‘Them’ Without ‘Us’
Negative Identities and Affective Polarisation in Brazil
Jo˜
ao Areal
jbarealneto@gmail.com
UvA ID 11754583
A thesis presented for the programme
Research Master Social Sciences
12,980 words
Supervisor and First Reader: Dr. Eelco Harteveld
Second Reader: Dr. Mike Medeiros
University of Amsterdam
Amsterdam, 23 July 2020
‘Them’ Without ‘Us’
Negative Identities and Affective Polarisation in Brazil
Jo˜
ao Areal
Abstract
High levels of hostility between those on opposing sides of politics have led to a burgeoning literature on the concept of affective polarisation. Though a globally widespread phenomenon, extant literature has generated theoretical expectations and empirical findings mostly inspired by the United States and Western Europe. By studying the case of Brazil, I argue and show that traditional explanations do not provide satisfactory accounts of affective polarisation in contexts where politics is only weakly structured by ideology or partisan attachments. I ar-gue and show that in such contexts the concept of negative political identities can provide a much better explanation for why politics is so divisive. Through a sequential design that leverages the 2014 and 2018 waves of the Brazilian Electoral Studies (BES) and independently collected survey data (N = 1,732), I provide robust empirical findings supporting the primacy of negative political identities over traditional hypotheses. Negative identification with the out-party/leader has a strong effect on dislike towards out-voters even when controlling for instrumental evaluations of political elites. This thesis makes a significant contribution to the field of affective polarisation and the study of political identities.
1
Introduction
After years of mass protests, economic crises and corruption scandals, scholars and journalists often remark that Brazil is engulfed in deep political polarisation (Kingstone & Power 2017; Mignozzetti & Spektor 2019). The 2018 electoral success of the far-right president Jair Bolson-aro, amidst a campaign filled with rash rhetoric and hostility between political opponents, has epitomised the divisive nature of Brazilian politics. Bolsonaro’s continuously antagonising rhet-oric towards the opposition is often met with a strong backlash, and reports of protests turning violent have abounded in recent years (e.g. O Globo 2019; Folha de S˜ao Paulo 2020b). Politics has come to divide friends and families like perhaps never before in Brazil, and polarisation shows no signs of abating: Brazilians of different sides of politics dislike each other to increasing and alarming levels.
This animosity between political opponents has been termed affective polarisation, understood here as inter-citizen political hostility. This hostility between individuals from different sides of politics has been shown to have pernicious effect for social integration and democracy at large (e.g. Kalmoe & Mason 2018; Martherus et al. 2019), with a vast body of work dedicated to understanding the factors that turn politics into a nasty game between enemies. The bulk of extant literature on affective polarisation, however, is concentrated either on the specific cases of the United States (Iyenger at al. 2012; Webster & Abramowitz 2017; Mason 2018b; Robison & Moskowitz 2019) or Western Europe (Wagner 2017; Westwood et al. 2018; Hobolt et al. 2020). Such contexts are characterised by highly institutionalised party systems (Dalton & Weldon 2007), programmatic patterns of political competition (Kitschelt 2000), strong linkages between social groups and political parties (Lipset & Rokkan 1967), and more entrenched and stable partisan identities (Neundorf et al. 2011; Huddy et al. 2015; Bankert et al. 2017). Like most of Latin American (Kitschelt et al. 2010; Lupu 2016) and Eastern European (Rose & Mishler 1998; B´ertoa 2012) countries, Brazil displays none of these features. Brazilian politics is only weakly structured by ideology (Oliveira & Turgeon 2015), stable partisan identities have largely failed to take root (Samuels & Zucco 2018), and parties lack clear linkages with distinct social groups (Mainwaring 1999). These considerations present a significant challenge to traditional mechanisms, which posit that people dislike each other either because they hold vastly contrasting views, or because they are deeply attached to their respective ‘side’ of politics. By systematically operationalising and testing these classic explanations for the first time in Brazil through analyses of the 2014 and 2018 waves of the Brazilian Electoral Study (BES), I show that these lines of explanation perform particularly poorly in the Brazilian context. If not through ideology or strong partisan identities, how can politics be so divisive? I build on
the concepts of negative partisanship (Medeiros & No¨el 2014; Mayer 2017; Samuels & Zucco 2018) and negational identities (Zhong et al. 2008), and propose a theoretical framework un-derpinning the notion of negative political identities in order to explain affective polarisation in Brazil. Negative political identities emerge independently out of a strong rejection of the opposing side’s party or political figurehead, and are particularly likely to arise in contexts of low partisan attachments. By leveraging independently collected survey data (N = 1,732) amongst Brazilian respondents, I put forward an innovative empirical measure of negative polit-ical identities. I show that a deep rejection of the the opposing side’s party/leader constitutes a prevalent form of social identity, and that these negative identities are the primary mechanism through which Brazilians dislike their fellow citizens. These findings are robust to models that control for negative instrumental evaluations of the out-party/leader, isolating the independ-ent effects of negative political idindepend-entities in generating out-voter dislike. Further, the effects of negative identities far surpass those of ideological disagreements and in-group attachments, indicating that Brazilians dislike ‘them’ even without a clearly defined sense of ‘us’.
The strength of negative political identities amongst the Brazilian electorate bespeaks an in-creasingly hostile and uncompromising polity, which should worry political practitioners and scientists. As citizens turn away from parties (Dalton & Wattenberg 2002), negative political identities are an increasingly important factor structuring political competition both in new (Fomina 2019) and advanced democracies alike (Boonen 2019; Bankert 2020). The theoretical and empirical contributions on the concept presented by this paper should thus be of significant relevance to the study of comparative political behaviour.
This paper is structured as follows. I begin by conceptualising affective polarisation and out-lining the recent divides that have marked Brazilian politics over the past decade. Thereafter, I review the existing literature and derive testable hypotheses that, as I shall argue, should provide an unsatisfactory account of affective polarisation in Brazil. I then put forward an al-ternative hypothesis couched in the idea of negative political identities, significantly developing the theoretical framework behind this oft-overlooked concept underpinning political competi-tion. Empirical results are first presented for the BES data, which are followed by results from my online survey. The survey allowed me to develop an innovative empirical measure of neg-ative political identities, as well as to provide a robust test of the mechanism linking negneg-ative identification with affective polarisation. In the conclusion I highlight several avenues for future research, which can draw on the theoretical and empirical contributions made by this paper.
2
Theoretical Framework
2.1
Affective Polarisation: When Adversaries Become Enemies
Political conflicts have always been at the heart of liberal democracies, and competing political agendas are thought to be a hallmark of a healthy competitive polity (Lipset 1959; Mair 2013; Lupu 2015; Mainwaring et al. 2018). Over the past few years, however, US scholars have noted that Americans from different sides of politics do not simply disagree - they increasingly fear, distrust, and even hate each other (e.g. Iyengar et al. 2012; Iyengar & Westwood 2015; Webster & Abramowitz 2017; Mason 2013, 2018b). These developments have led to a burgeoning literature on the concept of affective polarisation, describing the levels of hostility individuals feel towards the those from the ‘other side’ of politics. Though initially restricted to the United States, more recent works by comparative scholars have translated the concept of affective polarisation to multi-party systems in Europe, highlighting that the US is not alone in its levels of political divisions (e.g. Wagner 2017; Westwood et al. 2018; Harteveld 2019; Gidron et al. 2019; Reiljan 2019; Hobolt et al. 2020).
