Tropical Trump, or a very Brazilian tale? Social class resentment as a moderating factor between anti-PT sentiment and the vote for Jair Bolsonaro in 2018

Trump tropical, ou um conto muito brasileiro? O ressentimento da classe social como fator moderador entre o sentimento anti-PT e o voto em Jair Bolsonaro em 2018

Trump tropical, ¿o un cuento muy brasileño? El resentimiento de clase social como factor moderador entre el sentimiento anti-PT y el voto a Jair Bolsonaro en 2018

Trump tropical, ou un conte très brésilien ? Le ressentiment de classe sociale comme facteur modérateur entre le sentiment anti-PT et le vote pour Jair Bolsonaro en 2018

Prior analyses of authoritarian populism have linked it to hypotheses referring to cultural backlash and negative partisanship, suggesting that conservative values and hatred of opposing parties fuel the desire for strong leadership. This article adds to the previous literature by testing the influence of social class resentment on Bolsonaro coming to power in Brazil. Based on the AmericasBarometer 2018/19 survey, the analysis highlights the central role of the rejection of the Partido dos Trabalhadores (Workers’ Party) in explaining the propensity to vote for the far-right candidate, and to a lesser extent, the influence of authoritarian values in this case study. Our findings lend some support to the hypothesis of social class resentment as well. Social class resentment was found to significantly moderate the relationship between anti-PT sentiment and voting behavior, which sheds light on Bolsonaro’s ability to capture resentful voters who were disposed to subordinate social redistribution to the defense of previously acquired privileges.

2018 Brazilian elections; social class resentment; cultural backlash; antipetismo; authoritarianism

Resumo

Estudos anteriores sobre populismo autoritário colocaram a hipótese de uma relação entre o “cultural backlash” e o partidarismo negativo, sugerindo que os valores conservadores e o ódio aos partidos de oposição alimentam o desejo de uma liderança forte. Este artigo contribui para a literatura ao testar a influência do ressentimento de classe social na ascensão do Bolsonaro ao poder no Brasil. Com base na pesquisa AmericasBarometer 2018/19, a análise revalida as conclusões de estudos anteriores, destacando o papel central da rejeição ao Partido dos Trabalhadores na explicação da propensão para votar no candidato de extrema-direita e, em menor grau, a influência dos valores autoritários neste estudo de caso. Nossas conclusões dão algum apoio à hipótese de ressentimento de classe social também. O ressentimento de classe social foi encontrado moderar significativamente a relação entre sentimento antipetista e comportamento dos eleitores, o que esclarece a capacidade de Bolsonaro de captar eleitores ressentidos dispostos a subordinar a redistribuição social à defesa de privilégios previamente adquiridos.

2018 eleições brasileiras; ressentimento de classe social; cultural backlash; antipetismo; autoritarismo

Resumen

Estudios previos sobre populismo autoritario han formulado hipótesis relacionadas con el “cultural backlash” y el partidismo negativo, sugiriendo que los valores conservadores y el odio a los partidos opositores alimentan el deseo de un liderazgo fuerte. Este artículo contribuye a la literatura poniendo a prueba la influencia del resentimiento de clase social en la llegada al poder de Bolsonaro en Brasil. El análisis, basado en datos de la encuesta del Barómetro de las Américas 2018/19, revalida las conclusiones de estudios anteriores al destacar el papel central del rechazo al Partido dos Trabalhadores (Partido de los Trabajadores) en explicar la propensión a votar por el candidato de extrema derecha y, en menor medida, la influencia de los valores autoritarios en este caso de estudio. Nuestros resultados también apoyan la hipótesis del resentimiento de clase social. Se encontró que el resentimiento de clase social modera significativamente la relación entre el sentimiento anti-petista y los comportamientos de los votantes, lo que arroja nueva luz sobre la capacidad de Bolsonaro para captar electores resentidos y dispuestos a subordinar la redistribución social a la defensa de los privilegios previamente adquiridos.

