Princeton Studies in Global and Comparative Sociology

Social Communication and Voting in Latin America

Andy Baker, Barry Ames, Lúcio Rennó

• Publisher: Princeton University Press
• Year: 2020
• ISBN:
  9780691205793
• Topic: Social Sciences
• Subtopic: Sociology
• Index: Social Sciences

How voting behavior in Latin America is influenced by social networks and everyday communication among peers In Latin America's new democracies, political parties and mass partisanship are not deeply entrenched, leaving many votes up for grabs during election campaigns. In a typical presidential election season, between one-quarter and one-half of all voters—figures unheard of in older democracies—change their voting intentions across party lines in the months before election day. Advancing a new theory of Latin American voting behavior, Persuasive Peers argues that political discussions within informal social networks among family members, friends, neighbors, coworkers, and acquaintances explain this volatility and exert a major influence on final voting choices.Relying on unique survey and interview data from Latin America, the authors show that weakly committed voters defer to their politically knowledgeable peers, creating vast amounts of preference change as political campaigns unfold. Peer influences also matter for unwavering voters, who tend to have social contacts that reinforce their voting intentions. Social influence increases political conformity among voters within neighborhoods, states, and even entire regions, and the authors illustrate how party machines use the social topography of electorates to buy off well-connected voters who can magnify the impact of the payoff. Persuasive Peers demonstrates how everyday communication shapes political outcomes in Latin America's less-institutionalized democracies.

PART I

Introduction and Descriptive Background

1

Social Communication and Voting Behavior

The gift of speech … proves that man is a more social animal than the bees.

—Aristotle¹

THE PUZZLE OF VOTE CHOICE IN BRAZIL

Even by Brazil’s lofty standards, the October 2014 presidential campaign was exciting. Though surges by outsider candidates and other kinds of momentum swings have been frequent in the country’s recent presidential contests, the 2014 contest contained all the drama of a telenovela and a closely fought soccer match. Early in the year, President Dilma Rousseff of the left-of-center Workers’ Party (PT) held a commanding lead in the polls, a lead so safe that she was seemingly on course to win in the first round of Brazil’s majority runoff system (see figure 1.1). In a distant second place was Aécio Neves of the center-right Brazilian Social Democratic Party (PSDB). As in Brazil’s previous five elections, it thus seemed that the top two finishers would come, despite the country’s 30-plus political parties, from these two parties.

Once campaigning had officially begun in July, however, Eduardo Campos of the Brazilian Socialist Party (PSB) began to surge in the polls. After nearly a month of steady gains, Campos was nipping at Aécio’s heels for second place, contending for a spot in the now increasingly likely second round. Tragically, Campos died in a plane crash on August 13, but his running mate, Marina Silva, immediately grabbed the party’s baton and continued its ascent. Marina was the ideal replacement.a She was a known quantity who four years earlier had earned 19 percent of the first-round vote atop her own presidential ticket (for a different party). Capitalizing on a surge of sympathy for the deceased Campos, she zoomed past Aécio in the polls. By early September, with just a month remaining, Marina had a 20-point lead on Aécio, even reaching a statistical tie with the front-runner Dilma in a few polls.

FIGURE 1.1 The Evolution of Vote Intentions for Major Presidential Candidates in Brazil 2014

Notes: Each line (LOWESS-smoothed with a bandwidth of .25) summarizes results from nationally representative polls and represents a candidate’s estimated percentage of valid vote intentions (dropping undecided respondents). Actual election-day results are indicated with circle markers. Only the top three finishers are depicted.

Sources: Datafolha, IBOPE, Vox Populi.

Then Marina’s fortunes began to unravel. No single gaffe or mistake triggered her slide. In fact, Marina ran a disciplined campaign. Over the final 30 days, however, her support cratered almost as quickly as it had risen. By the eve of election day, she had fallen into a tie with Aécio in the contest for a second-round spot, and voters went to the polls on October 5 genuinely uncertain which of the two would earn the right to face the incumbent Dilma in the second-round election (also “runoff”). Was Brazil on the verge of producing the world’s first female-only presidential runoff? Or would the country see its fourth consecutive PT-PSDB runoff?

In the end, the result was another PT-PSDB runoff, and it was not even close. Aécio defied the final polls by a wide margin, beating Marina by 12 points and finishing closer to Dilma than to Marina. After peaking near 40 percent of vote intentions in early September, Marina fell to just 21 percent on election day, only a minor improvement over her performance four years earlier. The second round began in a tie, but Dilma opened a small lead toward the end of the three-week campaign period and, in Brazil’s closest-ever presidential contest, squeaked out a three-point victory.

This horse-race intrigue poses a major puzzle about the nature of Brazilian voting behavior and election outcomes. On the one hand, the election featured a huge and fascinating amount of change in voter preferences. It is clear from figure 1.1 that many voters were changing their minds as the contest unfolded. By our own panel-data estimates, at least 40 percent of the electorate shifted their vote intentions across party lines at some point during the campaign. This represents a high degree of volatility by international standards. On the other hand, this volatility ended in a predictable outcome. Aggregate vote intentions before the campaign’s start were similar to the eventual first-round vote totals. Moreover, because pre-campaign vote intentions favored the two traditional parties, the campaign restored os mesmos de sempre, the same ones as always, to their usual places in the second round. What explains this aggregate predictability despite the high rates of individual-level voter volatility?