A crucial contribution of this body of literature is the differentiation between affective and ideo-logical polarisation: while an ideoideo-logically polarised polity need not descend into acrimonious conflicts, high levels of affective polarisation can be found even in the absence of significant ideological disagreements (Iyengar et al. 2019). If political adversaries see each other as evil enemies, deliberation and cooperation are less likely to occur (Hobolt et al. 2020), threaten-ing the quality and stability of liberal democracies. Perhaps more worrythreaten-ingly, past research has shown that affective polarisation can affect interpersonal relationships (Huber & Malhotra 2017), foster the dehumanisation (Martherus et al. 2019) and increase tolerance for violence against one’s opponents (Kalmoe & Mason 2018).
I formally define this dislike towards the political ‘other’ as hostility towards political out-groups. This definitions captures animus between partisan (Webster & Abramowitz 2017), ideological (Mason 2018), social (Harteveld 2019) or even issue-based (Hobot et al. 2020) groups. This differs from initial definitions that conceptualised affective polarisation as the tendency of partisans to see co-partisans positively and out-partisans negatively (Iyengar et al. 2012), since this imbues the notion of out-group hostility with in-party bias. Individuals might well prefer their ‘own’ side without necessarily disliking their opponents, with longitudinal (Abramowitz & Webster 2016) and comparative (Reiljan 2019) evidence showing that affective polarisation is primarily driven by negative ratings of the out-group. Including in-party bias in the definition of affective polarisation would also obfuscate the prevalence of out-group hostility
in contexts, such as Brazil, of low partisan attachments. A second necessary clarification is that affective polarisation is understood here as an inter-citizen phenomenon. Past research has shown that individuals’ feelings towards political parties are significantly different to feelings towards their fellow citizens (Druckman & Levendusky 2019; Kingzette 2020), and might be capturing generally ‘anti-politics’ feelings amongst politically disillusioned voters (Klar et al. 2018). Since Brazilian parties lack clear social bases of support (Mainwaring 1999; Samuels & Zucco 2014) and tend to be highly distrusted by the electorate (Bargsted et al. 2017), it is all the more likely that Brazilians’ attitudes towards parties or political elites might not capture inter-citizen hostility. The principal goal of this paper is therefore to understand why Brazilians dislike their fellow citizens on the opposite side of the political divide.
Affective Polarisation in Brazil
While, to the best of my knowledge, there are no published works that attempt to study these divides through the lenses of affective polarisation, it is widely acknowledged that Brazil is engulfed in deep polarisation since at least 2014. The Car Wash corruption scandal and the deep economic crises that started to take hold in 2014 culminated in widespread mass protests, expressing high levels of anger and disillusionment amongst Brazilians. The left-wing Workers’ Party (Partido dos Trabalhadores, henceforth PT), in office from 2003 to 2016, bore much of the brunt of this popular anger. President Dilma Rousseff (PT) would end up being removed from office in a bitterly divisive process that triggered accusations of a coup d’´etat on the part of the PT and its supporters (Kingstone & Power 2017). As a symbol of this deeply polarised context, demonstrations against and in favour of the presidential impeachment had to be kept apart by a dividing barrier lest they descended into violent conflicts (Folha de S˜ao Paulo 2016). With much of the political class discredited by their association with corruption scandals, far-right populist Jair Bolsonaro took over one of Brazil’s many ’parties for rent’ (Desposato 2006), the formerly tiny Social Liberal Party (Partido Social Liberal, henceforth PSL), emerging as a credible alternative to Brazil’s mainstream parties at the 2018 presidential contest. Bolsonaro capitalised on a strong rejection of the PT (a phenomenon know as antipetismo), popular disillusionment with politics, a culturally conservative agenda, and an extremely hostile rhetoric against his opponents (Hunter & Power 2019). Though ultimately victorious, Bolsonaro’s brash style and extreme policies were decried by many Brazilians, and both he and the PT’s presidential candidate had high levels of rejection amongst the opposing groups of voters (BBC Brasil 2018).1
In the years since the acrimonious contest of 2018, protests both against and in support of
1For a more comprehensive overview of the last few years of crises and polarisation in Brazil, see Mignozzetti
Bolsonaro’s government have become commonplace (e.g. DW Brasil 2019; Folha de S˜ao Paulo 2020a), political disagreements have entered people’s personal networks (Machado et al. 2019), and stories of prejudice and even violence proliferated (e.g. G1 2018; Folha de S˜ao Paulo 2020b). Studying the case of Brazil, however, is of significant importance to the broader field of affective polarisation not simply because it is yet another country suffering from acrimonious political divides. As I shall detail in the next section, traditional mechanisms that attempt to account for affective polarisation in contexts such as the United States or Western Europe are unlikely to hold much explanatory power in Brazil. Understanding how affective polarisation can take place through non-traditional ‘sources’ can thus significantly advance our understanding of this phenomenon.
2.2
Traditional Sources of Affective Polarisation
In reviewing the extant literature on the individual-level mechanisms behind affective polarisa-tion, I start by deriving testable hypotheses that have been employed by scholars in the United States and Western Europe. These mechanisms are broadly based on ideological disagreements and conflicts arising out of individual’s attachment to socio-political groups. After presenting each hypothesis I reflect on their applicability to Brazil, arguing that, owing to the weak ideo-logical structuration of Brazilian politics and low levels of partisan identification amongst the electorate, mainstream explanations are unlikely to provide a satisfactory account of affective polarisation in Brazil. I thus propose an alternative hypothesis couched on the idea of negative political identities. I hypothesise that Brazilians dislike ‘the other side’ not through ideolo-gical disagreements or identification with their ‘own side’, but rather because the opposing party/leader provides them with a negative identity. These negative identities, I shall argue, are the primary driver of out-group hostility in Brazil.
2.2.1 Ideological Distance
Even though ideological polarisation and affective polarisation are theoretically distinct con-cepts, it is expected that a certain level of hostility should be present if political opponents are ideologically distant from each other to a significant extent (Abramowitz & Webster 2017). In the classic rational choice paradigm, the utility ‘costs’ of seeing the other side win will increase as a function of how ideologically distant the opponent is (Downs 1957). In this approach, ideo-logical and affective polarisation would closely mirror each other: the greater the ideoideo-logical distance, the greater, too, the affective distance.
The empirical evidence on this line of explanation, however, is mixed. Studies by Rogowski and Sutherland (2016), Webster and Abramowitz (2017), Lelkes (2019), and Orr and Huber (2020) show that citizens do tend to dislike those who they perceive as being too ideologically distant from themselves, whilst comparative evidence provided by Reiljan (2019), Harteveld (2019) and Gidron et al. (2019) find only weak associations between ideological and affective polarisation, particularly in less institutionalised party systems such as in Latin America and Eastern Europe. We should thus cautiously expect that affective polarisation is at least partly explained by the ideological distance between citizens of opposing sides. In concrete terms, this means that individuals whose policy preferences are more distant to those held by citizens on the ‘other side’ should display higher levels of out-group hostility.
Hypothesis 1: Higher levels of ideological distance from the opposing side are associated with higher levels of affective polarisation.
