks. Therefore, information accessed through these virtual environments should be more diverse and often contradictory than what people encounter in their real-life political discussions. We believe that social networks (and social media) are especially relevant during electoral campaigns, when less interested citizens need to engage in politics to decide their vote. 3 Internet use and political heterogeneity: evidence and possibilities The possible impact or the effect of the Web over political affairs has been a subject of study for many researchers in the fields of Political Communication. To cite some Brazilian researchers, there are studies focused on the Web as a public sphere and the possibilities of this medium for enhancing democracy (Gomes and Maia, 2008); the effect of the virtual environment on political participation (Aggio, 2011; Marques and Maia, 2010); political engagement and deliberation online (Gomes, Maia & Marques, 2011); and finally the use of social media by political campaigns (Aggio, 2011; Gomes et al., 2009; Rossini and Leal, 2010; 2011; 2012). Being a highly segmented, non-hierarchical, interactive and purposive media (i.e. the receipt of information is a result of the users choices), the Internet dramatically changes how people access media and therefore represents a challenge for political parties and campaigns as it provides its users with access to multiple sources of information (Bimber and Davis, 2003: 5-7). In this sense, recent findings by Kelly Garrett (2009a, 2009b) on what motivates the consumption of political information online and Jennifer Brundidge s inadvertency thesis (2010) support our hypothesis by proving that, although people tend to seek information online that reinforces their beliefs, as predicted by the selective exposure thesis (Sears; Freedman, 1967), it does not mean that they avoid differences. Rather, the virtual environment affects the overall heterogeneity of people s political discussion network. Both Garrett and Brundidge tested the selective exposure thesis in their studies of how people access information online and how they respond to political differences. The selective exposure thesis assumes that people will seek information that reinforces their beliefs and will systematically avoid challenging opinions. Applying these assumptions to the Internet, a negative perspective suggests that when people can control and filter information, they will increasingly seek like-minded others and access consonant information,

10 thus avoiding challenge and heterogeneity. However, a positive approach suggests that the Internet contributes to democracy because people are more exposed to political differences through the weakening of social boundaries and the bridging of geographical divides (Brundidge, 2010: 681). Kelly Garrett investigated how Americans access political information online, considering that this medium provides users with novel mechanisms for controlling content and selecting sources when compared to traditional, oneto-many media. His findings are consistent in two different studies (2009a, 2009b): although people indeed tend to access biased information that is consonant with their beliefs, they do not reject or avoid difference. The exposure to political difference is not reduced by selective exposure. Garrett s proposal for reframing the selective exposure debate is based on a study he conducted on online media consumption during 2004 US national elections. His analysis of cross-national data shows that individuals with strong party identification access more biased sources than those who have weaker identification. Nevertheless, those who seek reinforcement do not avoid political differences, contrary to what would be expected according to one of the assumptions of the traditional selective exposure perspective - that people who seek reinforcement will be less exposed to heterogeneity (2009a: 692). Data also shows that people who access political information online also follow campaigns on offline sources (2009a: 690). Garrett concludes that, given more control, people tend to choose sources of political information that are consonant with their beliefs. However, this attitude does not affect exposure to political differences. In fact, as mainstream news media are the preferred source for 83,9% of Americans and only 10,8% of Internet users claim to prefer alternative news media, it does not appear that these users are trying to avoid sources that include other viewpoints (2009a: ). The author concludes that people who access political information online tend to have more knowledge overall, that is, they know more about their own beliefs and also about challenging perspectives. A qualitative web-administered behavior-tracking study coordinated by Garrett (2009b) provides further evidence that the desire for opinion reinforcement does not reduce contact with political difference. In this case, participants were volunteers who had the habit of accessing political news online. They were exposed to a high number of news articles and they had to choose what to read, and evaluated the articles before and after reading. The experiment also tracked time spent reading each article. The results show that people will preferably choose to read information that is consonant with their interests, but the aversion to opinion challenging news is weak. Overall, people spent more time reading news contrary to their viewpoints an outcome that could be explained by the need for learning and evaluating novel arguments and the interest in feeling better informed about the scenario in order to engage in discussions (2009b: ). Garrett discards the hypothesis that consumption of controlled information on the Web would have negative consequences from the standpoint of tolerance for political difference.