Some Incorrect Answers

To scholars of US campaigns and voting behavior, this puzzle of short-term change amidst medium-term stability has a familiar ring to it: enlightened preferences.² During political campaigns, many US voters vacillate between indecision and a particular candidate, thus making aggregate vote intentions somewhat volatile. Election outcomes, however, are highly predictable based on the political and economic fundamentals (e.g., macroeconomic health and the distribution of partisanship) in place before a campaign’s start. Moreover, virtually all US voters end up choosing the candidate in line with the partisan predispositions and group affiliations they held before the campaign began.³ Thus in the US a campaign serves to bring voters home to the partisan choice that has lurked beneath the surface all along—the one based on their most informed, and thus enlightened, preferences.

Although a powerful predictor in the US context, this explanation does not fit the Brazilian case. Because Brazil is a new democracy with an extremely high number of parties, Brazilians’ partisan predispositions are weak.⁴ In any given cross-sectional survey in 2014, just 15 to 20 percent of the electorate identified with the PT, less than half of Dilma’s first-round total.⁵ Panel data show that only half of Dilma voters declared themselves to be PT sympathizers at some point during the campaign, and the PT is the party that had Brazil’s largest base of citizens. Aécio’s PSDB and Marina’s PSB had virtually no mass bases. Just 13 percent of his voters and 7 percent of hers declared themselves to be PSDB or PSB sympathizers, respectively, at least once during the campaign. Clearly, most voters were not returning to an underlying partisan leaning over the course of the campaign. Moreover, the sorting of voters into pro- and anti-incumbent camps based on stable economic perceptions cannot explain the volatility: most of the switching occurred between the nonincumbent candidates, Aécio and Marina. Finally, unlike US voters, Brazilian voters do mull different party choices, as evidenced by the 40 percent who crossed party lines. They do not simply end up where they began. The group of individuals lined up to vote for Aécio at the campaign’s start was very different from the group that ultimately voted for Aécio on election day. In summary, the enlightened preferences argument does not export well to Brazil.

Strategic voting poses another possible answer to the puzzle. Perhaps the high degree of volatility arose as voters, and particularly strongly anti-Dilma voters, sought to coordinate around the most viable opponent to her candidacy, an instinct that would maintain Brazil’s stubborn two-party duopoly in presidential elections.⁶ Yet this hypothesis contains a logical failure: Brazil’s president is chosen by majority, not plurality, rule. Two-round systems with strategic voting should reduce to three-party, not two-party, competition. In other words, there was no strategic reason for sincere Marina supporters to abandon her for Aécio, especially when she had a 20-point lead on him. At that point, the contest had reduced to a battle between the two for a second-round slot. Moreover, polling results for simulated second-round scenarios showed Marina to be the better competitor against Dilma. Strategically minded Dilma-haters should have voted for Marina.

Perhaps Aécio rode a wave of vote buying into the second round. After all, he had the backing of most of Brazil’s conservative, and largely clientelistic, establishment parties, while Marina had virtually no machine infrastructure supporting her. Yet no empirical evidence or even speculation exists that a vote-buying operation of this scope occurred on Aécio’s behalf. It would be a hard operation to keep secret, as Marina lost an estimated 20 million votes, nearly one-fifth of the electorate, in five weeks. In fact, research on Brazil shows that targeted vote-buying attempts have limited relevance for presidential politics because of the country’s federalized and fragmented party system. Almost all vote-buying and party-contacting efforts are made on behalf of municipal and legislative candidates,⁷ and these lower-level candidates avoid making upward endorsements of presidential contenders who are lagging in the polls.⁸ Evidence from the broader literature on Latin American clientelism shows that vote-buying attempts are ineffective at producing this amount of preference change because parties do not even target swing voters.⁹

Perhaps the solution to the puzzle lies in the mass media and the relative balance of partisan information broadcast during the campaign. On this front, Marina’s campaign did face a major disadvantage. In Brazil campaign commercials in the traditional electronic media (television and radio) are strictly regulated. The Federal Electoral Tribunal prohibits commercials before the final seven weeks of the campaign (figure 1.1), and airtime is allocated to candidates in proportion to the size of the legislative coalitions backing them. Hailing from a smaller party, Marina received half the airtime that Aécio did, and just one-sixth that of Dilma. Media consumption thus exposed citizens to relatively little pro-Marina content, potentially spelling her eventual demise. Two facts, however, undermine this hypothesis. First, figure 1.1 shows that Marina’s surge continued several weeks after electronic media campaigning began, a fact leading some observers to conclude at the time that the amount of media exposure was unrelated to success.¹⁰ Second, among voters who had expressed a mid-campaign intention to vote for Marina, those least exposed to campaign commercials were more likely to abandon her, according to our panel survey data. In other words, attention to electronic media correlated negatively, not positively, with changing one’s mind. Direct exposure to the partisan balance in campaign commercials does not explain vote switching.

Socially Informed Preferences

Our solution to the puzzle lies in an arena of political life that scholars of Lat...

Table of contents

• Cover Page
• Series Page
• Title Page
• Copyright Page
• Dedication
• Contents
• List of Illustrations
• List of Tables
• Preface
• Acknowledgments
• List of Abbreviations
• Part I: Introduction and Descriptive Background
• Part II: Social Influence and the Vote
• Part III: Implications of a Horizontally Networked World
• Appendix A: Statistical Results
• Appendix B: Measurement of Variables
• Appendix C: Details of Correct-Voting Analyses
• Notes
• References
• Index