It is unlikely that this hypothesis will hold much explanatory power in Brazil, however. Findings by Borges and Vidigal (2018), Samuels and Zucco (2018), Kearney and Machado (2018), and Mignozzetti and Spektor (2019) tend to show little and decreasing ideological differentiation between voters and sympathisers of the PT and its main rival on the centre-right until 2014, the PSDB (Brazilian Social Democratic Party, Partido da Social Democracia Brasileira). Brazilian parties themselves converged around the centre for much of the democratic period (Lucas & Samuels 2010; Power & Zucco 2012), which might have contributed to this lack of ideological distance in the electorate (Adams et al. 2012). More recently, Zucco and Power (forthcoming) do notice an increase in ideological polarisation at the elite-level, and Amaral (2020) reports a growing number of Brazilians identifying on the far-right end of the ideological scale. Though more polarised elites can indeed generate more mass-level ideological polarisation (Zechmeister 2015), past research has consistently shown that ideological self-placement is highly endogenous to presidential candidate affect in Brazil (Ames & Smith 2010; Renn´o 2010 & Cabello 2010; Baker et al. 2016; Borges & Vidigal 2018; Avelino et al. 2020), such that the rise in the number of ‘far-right’ Brazilians might be an artefact of a growing identification with Jair Bolsonaro rather than a meaningful change in ideological beliefs.
2.2.2 Party-Issue Constraint
Ideology can also lead to affective polarisation based on the extent to which disagreements are captured by the partisan divide rather than overall distance. This alternative approach builds on the concept of ‘issue constraint’ (Converse 1964), i.e. the extent to which citizens have consistent positions across political issues. A highly constrained individual will have issue
preferences that fall on the same ‘side’ of the ideological divide - for instance, by holding right-wing views on taxation, privatisation and same-sex marriage.2 We can thus think of ‘party-issue constraint’ as the extent to which division over issues and policy are structured along partisan lines (Baldassari & Gelman 2008). For example, if someone holds views that tend to fall on the same side of the partisan divide, that person shows a high level of alignment between their ideology and party preferences. Crucially, this party-issue alignment is independent of the overall relative ideological distance between voters: partisan camps might not be so ideologically distant from each other, but, due to strong party-issue alignment, they might disagree on nearly everything (see also Sniderman & Stiglitz 2012).
If the partisan divide encompasses every other potential line of ideological conflict in society, voters of opposing partisan groups will have little in the way of ideological common ground. Drawing on belief congruence theory (Tajfel 1982), Bougher (2017) argues and shows that strong party-issue alignment leads to prejudice and dislike towards out-groups couched in the perceived dissimilarity of beliefs. Similarly, Sanders and Abramowitz (2017) show that higher levels of ideological consistency amongst voters predict more negative evaluations of the out-party. We should thus expect an individual who holds issue preferences that are consistent with those of their party of choice to dislike the opposing side more than someone who is less consistent in their party-issue alignment.
Hypothesis 2: Higher levels of party-issue constraint are associated with higher levels of af-fective polarisation.
This hypothesis requires that voters be consistent in their ideological beliefs, and that those beliefs be captured by mutually exclusive partisan camps. Neither condition is satisfied in Brazil. Brazilians have historically shown low levels of consistency and stability in their ideolo-gical beliefs, regardless of individual differences in political sophistication (e.g. Ames & Smith 2010; Oliveira & Turgeon 2015). Brazilian voters tend to lack a basic understanding of left and right, and issue domains are largely orthogonal to each other (Kearney & Machado 2018). Per-haps more importantly, ideological preferences are only weakly related to sympathy or electoral support towards mainstream parties (Carreir˜ao 2007; Nicolau 2014; Ribeiro et al. 2016; Izumi 2019), in contrast to what is observed in the United States (e.g. Baldassari & Gelman 2008; Webster & Abramowitz 2017) and Western Europe (e.g. Rekker & Rosema 2019; Oscarsson & Rosema 2019). This is unsurprising, given that Brazilian parties are notoriously devoid of a consistent or stable ideological agenda (Mainwaring et al. 2018), with constant party switching amongst legislators (Deposato 2006) and candidate-centred campaigns (Ames 2001) hindering
2In contrast, lower issue constraint would mean holding right-wing views on certain issues, but having a
the importance of ideology in structuring party support. Instead, party support and vote choice are typically understood through the lenses of pork-barrel politics (Ibid.), leader’s evaluations (Baker et al. 2016), clientelism (Mainwaring 1999), or the state of the economy (see Carreir˜ao & Renn´o 2018 for a review).
If voters do not have consistent ideological beliefs, and if politics itself is only loosely structured by ideology, the alignment of issue and party preferences is unlikely to be a significant driver of affective polarisation in Brazil. Nonetheless, increased polarisation and consistency amongst political elites in recent years (Zucco & Power forthcoming) might have changed this scenario. Elite polarisation can increase ideological consistency amongst the public (Levendusky 2010), and the increased usage of ideological labels by (especially right-wing) politicians (Quadros & Madeira 2018) might have ‘sorted’ voters into partisan camps that are more closely aligned with their beliefs.
2.2.3 The Interplay of Social Identities
Though it is expected that diverging ideological preferences might lead to negative feelings towards political opponents, the bulk of the literature on affective polarisation has emphasised the role played by social identities (Iyengar et al. 2019). This approach is couched in Social Identity Theory (SIT), which posits that individuals derive a sense of identification from social groups to which they see themselves as belonging (Tajfel & Turner 1979), categorising the world as ‘us’ (group members, that constitute an in-group) and ‘them’ (nonmembers, or members of out-groups). Building on the SIT theoretical framework, political scientists have developed the notion of partisanship as a social identity, or expressive partisanship (Huddy et al. 2015), whereby partisan attachments can become part of an individual’s identity akin to traditional group identities such as race, gender or religion (see Greene 1999, 2002)3. This conceptualisation of partisanship has been crucial to the literature on affective polarisation due to negative biases against out-groups that are said to follow from identifying with a social group (Huddy et al. 2018). Social Identity Theory posits that individuals seek to maintain the group’s positive image and higher status in the face of out-groups (e.g. Abrams & Hogg 1988), such that belonging to an in-group can induce hostility and prejudice against out-groups even if group divisions are based on trivial or minimal criteria (e.g. Billig & Tajfel 1973). The behavioural effects of social identities, however, vary according to the the strength of individual’s identification
3In contrast, an alternative, rational choice approach to partisanship focuses on the expected utility an
indi-vidual might gain from supporting a particular party, which should be highly responsive to a party’s agenda and performance (Downs 1957; Fiorina 1981). This instrumental (Huddy et al. 2015) model of partisanship implies less stable partisan attachments, and out-group hostility, if present, would arise out of rational evaluations rather than identity-based mechanisms
(Huddy 2001; Huddy et al. 2018). As empirically corroborated by Mason (2015) and Miller and Conover (2015), we should expect out-group hostility to be higher amongst strong party identifiers.
Hypothesis 3: Stronger partisan identification is associated with higher levels of affective polarisation.