11 Jennifer Brundidge develops an alternate explanation about the contribution of the Internet to the heterogeneity of political discussion networks, which she names the inadvertency thesis. From this standpoint, the Internet contributes to heterogeneity of political discussion because of its weaker or blurred boundaries between informational and interactional environments, and opportunities for interpersonal communication among weak ties. Inadvertency is facilitated online through (a) less than perfect online selective exposure strategies, (b) nonavoidance of encounters with political difference, and (c) weakened social boundaries between far flung geographic locations, between one discursive space and the next (blurred and porous boundaries creating interspatiality ), between political and apolitical spaces of communication, and between the private and the public spheres. (Brundidge, 2010: 687). Because people can easily traverse across communication spaces, the processes of receiving, interpreting and discussing news have blurred boundaries. People access and interact with content in various ways: they can comment on what they have read directly at the source, share links on a social network site, discuss on blogs and forums, and so on. Brundidge uses nationally representative data of the US to test her inadvertency thesis and finds that online political discussion and online news use (directly and indirectly, as it also affects political discussion offline) contribute to the heterogeneity of political discussion networks beyond the influence of offline discussion and traditional media. This does not mean that people seek political difference online, but that they find heterogeneity inadvertently (2010: ). The inadvertency exposure thesis is consistent with our proposal to investigate the relevance of Twitter for accessing political information and encountering diversity online, as its users are generally connected to a chosen set of sources (where weak ties are predominant) from which they receive information. As opposed to candidate and party web sites, where a common practice of the campaign staff is to preach to the converted to try to elevate the commitment of those converts (Bimber and Davis, 2003: 104), social network sites and social media are platforms where people interact with heterogeneous contacts and therefore are more susceptible to receive all kinds of information. The exposure to political information in these spaces should occur both by selection (i.e. the choice to connect with sources) and by inadvertency, as one cannot predict what kind of information flows through his social ties. 4 Twitter use during Brazil s presidential race: the voter s view Cheaper computers and communication technologies, broader mobile access and Internet use on universities, community centers and workplaces are some factors that explain the increase in the number of Internet users in Brazil.

12 Although national data show that this access is still stratified by age, education and income, these differences are decreasing over the time. According to the Brazilian Institute of Geography and Statistics (IBGE National Household Sample Survey, 2009) more than 68,6 million people had access to a computer at home in 2009, while 53,6 million of those also had access to the Web. The social difference becomes clear when family income is taken into account: Internet at home was a reality for 92% of families with monthly income above 20 minimum salaries, 78% of families with monthly income between 10 and 20 minimum salaries and 22% of families with monthly income below 10 minimum salaries (IBGE National Household Sample Survey, 2009). The Web was found to be the second main source for voting decision making (12%) in the 2010 elections, only behind TV, which was preferred by 72% of the electorate according to a cross-national survey conducted by Ibope (Rezende; Chagas, 2011). It represents an increase of 10% when compared to data from the same institute in When education and wealth are considered, the importance of Internet increases to 33% for electors with college education and 29% for people whose family income is above five times the minimum salary 5 (Rezende; Chagas, 2011: 4). Data show that 25% of the electorate accesses the web every day, which means that 27 million people are constantly connected. Their declared activities on the Web were divided as follows: news consumption (30%-33%), (27% - 30%), instant messengers (25%-28%), social network sites (22%-25%) and maintenance of personal weblogs or sites (4%) (Rezende; Chagas, 2011: 6-8). Findings in Brazil are consistent with data from US 2008 elections provided by Pew Research Center and represent strong evidence of the growing importance of the Web for political purposes. While data provided by Pew Reseach provide more insights into the political uses of the Web during campaigns and Ibope focuses on general media use. According to the Pew report, 46% of Americans accessed the Internet to obtain information and share impressions on the 2008 presidential campaign, while 35% watched political videos online and 10% engaged in political activities through social network sites. The use of social network sites and social media in contemporary political campaigns is a consequence of the growing amount of time citizens spend on these platforms. Social media sites, like Twitter, insert the political sphere into a non-hierarchical interactional space that allows people to communicate with candidates and receive information without media interference. Moreover, it allows citizens to actively participate on campaigns by creating, engaging and sharing content on the network. As Twitter users are connected to many others through one-sided ties (one can follow others without being followed back), relationships are predominantly weak. The environment is thus susceptible to political 5 The minimum salary is fixed by the government and represents the minimum income allowed for those who are registered employees. In 2010, the amount was R$ 510.

