Since the publication of The American Voter (Campbell et al. 1960), political scientists have regarded positive partisanship as critically important – almost a “master variable” for explaining political participation, vote choice, candidate evaluation, and attitudes toward public policies. By contrast, relatively little research on negative partisanship exists. Does negative partisanship have the same effects in other countries as it does in Brazil?

This chapter demonstrates that our claim about the importance of negative partisanship can be generalized. Individuals can have either positive or negative partisan attitudes – or both – in any country around the world, and such attitudes play an important role in vote choice. Scholars of parties and voting behavior who ignore the independent role that negative partisanship can play are missing a potentially large part of the story about individual motivation to vote or otherwise participate in politics.

Most research on negative partisanship has been conducted in a single country (e.g., Maggiotto & Piereson 1977; Caruana, McGregor, & Stephenson 2015; McGregor, Caruana, & Stephenson 2015) or in countries with single-member districts (Medeiros & Noël 2013). (Exceptions include Rose … Mishler [1998] and Vlachová [2001]). Our exploration of Brazil suggests that antipartisanship can matter even in multiparty, proportional representation (PR) electoral systems. Mayer (2017) is the only large cross-national study that we know of. She used surveys from the third module of the Comparative Study of Electoral Systems (CSES) to explore negative partisanship in seventeen European PR systems.

Our approach here is similar to Mayer's, but differs in three ways. First, we incorporate a geographically broader sample, analyzing countries from Latin America, Asia, and North America as well as Europe. Second, we explore negative partisanship at two distinct points in time – about twenty years apart, using CSES surveys from the mid-1990s and the mid-2010s. And third, our conceptualization of negative partisanship departs from Mayer's. Whereas she examined the concurrent effects of positive and negative partisanship on individual voting behavior, allowing individuals to posses both positive and negative partisan attachments, in keeping with earlier chapters in this book we explore whether negative partisanship alone can predict the behavior of individuals who lack a positive partisan attachment.

About half of Brazilians are or have been either partisans or antipartisans. To what extent have these positive and negative partisan attitudes shaped voting behavior?

Every four years, on the same day in October, Brazilians cast ballots for president, senator, governor, federal deputy (members of the lower chamber of the national legislature), and state deputy (members of state legislatures). Scholars have already revealed a great deal about the factors that shape results of these elections. In presidential elections the incumbent government's provision of redistributive benefits seems to matter a great deal (Zucco 2013). In down-ticket elections candidates depend on their local name recognition, which is a function of their ability to “bring home the bacon” through pork-barrel politics, and on their own wealth and/or their personal connections to deep-pocketed sources of campaign finance, because candidates for all offices except president cannot rely on their party for funding (Ames 1995; Samuels 2002).

We also know that contextual factors matter. For example, aggregate economic performance is strongly related to outcomes in presidential elections. A strong economy boosted the PSDB's Fernando Henrique Cardoso to office in 1994 and helped Lula's reelection effort in 2006, while a weak economy undermined the PSDB candidate in 2002 (Mendes … Venturi 1995; Carreirão 1999, 2007). Campello and Zucco (2016) have even shown that Brazilian presidents’ popularity responds to international economic conditions, which they certainly cannot influence.

None of the foregoing factors are directly related to voters’ perceptions of candidates’ parties. To what extent does positive and/or negative partisanship also shape voters’ decisions, encouraging them to cast ballots for candidates only of “their” party – or to avoid casting ballots for the “other” party, across all levels of elections?

Consider Figure 5.1, which explores evaluation of government performance since 1989. In line with the notion that contextual factors matter, the data suggest most broadly that the state of the economy shapes presidential popularity. For example, President Cardoso's (1995–2002) approval went up as Brazil's economy stabilized early in his first term, but then plummeted as Brazil entered a recession circa 1998. The data also show declines in presidential evaluation as the economy entered a recession in 2014 under President Dilma, and during the crisis that led to the impeachment of President Collor in 1992.

Conventional scholarly wisdom has long assumed that mass partisanship is weak and immaterial in Brazil – that voters’ stated affinities or antipathies for this or that party have little effect on their political attitudes and behavior, particularly when compared against the weight of candidates’ personal qualities and performance in office, whether in terms of managing the economy or providing constituent service. This view is incorrect. Despite Brazil's confusing party system and candidate-centric electoral rules, both positive and negative partisanship have mattered a great deal, for a substantial proportion of Brazilian voters.

Most research in political science has focused on positive partisanship, but Social Identity Theory (SIT) teaches us that out-group antipathy may actually drive attitudes and behavior more than any in-group affinity. SIT also suggests that positive and negative partisan attitudes do not necessarily mirror each other. These two notions pushed us to focus on both sides of the partisanship coin, in an effort to better understand Brazil's party system in the electorate.