The ‘affectively polarising’ effects of social identities are further enhanced when other group-identities - such as race, class, religion etc - ‘overlap’ (Roccas & Brewer 2002). When individuals are distant on a range of group-based divides they become more socially isolated and less able to cope with threat, leading to more negative feelings about out-group members than those with cross-cutting identities (Mason & Wronski 2018). For example, if the majority of white and Protestant voters tend to identify with the Republican party, a white, Protestant Republican partisan will have their political, social and religious identities ‘aligned’ on the same side of the political divide. This might make one less likely to have social contact with out-partisans, and to overestimate inter-group differences (Ahler & Sood 2018). The alignment of partisanship and demographics - called social sorting (Mason 2016) - leads to a scenario where partisan divides capture other societal fault lines, potentially enhancing hostility between political opponents. We should thus expect individuals who are highly socially ‘sorted’ to hold more negative views of the opposing side.
Hypothesis 4: Higher levels of social sorting are associated with higher levels of affective polarisation.
Identity-based approaches have been central to our understanding of affective polarisation. In their seminal article, Iyengar et al. (2012) brought a wealth of evidence to show that out-group evaluations in the US have become increasingly negative over the past few decades, though this is only weakly associated with individuals’ policy positions. Similarly, Mason (2013, 2016, 2018a, 2018b) shows that aligning social and political identities have a stronger independent effect on dislike for the political out-group than ideological differences. Lending further credence to group-based explanations, Robison and Moskowitz (2019) show that negative evaluations of societal groups linked to an individual’s out-party tend to predict negative evaluations of the out-party itself, whilst Harteveld (2019) presents evidence of the effects of social sorting on affective polarisation across American, European, and Asian countries. If the interplay of identities is the main driver of affective polarisation, however, it follows that social identities must exist amongst a significant part of the electorate, playing an important role in structuring how citizens engage with politics.
comparatively low, reaching 50% of the electorate in its peak, but generally hovering at around 30% (Samuels & Zucco 2018). Only the PT managed to claim a significant portion of these partisans, though, for the most part, PT identifiers rarely amounted to more than a quarter (25%) of the electorate (Ibid.), and the recent political and economic crises severely eroded support for the party (Amaral 2020). The instability in party identification amongst Brazilians points at another factor countering the importance of party identities as drivers of out-group hostility in the country: party attachments are primarily connected to leaders’ personal appeals and party performance rather than a strong sense of identity (Baker et al. 2006; Baker et al. 2016).4 Some Brazilian voters might have positive feelings towards a particular party, but, since these attachments are by and large the result of instrumental evaluations, they do not fit within explanations couched in Social Identity Theory. In sum, Brazilian partisans are few in quantity and low in terms of social identification, making partisanship an unlikely explanation for out-group hostility in the country.
Secondly, Brazil, like much of Latin America, stands in stark contrast to the United States (Campbell et al. 1960; Mason 2018b) and Western Europe (Lipset & Rokkan 1967; Helbling & Jungkunz 2020) in the extent to which social cleavages structure partisan competition. Though racial, income and regional inequalities run deep in Brazilian society, these cleavages have not been articulated into politics, neither have they been properly mobilised by political actors (Samuels & Zucco 2014). Political parties have seldom tried to convey group-based appeals and have been largely unsuccessful when they did try, resulting in rather heterogeneous social bases of support without clear linkages with specific social groups (Mainwaring 1999). Two relevant exceptions could be made with regards to the PT’s association with the Northeast (Hunter & Power 2007), and the growing alignment of conservative Evangelicals with the right-wing, and, especially, Jair Bolsonaro (Quadros & Madeira 2018; Smith 2019). In spite of these two examples of regional and religious political alignment, the fact that social identities are generally ‘only weakly politicised at best’ (Zucco & Power forthcoming, p. 2) means that, even if some social groups are aligned with a particular partisan camp, the social sorting mechanism is unlikely to be a substantive driver of hostility in Brazil.
In sum, traditional explanations for affective polarisation seem to offer little in the way of explanatory power when applied to the Brazilian context. Ideological preferences play only a small role in dividing voters into opposing partisan camps, and ideological distance is not particularly high. The lack of widespread and stable partisan identities, coupled with the
4A fitting example here is that of the PSL. As one of many ‘parties for rent’ (Desposato 2006), the PSL acted
mostly as a vehicle for Jair Bolsonaro to propel his candidacy and that of his allies, being formerly a largely unknown party. As Amaral (2020) shows, however, the PSL already captured 6% of the Brazilian electorate in 2018, second only to the PT at 10%.
weak politicisation of social cleavages by political actors, pose serious challenges to mechanisms originating in in-group attachments. To be clear, the point is not that those mechanisms are wrong at the individual-level, but rather that they would explain very little of aggregate-level affective polarisation in Brazil due to low scores on the independent variables. How, then, can politics be so divisive? In attempting to solve this puzzle I put forward a hypothesis couched in the idea of negative political identities. I argue that a strong rejection of the ‘other side’s’ party or presidential candidate informs Brazilian’s own self-concept, constituting a meaningful form of social identity. Not only are these negative identities more prevalent than positive ones amongst the Brazilian electorate, they are also theorised to have a much stronger effect on out-group hostility. Crucially, this explanation eschews the need for strong in-out-group attachments, and go beyond rational ideological evaluations of one’s out-party. As such, I expect negative identities to provide a much better overall explanation for affective polarisation in Brazil.
2.3
‘Them’ Without ‘Us’: How Negative Identities Fuel Affective
Polarisation
The concept of social identities has long been couched in the idea of an individual’s belonging or attachment to a particular social group (Tajfel & Turner 1979). From this perspective, in-group identification is psychologically prior to the definition - and derogation - of out-in-groups. Negative feelings and biases towards ‘the other’ must then be fundamentally couched in a prior sense of identification to an in-group. Zhong et al (2008), however, argue that individuals can also define themselves negationally, in terms of who they are not, even if no prior in-group identification is present. This means that, for instance, individuals might define themselves as ‘not-Republican’ without having any attachment or identification with the Democrats (see also Weisberg 1983). In theorising about the psychological roots of negative identities, Zhong et al. (2008) draw on Optimal Distinctivness Theory (Brewer 1991), which posits that social identities satisfy two basic human needs: a need for belonging (to an in-group) and a need for distinctiveness (i.e. differentiating oneself from out-groups). Positive identities can satisfy both needs, and should thus be preferred by individuals. In the absence of in-group attachment, however, negative identities can emerge out of an independent and primary dislike towards the out-group. Indeed, positive and negative group evaluations are not simply mirror images of each other, and individuals do not need a strong sense of identification or in-group attachment to display out-group derogation (Medeiros & No¨el 2014).
Since negative identities are primarily informed by one’s need to differentiate oneself from a particular group, they are likely to have significant effects on out-group hostility. In fact, Zhong
et al. (2008) find that negative identities have a higher effect than positive ones in fostering negative feelings towards individuals associated with the out-group label. Whilst in-group attachment does not necessarily lead to out-group derogation (Brewer 1999), this should flow much more directly from negative identities. Two crucial points to be made here are that, firstly, this hostility should arise out of a true sense of negative identification with the label associated with the out-group (e.g. a party). People might like or dislike political parties for instrumental reasons connected to particular policies or performance (Huddy et al. 2015; Rosema & Mayer 2020), and, though this might lead to more negative feelings towards citizens associated with that party, this is not predicted by the negative identity mechanism. Testing the link between negative identities and out-group hostility should thus attempt to separate out the effects of instrumental evaluations from those that arise out of a negative identification. Secondly, this hostility should be measured towards other individuals associated with that label. For example, a negative identification with the Republican Party should predict negative feelings towards other citizens associated with the party (e.g. Republican voters), even when controlling for negative instrumental evaluations of the Republican Party. This effect should also be higher than the out-voter dislike predicted by identifying purely as a Democrat.