13 heterogeneity because, even though people choose to connect with others according to particular interests (selective exposure), they are susceptible to receiving political content through their weak ties. It happens because every Twitter user is a bridge between the two segments of their networks, as we will explain below (see Rossini and Leal, 2011, for further discussion). It is important to consider the contextual differences between the US electorate and the Brazilian electorate, especially regarding partisanship and political participation. In Brazil, party identification and political interest are low, and people tend to have a negative perception of their political representatives (Aldé, 2004). As such, they should have low incentive to seek political information and therefore are susceptible to information shortcuts (Figueiredo, 1991; Popkin, 1994; Leal, 2002) to learn about politics and make political decisions. Community, relationships and media, in this context, should be more reliable shortcuts than party identification and campaigns. We believe that social media and social networks are becoming relevant for research in political communication because these structures constitute rich spaces of human interaction that may be relevant for many electors as a source for political decision-making processes. Our hypothesis is that social media sites are important variables for political campaigns because of three main characteristics: (1) they provide electors with multiple sources of political information; (2) allow them to interact directly with political agents that use these platforms and (3) increase the chances of inadvertent exposure to political information through weak ties. To approach the extent in which social media sites, particularly Twitter, influence the electoral decision, we conducted an online survey to understand how citizens used and perceived social media use during the 2010 elections in Brazil. 4.1 Research questions and methods There are many ways in which Twitter use can influence voters access to political information. First, Twitter allows users to follow their interests through one-sided connections, meaning that they will receive information from the sources they chosen. A user network on Twitter is thus divided into two segments: the profiles that he chooses to follow (his information sources) and the profiles who have chosen to follow him (for whom he acts as an information source). Second, Twitter networks are mainly formed by weak ties as a real time information network 6, connections are driven by people s interests and have low reciprocity. According to Kwak, Lee, Park and Moon, who analyzed 41.7 million user profiles, 1.46 billion social relations and 106 million tweets, only 22,1% of the connections were reciprocal (2010: 3). When we combine the high presence of weak ties with the finding that the average distance between any two users of the network is 4.12 (Kwak, Lee, Park and Moon, 2010: 4), meaning that the information is to flow over less than 5 hops between 93,5% of user pairs, we can infer that even those who do not follow political agents, candidates or parties are susceptible to receiving political information through 6

14 their weak ties, who can retransmit (retweet) to their followers content shared by their sources, thereby acting as bridges between their followers and their chosen sources. Our questions focused on two main purposes: (1) interest in political campaigns on Twitter and access (direct and indirect) to political information through the network, and (2) perception and evaluation of candidates performance on Twitter. The online survey was hosted online by Google Docs from 14 July 2011 to 11 August 2011 and had 559 respondents. The sample was comprised of Brazilian Twitter users who had volunteered to respond to 12 multiple-choice questions, designed to access their habits regarding access to political information on Twitter and perceptions of candidates performance. Respondents were recruited through Twitter, where many users broadly disseminated the link for the survey across the country, including TV personalities and well-known journalists, as well as corporate profiles. As a convenience sample, it did not allow us to make accurate inferences that extend to the population as a whole, although some results are consistent with data from cross-national samples and may help understand the profile of voters who seek political information through networks. 4.2 Analysis The group of 559 respondents that constitute our sample is demographically described as follows: Table 1: Age x Gender Age-range Men Women Total (27%) (52)% (13%) (7%) (1%) Total 333 (60%) 226 (40%) 559 Table 2: Education levels (%) Higher education 59% Incomplete superior education 33% High school 7% Incomplete high school 1%

15 Primary education - Incomplete primary education - Total 100% These distributions reinforce two previous expectations derived from the literature review (Bimber and Davis, 2003 Popkin, 1994): men represent the majority when it comes to political discussion and there is a strong presence of individuals with higher levels of education. Nevertheless, there is evidence that relatively contrasts the supposition that younger people represent the majority of social media users, as most of responses are concentrated in the age range of years (52%). A reflection that emerges from that evidence and should be further explored is: the so-called new communication technologies, expression still widely used these days, are becoming older every year that passes, and generations that are included in those environment are increasingly becoming older. The gender differences reflect the institutional world of Brazilian politics, which is still almost entirely male. Although the president is a woman, the presence of women in other political bodies such as state government, legislative chambers, the senate, city halls and so on is small, compared to the number of female voters. In this scenario, the majority of respondents are male (60% x 40%)which is convergent with the expectation that more men will have participated in the survey as we are talking about a topic still socially associated with the masculine world As for the association between higher educational levels and broader use of digital environment, cross-national data cited above indicate that, in spite of the total growth of Internet use, its access is still elitist for several reasons, since the asymmetric training to use these technologies point to the high statistical correlation between higher education and higher income (which means easier access for the more educated / rich). However, our main interest is in understanding how respondents use the Internet to build their political opinions and make electoral decisions. In this sense, table 3 summarizes information sources on the web that people accessed during the presidential campaign in People could check more than one option, therefore percentages add up more than 100%. Table 3: access to political information online Did you access political information online? How? (allows more than one choice) News sites, websites of newspapers and magazines 55% Parties and candidates websites 22% Social Media (Twitter, Facebook, Orkut, etc) 48%