elecciones brasileñas de 2018; resentimiento de clase social; cultural backlash; antipetismo; autoritarismo

Résumé

Des études antérieures sur le populisme autoritaire ont émis l'hypothèse d'une relation entre le “cultural backlash” et la partisanerie négative, suggérant que les valeurs conservatrices et la haine des partis opposés alimentent le désir d'un fort leadership. Cette article contribue à la littérature en testant l'influence du ressentiment de classe sociale sur l'accession au pouvoir de Bolsonaro au Brésil. L'analyse, basée sur les données de l'enquête AmericasBarometer 2018/19, revalide les conclusions des études précédentes en mettant en évidence le rôle central du rejet du Partido dos Trabalhadores pour expliquer la propension à voter pour le candidat d'extrême droite, et dans une moindre mesure, l'influence des valeurs autoritaires dans cette étude de cas. Les résultats apportent également un certain soutien à l'hypothèse du ressentiment de classe sociale. Le ressentiment de classe sociale modère de manière significative la relation entre le sentiment anti-PT et les comportements des électeurs, ce qui apporte un nouvel éclairage sur la capacité de Bolsonaro à capter des électeurs rancuniers prêts à subordonner la redistribution sociale à la défense de privilèges précédemment acquis.

élections brésiliennes de 2018; ressentiment des classes sociales; cultural backlash; antipétisme; autoritarisme

Introduction

Voting behavior varies according to both structural (e.g., party identification and sociological features) and more circumstantial factors (e.g., economic grievances) ( Arzheimer et al., 2017 ). Similarly, the economic and political context is also likely to influence the motivations behind the vote, and the literature has shown how voters might behave differently depending on whether elections take place in ordinary rather than extraordinary times, the latter entailing moments of widespread economic and/or political crisis ( Lewis-Beck; Costa Lobo, 2017 ). The aim of the present research is to make further progress in understanding this issue by addressing the case of the Brazilian presidential election in 2018 and the victory of Jair Bolsonaro, which took place in the context of a combination of profound political and economic crises. Setting aside the temptation “to write off Bolsonaro’s behavior as that of an unhinged provocateur” ( Hunter; Power, 2019 , p. 76), from an analytical point of view, it is intriguing to try to make sense of his coming to power.

In the remainder of this paper, we enter into a dialogue with the most recent international literature on authoritarian populism while at the same time considering the specifics of our case study and the main reasons that drove Bolsonaro’s voters to support him in 2018. In line with preceding attempts in this field, we evaluated the hypotheses of negative partisanship (hatred of opposing parties), and a cultural backlash as the propellers of far-right voting in Brazil. However, the novelty of our approach lies in proposing a third possible explanation grounded in the concept of social class resentment, defined as a negative attitude towards a fairer distribution of wealth. Data for this study are retrieved from the 2018/19 edition of LAPOP’s AmericasBarometer survey, and hypothesis testing is based on binomial logistic regressions.

We found that, indeed, negative partisanship, and most specifically, hostility towards the Partido dos Trabalhadores (PT or Workers’ Party) is key to understand voting behavior during the presidential elections in 2018. Likewise, the significant correlation between conservative attitudes and the decision to vote for Bolsonaro provides support for the cultural backlash hypothesis. As for social class resentment, it emerges as a compelling argument in our model, especially for its moderating role in the relationship between antipetismo and the outcome variable. It seems, therefore, that Bolsonaro’s ascent was the consequence of his ability to capitalize on latent dissatisfaction against social redistribution policies promoted by PT-led governments.

The hypothesis of social class resentment and its relevance to the Brazilian case

Jair Bolsonaro’s election in October 2018 has led to a proliferation of studies concentrating on two main hypotheses in explaining his coming to power: the cultural backlash hypothesis ( Norris; Inglehart, 2019 ) – that is, that the election – was a cultural reaction to the disruptive forces that had been unchained by modernization; the so-called antipetismo , referring to hostility towards second being the so-called antipetismo hypothesis, referring to hostility toward the Workers’ Party.