Hypothesis 5: Negative identities have a stronger effect on affective polarisation than positive identities.
An important advantage provided by the concept of negative identities is that they do not require any sense of in-group belonging at all. Understanding affective polarisation through the lenses of positive identities is likely to return unsatisfactory results in countries - like Brazil - where partisan identities are rare. But why might negative identities emerge in the first place, and, more importantly, why do they apply in Brazil? Though political scientists have been paying increasing attention to the notion of negative partisanship, i.e. voters’ rejection of a particular party, the bulk of this literature has focused on identifying its effects on vote choice and political participation. (e.g. Garry 2007; Abramowitz & Webster 2016; Mayer 2017; Borges & Vidigal 2018; Iyengar & Krupenkin 2018) rather than potential causes, especially with reference to a clear theoretical framework. These studies are also generally agnostic as to whether negative partisanship acts a social identity or if it simply captures negative instrumental evaluations unrelated to one’s self-concept (Boonen 2019). In this vein, Bankert (2020) provides the first and only published study to measure negative partisanship as a social identity, though still primarily focused on its consequences for certain behavioural outcomes. Further, since this was a study conducted in highly partisan context of the United States, it could be argued that these negative political identities emerged out of previous partisan attachments, casting doubts over its usefulness in Brazil and other low-partisan contexts. We must thus go further
in theorising about the roots of negative political identities, with a view to establishing their applicability as a significant driver of affective polarisation in Brazil. While the concept might lack a clearly established theoretical framework, we can attempt to translate the mechanism proposed by Zhong et al. (2008) to the realm of politics, drawing on the case of Brazil for further theoretical and empirical leverage.
2.3.1 Towards a Theory of Negative Political Identities: (Anti) Petismo & (Anti) Bolsonarismo in Brazil
In the low-partisan context of Brazilian politics, scholars have long emphasised the relevance of antipetismo, a strong rejection of the PT, in structuring voters’ behaviour. As Samuels & Zucco (2018) argue, both positive and negative party attachments in Brazil have long revolved around the PT, as the party’s electoral success and deliberate efforts to cultivate a clear brand made the it a clear and salient party to either like or dislike. In contrast, the largely non-programmatic alternatives to the PT did not particularly attempt to develop a widespread partisan base (Samuels & Zucco 2014). This meant that non-supporters of the PT lacked any clear party to belong to and thus could not satisfy their need for in-group belonging. The high salience of their out-party, in turn, fostered a need for distinctiveness, the bases of negative identity formation. Indeed, antipetismo is not simply a rational negative evaluation of the PT: it has been shown to act as a source for perceptual bias and motivated reasoning (Samuels & Zucco 2018), and, though the PT is the standard-bearer of the Brazilian left (Hunter 2010), not to be primarily grounded in ideological beliefs (Ribeiro et al. 2016; Samuels & Zucco 2018). This argument resonates with past works on negative partisanship in Eastern Europe, whereby electorally dominant parties dichotomise competition between supporters and non-supporters (Rose & Mishler 1998; Fomina 2019). Similarly, the low levels of trust that Brazilians and Eastern Europeans place on political parties that are seen as corrupt and unrepresentative can further voters’ need for distinctiveness rather than belonging (e.g. Reiljan 2019). This can also explain why antipetistas now even out-number PT partisans (Amaral 2020), following the party’s involvement in recent corruption scandals and the general political disillusionment and anger amongst Brazilians in recent times.5
If antipetismo is a prevalent negative identity, and if negative identities lead to affective po-larisation, this hints at why Hypothesis 5 might apply so well in Brazil. However, this implies an asymmetrical affective polarisation and it does not explain why PT sympathisers (or, more broadly, left-leaning voters) might dislike ‘the other side’. While Samuels & Zucco (2018, p.
5A relevant criticism could be levelled here at the seemingly volatility of negative political identities. However,
41) state PT and anti-PT partisans dislike each other, the lack of empirical evidence on this front might be due to the fact that, at least until very recently, there was no clearly defined group to capture hostility emanating from left-leaning Brazilians or PT sympathisers. After all, antipetistas constituted a markedly different and more numerous group than partisans of the PSDB. The emergence of Jair Bolsonaro, encapsulating a blatantly anti-leftist and far-right political movement, however, fills this gap. Though we can only speculate how much being ‘not-being-right-wing’ would have constituted a negative identity for leftists or PT supporters, the notion of a clear bolsonarista identity group is far more concrete (Renn´o 2020), laying the grounds for a negative antibolsonarista social identity.
Though bolsonarismo is significantly informed by antipetismo, antibolsonarismo does not simply emerge out of Bolsonaro’s attacks against the PT or the left. Were that the case, this negative identity should only emerge amongst PT partisans or strong sympathisers. Rather, Bolsonaro’s extreme policies and rash populist rhetoric can foster a need for distinctiveness even amongst voters who did not particularly like the PT or the political left. Bischof & Wagner (2019), for instance, argue that far-right parties can motivate individuals to signal their disapproval for their extreme policies and discourse, fostering a need for differentiation. This is not to be understood as a simple instrumental evaluation couched in ideological distance: as Gidron et al. (2019) show, radical-right parties throughout Western Europe tend to be much more disliked than their ideological extremism would predict. We can couple this policy extremism with Bolsonaro’s blatantly hostile and populist rhetoric, which is also theorised to evoke strong neg-ative feelings amongst voters (Mudde & Rovira Kaltwasser 2018; Bankert 2020; Skytte 2020). The point, however, is not that ideology plays no role in triggering a need for distinctiveness and generating a negative identity. Rather, negative identities cannot be reduced to ideological disagreements alone, and lead to different behavioural outcomes than those predicted by ra-tional evaluations6. Much like their positive counterparts, negative political identities, too, can induce motivated reasoning and induce hostility towards out-groups, i.e. individuals associated with the source of those identities.
The theoretical underpinnings of Hypothesis 5 should be much clearer now. Through a need for distinctiveness, Brazilians ‘negatively identify’ with the party/leader that symbolises the ‘other side’ of politics, which is theorised to have strong effects on out-group hostility. My argument is thus twofold. As per Optimal Distinctivess Theory, positive identities should satisfy both basic human needs (i.e. for belonging and differentiation) and be generally preferable to purely negative ones. However, contextual political factors - such as lack of efforts on the part of elites to develop partisan identification, non-programmatic political competition, extreme policies and
6Additionally, ideology-based mechanisms touch upon ideological divergences between citizens rather than
divisive rhetoric - can make in-group belonging less likely for a significant portion of voters, whilst at the same time fostering a need for these individuals to distance themselves from political parties. Even when positive identification is present, however, I expect that negative identities will have an independent and stronger effect on out-group derogation than positive identities. This explanation should go well beyond Brazil: not only are negative identities prominent in Eastern Europe and other Latin American countries, but they can also compound positive partisan attachments in more established polities (e.g. Bankert 2020 on the US). The point concerning Brazil is not that in-group attachments do not work as expected in generating out-group hostility, but that positive partisan identities are rarer than negative ones. Put together, this should make negative political identities a much better explanation for affective polarisation than the traditional hypotheses posited by mainstream literature.