Chapter 2 showed that observers have underestimated the extent of partisanship in Brazil's electorate. For about half of Brazil's voters – the nonpartisans – the conventional wisdom is true: partisanship plays no role in determining political attitudes or voting behavior. Yet this means that the other half have held either positive and/or negative partisan attitudes, or both. Chapter 2 also showed that since the 1980s, most partisan attitudes in Brazil have revolved around the PT, whether for or against, and demonstrated that the divide between petistas and antipetistas cannot be boiled down to sociological or demographic factors or to differences of opinion on important policies. Instead, the divide is a product of distinct normative views about how politics should work – particularly, about the value and purpose of democracy. Petistas have supported democracy because they want to use it to bring about social and economic change; antipetistas are less committed to democracy to begin with, and tend to fear change.

Chapter 3 turned to the relative strength of positive and negative partisan attitudes in Brazil. It showed that positive partisanship is as tightly “bounded” as it is elsewhere, particularly for the PT, and revealed that both positive and negative partisans engage in motivated reasoning.

introduction

Chapter 2 revealed that for a large proportion of Brazilian voters, one or both sides of the partisanship coin has mattered. At times more than half of Brazilians have expressed positive and/or negative partisan attitudes – and most of this sentiment has appeared as support for or opposition to the PT.

It is one thing to note that many Brazilians call themselves partisans or antipartisans.Yet how would we know whether such attitudes are “real?” It is possible that self-expressed partisan attitudes could be superficial and thus politically inconsequential. Social Identity Theory suggests that partisan attitudes – whether positive or negative – matter if they are relatively consistent over time and if they shape voters’ opinions and voting behavior.

To assess the temporal consistency of partisanship in Brazil, we consider its “boundedness” – the extent to which voters affirm an allegiance to a particular party over time – as assessed through responses in panel surveys. We find that petismo is bounded similarly to partisanship for major parties in other countries, while partisanship for the PSDB and PMDB is more weakly bounded.

We then explore two ways in which positive and negative partisanship shape voters’ attitudes. For example, if petismo and antipetismo are in fact powerful psychological attachments, then Brazilians who affirm one or the other attitude should be well primed to either agree or disagree with everything associated with the PT. We leverage two forms of empirical evidence – observational and experimental – to put this notion to the test. We show that motivated reasoning affects petistas, antipetistas, and tucanos similarly: it drives petistas to blindly support their party during good times and bad, just as it corners members of the latter two groups into holding contradictory views about the PT. However, we find that PMDB partisanship only weakly predicts political opinions.

We then explore results of a survey experiment designed to test whether partisans’ attitudes about public policies change when they are shown that “their” or “the other” party supports or opposes those policies. The results confirm that petismo and antipetismo have consistent and powerful effects on voter attitudes, while identification with the PSDB has weaker effects.1 In any case, taken as a whole, the evidence strongly supports the notion that partisan attitudes in Brazil are “real.”

If democracy is unthinkablewithout strong political parties (Schattschneider 1975), we all have good reason to worry. For decades, parties around the world have appeared to be in decline, with links of representation and accountability between voters and elected officials growing increasingly tenuous.After Brazil redemocratized in the 1980s, scholars quickly classified it as an important case study of partisan and party-system weakness: its political institutions, it was said, promoted individualism and undermined parties as agents of collective representation (Ames 2001), resulting in an “inchoate” party system (Mainwaring 1999). Most observers concluded that the weakness of Brazil's parties boded ill for the health of its nascent democracy (e.g., Lamounier 1989; Mainwaring 1992; Mainwaring & Scully 1995b; Weyland 1996; Kinzo 2004; D'Araújo 2009). Some scholars did see a glass half-full rather than half-empty, noting that Brazil's legislative parties were actually fairly cohesive, and that despite the party system's extreme fragmentation and relative ideological incoherence, democracy appeared to function about as well as in any other country in the region (Figueiredo … Limongi 1999; Melo … Pereira 2013; Montero 2014).

However, Brazil's recent political and economic crises – culminating in the 2016 impeachment of President Dilma Rousseff – has brought renewed attention to party and party-system dysfunction. After the 2014 elections twenty-seven parties held at least one seat each in the lower house of Brazil's legislature (the Chamber of Deputies), and the largest party held only 11% of the seats (Câmara dos Deputados 2016). This is an extraordinary level of fragmentation, especially given Brazil's lack of ethnic, linguistic, or religious cleavages, as for example in India.

The political crisis that erupted following that year's election was so profound that by the mid-2010s Brazilian voters expressed the lowest level of confidence in political parties of any country in the region (Latinobarómetro 2016b). Furthermore, by 2016 72% of Brazilians stated they felt close to none of Brazil's parties, the lowest level since survey firms started asking a partisanship question in 1989 (Datafolha 2016). Disillusionment with parties also damaged popular faith in democracy: by 2016 only 32% of Brazilians agreed that “Democracy is preferable to all other forms of government,” a decline of twenty-two points from the previous year and ahead of only Guatemala across Latin America (Latinobarómetro 2016a).