16 News sites and social media 55% I did not access political information 6% Data indicate that the combination of traditional media websites (newspapers, magazines, etc) and web portals; as well as the combination between traditional media websites and social media (both with 55%) were the main choices, although a significant number preferred only social media (48%). Most respondents have chosen more than one option. The fact that only 22% opted to access information directly from parties and candidates websites is revealing. A possible interpretation of this result is that electors generally mistrust discourses produced by political institutions instead, they prefer to access information interpreted by journalists in news sites and from their networks in social media, seen as more convincing and relevant. If we control for unique sources of information, the results for parties and candidates sites were even lower: only 7 respondents chose it as the only source of political information online, while 58 marked social media and 89, news sites and traditional media websites. This evidence confirms the theoretical assumptions discussed in this paper: even though they are constituted mainly of weak ties, contacts maintained through social media are relied upon by a significant number of users as information shortcuts for decision making processes. References acquired from these sources - news sites, social media, virtual friendships are considered as key variables when electors are deciding their vote. Another dimension in which survey data converge with the theoretical discussion concerns the potential for interaction provided by virtual networks: as table 4 indicates, most respondents accessed political information regularly. Among others, however, the larger group consists of those who, despite not having interest, become involved with the subject in conversations with others. In the social network structure, even if the user s goal is not originally a political discussion, it emerges from his connections, and thus allows these issues to reach individuals or groups who would not voluntarily seek political information, but are exposed to it inadvertently. In consonance with the inadvertency thesis (Brundidge, 2010), it is correct to assume that digital environments increase the likelihood of being exposed to political issues, whether or not they are originally interested in these issues when they use Twitter, Facebook and so on. As news portals are accessed by 76% of the sample and social media by 69%, indicating a complementary use of these sources, we can infer that respondents are more susceptible to receiving political information during the campaigns than those who did not use the Web in that period. Table 4: interest in political information Are you interested in political information?

17 Yes. I access political information regularly. I am only interested in political information during the elections. I am not very interested, but I talk about it with family and friends during the elections period. I am not interested and I do not seek for political information in order to make my vote decision. 72% 11% 15% 1% Table 5: sources of political information How do you consume/access political information? (allows more than one choice) I watch the news on TV 57% I read newspapers and magazines 47% I read news online in portals 76% I read blogs 37% I follow information through social media (Twitter, Facebook, Orkut) 69% Table 5 shows the simultaneous consumption of different information sources. As people could mark more than one option, there were 30 different combinations of response and percentages exceeded 100%. It is clear that social media, a tool that enable direct access to political actors, is combined with traditional mediated sources, where information is processed by journalists and media institutions. Specifically on Twitter, the majority of respondents point that their motivation to follow candidates on the network was to receive information directly by the candidate (37%). In other words, despite this answer does not indicate that people trust politicians, it reveals the potential of disintermediation represented by technologies based on network structure. It is no longer necessary to wait until mediating institutions (journalists, press, TV stations) to address an issue: direct contact becomes possible, even though, as demonstrated in table 3, political sources are seen as less trustworthy than traditional media institutions. Tables 6 and 7 below summarize this point: even though many users follow profiles to receive direct information, there is a low level of effective interaction. Users clearly perceive that, despite the potential for dialogue, candidates profiles are merely tools of electoral strategy, aiming at maximizing votes, and do not represent an exercise of effective interaction between the political sphere and the citizens. The importance of weak ties is

18 also evidenced in table 5 above: among those who did not follow any candidate on Twitter, 14% received political information through their weak ties another evidence of the inadvertency thesis explained above. Table 6: candidates on Twitter Did you follow any presidential candidate on Twitter? Why? Yes. To receive information directly from the candidate. Yes. To receive information and interact with candidates. Yes. And I also engaged in campaign activites on Twitter. No, but I received information about candidates through people I follow. No, but I followed other news profiles in order to receive political information. No, and I did not receive political information from other people I follow. 37% 14% 16% 14% 11% 9% Table 7: Interaction on Twitter Did you interact with political profiles on Twitter? Yes, I interacted and was responded 24% Yes, I interacted but I was not responded. 20% No. 56% Findings are in consonance with a previous study, focused on four Latin America presidents and their Twitter profiles: political agents that are present on social media sites are more interested in publicizing their actions and consolidating their images than in engaging in dialogue with other users (Rossini; Leal, 2012). Although structures based on the network architecture are potentially interactive and enable non-mediated contact with sources, being appropriate to a participatory environment, evidence shows that political agents are not interested in interacting. Instead, they would rather use new tools to reproduce old practices: they want to be seen, to be heard, to be popular and visible in the network, and they use their profiles to spread their ideas, to consolidate their political image and to publicize their perspectives apart from the media coverage (Rossini; Leal, 2012: 14-15). Table 8 indicates that people tended to follow more than one candidate, as the question allowed more than one choice and the three main candidates had a similar percentage, 45% for Dilma Rousseff (the elected candidate), 47% for José Serra (recurring contestant) and 49% for Marina Silva (first-time