Testing the mechanism that links negative political identities and affective polarisation presents a significant empirical challenge. Firstly, we need a direct measure of negative identities, which goes beyond affective ratings of parties or politicians that might simply capture instrumental evaluations. Bankert (2020)’s study aside, there is no such measure in any existing dataset, and works on negative partisanship have either used like/dislike scales or question-items indicating which party a respondent would never vote for. Since it would be pure tautology to show that ‘never voting for a party’ is associated with negative feelings towards that party, we also need to operationalise affective polarisation as hostility towards other citizens. Affective ratings of voters or other political groups, however, are hard to come by, and literature on affective polarisation has typically relied on party affective ratings. A proper test of Hypothesis 5, thus, requires a more fine-grained operationalisation of both negative identities and out-group hostility than the ones that are commonly found in existing datasets.
3
Analyses
The lack of appropriate measures in existing datasets hinder a test of Hypothesis 5, which I expect to be the key explanation for affective polarisation in Brazil. As such, I employ a sequential research design whereby results from survey data provided by the Brazilian Electoral Study (BES) are followed by results from data I collected through an online survey in April and May 2020. This analytic strategy can leverage the high quality of the BES samples and the fine-grained measures provided by my survey data. By ‘triangulating’ the empirical results, findings can be supported by both external and internal validity. I split this section by data source, starting with the operationalisation and analyses of the BES data and moving on to the online survey results.
3.1
Brazilian Electoral Studies (BES)
3.1.1 Data and Operationalisation
The Brazilian Electoral Study (BES) collects measures of political and social attitudes amongst a nationally representative sample of voting-age Brazilians in the weeks following national elections. Though these data have been collected since 2002, the lack of comparability on certain question-items means that model results are restricted to the last two BES waves, encompassing the 2014 and the 2018 elections. The final analytic sample consisted of 2,870 respondents.
Dependent Variable
Following the definition of affective polarisation as hostility towards political out-groups, I oper-ationalise the dependent variable as the reverse-coded affective rating respondents attributed to the party they did not vote for in the second round of presidential elections. The original scale measured how much respondents liked or disliked mainstream parties on a 0-10 affective scale, which I then reversed such that higher values should indicate greater dislike. Note that this op-erationalisation differs from that employed by Iyengar et al. (2012), who subtract in-party from out-party ratings. This approach, however, captures in-party favouritism as well as out-group dislike. Using this difference-in-scores approach does not return significantly different results. Though Brazil is technically a multi-party system, the astonishingly high number of polit-ical parties does not reflect a diversity of social cleavages, programmatic positions or partisan identities (see Power & Zucco forthcoming). I thus focus on the presidential elections, which captures the main political dividing lines in a country where the executive is the primary and most visible source of political power (Mello & Spektor 2018). Further, I define in- and out-parties based on second-round voting not only to increase sample size, but, more importantly, because the run-off has long pitted the most relevant competitors against each other. From 1994 to 2014 political competition revolved around the PT-PSDB duopoly, with the latter being ‘replaced’ by Jair Bolsonaro’s PSL in 2018. Nonetheless, restricting the following analyses to first-round voting does not change any of the results.
The BES does not have data on how respondents feel towards other voters or partisan groups in society. As previously alluded to, it is likely that this measure of affective polarisation be capturing hostility towards parties rather than other citizens, especially given the low trust placed on political parties amongst Brazilians. A second consideration here is that the PSL does not have an established party label and primarily relied on Bolsonaro’s personalist appeal,
further weakening the link between party ratings and feelings towards its supporters. Jair Bolsonaro himself has since left the PSL, though, at the time of the 2018 election, the party did include many of his prominent political allies and was inherently linked to his agenda. As a robustness check I also ran models where feelings towards the PSL are replaced with feelings towards Bolsonaro, which did not significantly change results.
Independent Variables
Hypotheses 1 and 2 require a measure of respondents’ ideological beliefs. The traditional left-right scale, however, is ‘notoriously noisy’ in Brazil’s non-programmatic context (Samuels & Zucco, p. 40), and has been shown to be highly endogenous to (especially presidential) affect. Further, ideological labels can also be understood as a source of social identity rather than reflecting a system of beliefs (Malka & Lelkes 2010; Devine 2015), which is not part of the mechanism behind these hypotheses. I chose four question-items that have been asked in both the 2014 and 2018 waves of the BES dataset, allowing for direct comparability. These questions tap into respondents’ attitudes on wealth inequality, government spending, support for democracy, and approval of affirmative action policies. These questions tap into both the economic (equality vs free-market) and cultural (libertarianism vs authoritarianism) axes of political competition (see Kitschelt 1994), which have also been broadly reflected in the agendas associated with the PT and the PSDB/PSL( Hunter 2010; Ribeiro et al. 2011; Power & Rodrigues-Silveira 2018). All questions share a 5-point disagree/agree scale, which have been recoded so that right-wing/conservative attitudes indicate higher values.
I measure ideological distance as the absolute value of the difference in attitude scores in the 5-point scale between the respondent’s own position and the mean position of their out-party, which is itself calculated as the mean position of that party’s voters.7 I then sum the ideological
distance scores across all four questions and standardise this final number into a final z-score. In operationalising party-issue constraint, I first calculated the fraction of issue positions that are on either the left or right-wing side of the ideological divide. Scores below the overall mean (calculated across all respondents) are on the left and those above it, on the right. I then subtracted the fraction of right-wing positions from the fraction of left-wing positions (for PT voters), and vice-versa (for PSDB/PSL voters). For example, a value of 1 would mean that a PT voter held only left-wing views, whereas a value of -1 would indicate a ideologically left-wing respondent voted for the right-wing partisan option8. Finally, I standardise this measure into
a z-score. In short, ideological distance measures the extent to which an individual is distant
7For example, if the mean position of PT voters on support for democracy was 2 - on the 1-5 scale -, a
PSDB/PSL voter who scored either 1 or 3 would have an absolute ideological distance of 1 on that question.
from the other side in terms of his or her stated issue positions, whereas party-issue constraint measures the extent to which that respondent’s beliefs are on the ‘correct’ partisan side of the ideological divide.
Social sorting reflects how ‘sorted’ into their own partisan a camp a respondent is in terms of social demographics (race, income, education, religion and region of residence). For each of these categories I calculate the proportion of respondents of the same demographic group as the respondent who voted for the respondent’s party. As an example, 70% of Evangelicals in the BES 2018 sample voted for Jair Bolsonaro, meaning that an Evangelical Bolsonaro voter would receive a score of 0.7 on a ‘religious sorting’ variable. I repeat this procedure for the other five social demographic dimensions and calculate a mean sorting score, which is likewise standardised into a z-score. This should thus represent the extent to which a respondent is ‘socially aligned’ with their party of choice.
To measure partisanship I make use of two separate questions. The BES first asked respondents whether there was any political party they liked or that represented their worldview. Respond-ents who answered ‘no’ to either question were classified as nonpartisans, whereas self-declared partisans were further asked about how much they liked their own party. I use this second question in order to classify partisans as weak, medium, or strong partisans. The final meas-ure is therefore a nominal variable with nonpartisans as the baseline category. Note that this measure of partisanship does not directly tap into a sense of identification (see also Baker & Renn´o 2019 for a criticism of this ‘filtering’ first-question approach). An alternative approach would be to use in-party affective ratings, but these are much more likely to capture instru-mental evaluations rather than a sense of affection or identification. Regardless, replacing the partisanship strength scale with in-party ratings does not return substantively different results. Control Variables
I control for standard socio-demographic variables, namely age (continuous), income (five cat-egories), education (categories ranging from no formal education to tertiary), region of residence, race, religion affiliation, and sex. I also control for vote choice (‘PT Vote’ dummy), political interest (ranging from low to high self-declared interestin politics) and survey-year (with 2014 as baseline).
3.1.2 Results
Though inconsistency in question-items restricted inferential analyses to 2014 and 2018, we can leverage past BES waves for some descriptive data. Figure 1 shows how in- and out-party
ratings have changed since 2002 in Brazil. As a frame of reference, I include the same measures for the United States in 2016, a context recognised as being deeply polarised, using data from the Module 5 of the Comparative Study of Electoral Systems (CSES). After remaining mostly constant between 2002 and 2014, out-party ratings plummeted in 2018 whilst in-party ratings saw no significant changes, serving as further circumstantial evidence that hating ‘the other’ is only weakly related to how much Brazilians like their own side. It is noteworthy that even this imperfect party-based measure still manages to capture the significant rise in out-group hostility at the 2018 elections, with Brazilians disliking their out-party significantly more than Americans did in 2016.
Figure 1 Affective Ratings of In- and Out-Party in Brazil, 2002-2018
Note: Affective ratings on 0 (dislike) to 10 (like) scale. Brazil data from BES 02-18, US data from CSES Module V. In- and out-party defined based on 2nd round presidential vote choice. Sample weights applied. Dotted lines represent 95% confidence intervals
In the Appendix I provide several descriptive measures related to key independent variables. In sum, there was a slight increase in both ideological distance and social sorting between 2014 and 2018, though both these factors remain at weak levels. Party-issue constraint, and the extent
and strength of partisan attachments are unsurprisingly low.9 Put together, these descriptive findings already suggest that Hypotheses 1-4 might indeed be poor explanations for affective polarisation in Brazil due to low overall scores on the independent variables.
Model Results
Figure 2 displays a coefficient plot for all variables in an OLS model, where the dependent variable is a 0-10 scale with higher values indicating stronger dislike for the out-party. All numerical independent variables (i.e. issue distance, issue constraint and social sorting) are standardised, such that their coefficients should be interpreted as the predicted effect on the 0-10 scale associated with an increase of one standard deviation in those respective variables. Sample weights provided by the BES dataset were applied, and error bars indicate a 95% confidence interval calculated from robust standard errors.
Hypothesis 1 posited that individuals who are more ideologically distant from the other side would exhibit higher levels of out-group hostility. Results are indeed aligned with this hypo-thesis: one standard deviation increase in the level of ideological distance is associated with a statistically significant (p ≤ 0.000) increase in out-party dislike (β = 0.193). Conversely, a higher level of ideological alignment with one’s own party of choice (party-issue constraint) was not associated with a statistically significant increase greater dislike for one’s out-party (β = 0.027, p = 0.61), running counter to Hypothesis 2. Put together, these findings suggest that ideology might drive out-group hostility not so much by respondents’ ideological alignment with their own ‘side’, but rather through being distant from the ‘other’. Indeed, this would be in line with the general weakness of in-party attachments and the weak connections between ideology and party support in Brazil. Nonetheless, the overall effect of ideological distance on out-party dislike is rather small. Holding every other variable in the model at their mean value and varying ideological distance across its interquartile range (i.e. 25th to 75th percentile in the sample distribution) only increases the predicted value of out-party dislike from 7.57 to 7.85. This follows from the low levels of ideological polarisation in the Brazilian electorate, and highlights the weak explanatory power of ideology-based mechanisms. A valid criticism here concerns whether the question-items used in building these two variables are salient enough to foster out-group hostility, especially since voters were virtually undivided across the four issues in 2014. One could argue that positions on, for example, same-sex marriage and issues of law and order would be far more divisive than affirmative action or disputes over economic questions. Unfortunately, the BES 2014-2018 waves do not allow for a comparison of these measures, but I return to this question when presenting results from the online survey.
9Though an imperfect measure of negative identification, around 51% of BES 2018 claimed to dislike a
Figure 2 OLS Model Results - Brazilian Electoral Studies (BES).
Note: Dependent Variable: Out-Party Dislike (0-10). N = 2,870. Adjusted-R2 = 0.10. 95%
confid-ence intervals calculated from robust standard errors. Sample weights applied. Referconfid-ence categories for categorical variables as follows. Political Interest : Low; Income: Up to 1x min. wage; Educa-tion: Primary School; Region: Southeast; Race: White; Religion: Catholic. See Appendix for full regression table.
The results are likewise mixed when it comes to the effects of social identities on affective polarisation. Following the Social Identity Theory (SIT) mechanism behind Hypothesis 3, stronger levels of partisan identification should be associated with higher levels of out-group dislike. With nonpartisans as the baseline category, weak, medium or strong partisan identities do not return higher levels of out-party dislike that are statistically significant at the 95% level. The coefficient for strong partisans does become significant when running models with
Bolsonaro ratings replacing those of the PSL as the dependent variable for PT voters (β = 0.45, p = 0.02. See Table 4 in Appendix). Given the relatively low number of strong partisans in the analytic sample (approximately 9%), however, strong partisans would account for only a very small share of the total variation in out-party dislike. Indeed, this would be in line with my main expectation that, whilst the proposed mechanisms behind Hypotheses 1-4 could work at the individual-level, they would explain very little of the variation in affective polarisation in Brazil. Further, treating the partisanship variable as a numeric interval (i.e. ranging from 1 to 4) did not produce statistically significant results, and neither did replacing partisanship strength with a partisan dummy (i.e. simply separating partisans from nonpartisans). Again, a valid criticism here might be that the partisanship measure employed by the BES does not directly tap into a sense of identification, which might underestimate the effects on out-group hostility. Finally, social sorting does return a statistically significant association with out-party dislike (β = 0.14, p = 0.01), suggesting that affective polarisation could be partly explained by social divisions amongst the electorate. Amongst the control variables, it is worth noting that PT voters show a significantly lower level of dislike towards their out-party (β = -0.75, p ≤ 0.000) if compared to PSDB/PSL voters, whilst individuals at the highest levels of income and education are more affectively polarised than those at the lowest levels.
Perhaps the most important finding of the BES model is the low level of explained variance, as evidenced by an adjusted-R2 of 0.10. This means that, even when all variables are inserted into
the model, we can only explain around 10% of the total variation in the outcome measure (out-party dislike). Put simply, the BES model, couched in a test of Hypotheses 1 to 4, does very poorly in accounting for affective polarisation in Brazil. This becomes even more evident when running the model without any of the key independent variables (i.e. issue distance, party-issue constraint, social sorting and partisanship), resulting in a hardly-changed adjusted-R2 of 0.09. In other words, we can explain just as much of affective polarisation in Brazil simply in terms of the control variables related to vote choice and socio-demographics, eschewing the traditional sources of affective polarisation found in the literature10. As a final robustness check, I re-ran
the BES model restricting it to the 2018 election to stave off potential suggestions that, if affective polarisation did only emerge in 2018, perhaps an aggregate model might obfuscate the relevant mechanisms. This, however, is not the case, with the model returning an even lower shared of explained variance (See Table 5 in Appendix).11
In short, the results of the BES models are aligned with my overall expectation that the tradi-tional ‘sources’ of affective polaristion do not properly account for the phenomenon in Brazil. It
10In fact, removing all socio-demographic variables does not substantively decrease the explained variance of
this ‘empty’ model. The highest influence is by far exerted by the vote choice and the survey-wave dummies.
is unclear, however, whether this is at least in part due to a lack of proper operationalisation on the dependent variable. Disliking the out-party might not respond to the mechanisms predicted by existing theories. Additionally, the BES dataset imposes constraints on constructing import-ant independent variables, namely by lacking questions on potentially more divisive political issues and a measure of partisanship as a social identity. More importantly, the BES does not contain a measure of negative political identities, which I expect to reflect the most import-ant driver of affective polarisation in Brazil. In order to remedy these issues, I conducted an online survey and collected better measures of out-group hostility, ideological disagreements, and, crucially, an innovative measure of negative political identities. These survey data can provide a more robust test of the traditional mechanisms, whilst also allowing for the testing of Hypothesis 5.
4
Online Survey
4.1
Data Collection and Sample Characteristics
The online questionnaire was circulated throughout April and May 2020, reaching a total of 1,732 Brazilians aged 18 or above. To encourage participation, respondents were given the opportunity to take part in a raffle awarding two gift-cards worth 50 EUR each. Respondents came from two main sources: personal networks and the online forum reddit. My primary mode of distribution through personal networks was by sharing the survey in WhatsApp groups, spe-cifically targeting groups that are not related to politics and encompass a diverse demographic profile. WhatsApp groups are considerably popular in Brazil, especially amongst lower socio-economic classes (Spyer 2017), and are characterised by a culture of open political discussion amongst participants (Machado et al. 2019). This facilitated the sharing of the online ques-tionnaire on this platform, and I estimate that around 600 respondents (or 35% of my final sample) were reached via WhatsApp. Engagement with the questionnaire was also significant on reddit, with around 1,100 Brazilian redditors taking part in the survey over the course of a week. Though I also attempted to distribute the questionnaire via Twitter and Facebook groups, this approach did not yield many respondents.
Table 6 in the Appendix shows how my sample compares to the BES 2018 sample on key demographics. In short, the online survey respondents were more male, white, of higher income and educational levels, and younger. The biggest source of bias is, by far, education - while only 28% of BES 2018 respondents had some level of tertiary education, this was true for 90%
of survey respondents. The sample is also skewed in terms of 2018 vote choice, with only 23% (n = 289) of reported valid votes going towards Jair Bolsonaro (in contrast to the actual 55% vote share at the election). Interestingly, WhatsApp survey respondents were more varied in their vote choice and key demographics than those coming from reddit, evidencing the diverse nature of Brazilian WhatsApp users in comparison to the overwhelmingly young and left-wing reddit population. To account for some sources of unrepresentativeness, I apply weights based on gender, race, education, income, region of residence, and self-reported political interest. I use the BES 2018 rather than national census data as a proxy for the true population values in order to weigh the sample by political interest. The high skewness of certain weighting variables (in particular, education), however, means that respondents from underrepresented categories might receive extremely high weighting scores, exerting undue influence on empirical results. To remedy this issue I apply lower and upper constraints on the post-stratification weights, such that no respondent has a weight lower than 0.3 or higher than 312.
This does mean that the weighting procedure cannot satisfactorily account for sample bias in-troduced by the high baseline levels of education in my sample. Highly educated individuals tend to be more interested in - and knowledgeable about - politics (Verba et al. 1995), boasting greater cognitive resources that allow them to better process political information (Zaller 1992). This poses a particular problem for point estimates of descriptive findings. Since the survey data were collected primarily for the purposes of testing correlational relationships between variables, however, sample bias would skew the results to the extent that these relationships would behave differently amongst the higher educated. I address this issue by interacting edu-cation with all independent variables inserted into the previous BES models (i.e. issue distance, issue constraint, social sorting and partisanship strength), and find no statistically significant effects. This indicates that, at least when other variables and demographics are controlled for, highly educated individuals do not behave any differently when it comes to the hypothesised relationships. A second consideration on sample bias is that the survey respondents do show significant variation in their self-reported levels of political interest. Political interest should capture much of the biases introduced by high levels of education, and the variation in responses allows weights to reasonable account for over-representation of highly educated respondents. In sum, though unrepresentative of the Brazilian population, the survey data should allow for relatively robust tests of the mechanisms under analysis.
12In practice, a weight of 3 means the respondent would count 3 times as much as their original value.
Estimating the following analyses using different upper and lower bounds did not produce substantively different results
4.2
Operationalisation
Dependent Variable
Using the same 0-10 affective scale employed by the BES (and other CSES datasets), I collected data on how respondents felt about mainstream political parties (the PT, PSL, and PSDB), political elites (Jair Bolsonaro, Fernando Haddad, and Lula), partisans (PT and Bolsonaro supporters), and voters (2018 PT and Bolsonaro voters). I operationalise affective polarisation as the reverse-coded rating assigned to respondents’ out-voters, i.e. voters of the party/leader the respondent did not vote for in the second round of the presidential election of 2018. Whilst both out-partisans and out-voters directly capture the notion of inter-citizen hostility, I focus on the latter because, since the goal is to separate out feelings towards political elites and towards fellow citizens, ‘voters’ are more distant from the image of the party/leader than partisans are. This is of particular importance given that Hypothesis 5 is couched in a negative identification with the out-party/out-candidate, presenting the risk of endogeneity in the proposed relationship. In addition, out-voters represent a more numerous group of Brazilians, and should thus reflect a more encompassing notion of affective polarisation in the country. In any case, running the following models using out-partisans as the dependent variable does not substantively change the results.
Measures of Political Identities
An appropriate test of Hypothesis 5 requires measures of both positive and negative polit-ical identities amongst Brazilians. I build on past work that has measured partisanship as a social identity through the adaption of Mael and Tretick’s (1992) Identification with a Psy-chological Group Scale (see Greene 2002). Partisan identity scales have been used both in the United States and in multi-party systems in Europe, returning robust measurement results and outperforming traditional questions of partisan attachments in predicting political behaviour (Huddy et al. 2015, 2018; Bankert et al. 2017). In building a measure of negative political identities (NID ) I attempt to ‘reverse’ the items used in the partisanship scale (PID ) to keep these scales directly comparable, drawing on the ‘need for distinctiveness’ mechanism that un-derpins negative identities. Respondents were asked about their level of positive identification with the party (or candidate, in the case of Bolsonaro voters13) they voted for in the second
round of the presidential election, and their level of negative identification with their out-party. Respondents who indicated no vote choice in 2018 (around 25% of the final sample) were not
13To reiterate, Jair Bolsonaro does not currently belong to a political party, and his former party, the PSL,
was mainly used as a vehicle for his personalistic campaign. Negative or positive identification with the PSL would thus not capture the relevant divides in Brazilian politics.