As for the first of these hypotheses, the rise of authoritarian populism has been interpreted by Norris and Inglehart (2019) , as a post-WWII generational shift that caused deep changes in values in terms of environmental issues, sexual liberation, gender equality, respect for LGBTQI+ rights, the recognition of migrants and ethnic and racial minorities, cosmopolitanism, and the promotion of democracy and human rights – in summary, what has been defined as the emergence of “self-expression values” ( Inglehart; Welzel, 2005 ). Voting for populist candidates and parties is thus connected to a broader cultural backlash to this process of modernization, led by the social segments who feel impacted by it. Using data from three waves of the Brazilian Electoral Panel Study, Rennó (2020) points to the relevance of the cultural backlash hypothesis to Bolsonaro’s election, along with punitive views of crime and public security, thus echoing the law-and-order discourse, as well as the precepts of economic liberalism. Furthermore, the results of this research indicate that the perception of corruption as the main national problem was an important predictor for supporting Bolsonaro, in line with what happened in previous elections ( Jucá et al., 2016 ). According to the author, Bolsonaro and his political discourse was able to capitalize on anti-corruption feelings in the electorate, mediated by strong anti-PT feelings. Overall, and taking Rennó (2020 , p. 19) into account, Bolsonaro’s discourse had “special resonance with poorer, less educated, and younger Brazilians”, a finding that rejects the hypothesis of a cultural backlash driven by older generations. Further contributions within the framework of the cultural backlash hypothesis examined the role of anti-Black backlash in Brazil ( Bacelar da Silva; Larkins, 2019 ) showing how the 2018 presidential election functioned as a detonator for a more overt expression of racial prejudice. Similarly, Mayka and Smith (2021) consider that Bolsonaro’s rhetoric aligned with the grassroots right , defined as a socio-political movement involving a heterogenous set of actors who mobilize against abortion or LGBTQI+ rights, and in favor of gun rights and tough-on-crime policies.

A second consistent explanation, both in Brazil and internationally, is aligned with the debate on negative partisanship ( Medeiros; Noël, 2014 ; McGregor et al., 2015 ; Abramowitz; Webster, 2018 ; Mason, 2018 ), according to which negative attitudes toward an opposing group are a constitutive element of that group’s own identity. The theory of negative partisanship has been used to explain voting behavior in different contexts ( Mayer, 2017 ; Reiljan, 2019), and the debate has been rekindled by Abramowitz’s (2018) book on Donald Trump’s election. While the traditional argument on polarization had relied firmly on a division of groups based on differences of opinion regarding political issues and ideology ( Fiorina; Abans, 2008 ; Bafumi; Shapiro, 2009 ; Hill; Tausanovitch, 2015 ), Abramowitz (2018) draws attention to a new division based on ganging up against one party, instead of identifying positively with another party, which is supposed to be the most relevant political attitude in term of its capacity to account for recent American voting behavior. In line with expectations derived from Abramowitz’s conclusions, a study based on data from the Estudo Eleitoral Brasileiro ( Amaral, 2020 ) found that the main variable explaining the vote for Bolsonaro in 2018 was rejection of the PT’s political and social legacy (see also Setzler, 2020 ). Furthermore, the model proposed by Amaral (2020) highlights a significant relationship between the populist vote in Brazil, income, and educational attainment, showing that individuals with higher incomes and longer periods in formal education were more likely to have voted for Bolsonaro in 2018. Also, age and adherence to Pentecostalism emerge as significant predictors, as does the unprecedented role played by gender differences in voters’ behavior compared to previous elections, with Bolsonaro garnering much greater support from men than women. Along the same lines, Layton et al. (2021) suggested that the 2018 elections constitute a turning point in Brazilian democracy, not just because of the resurgence of the far right in the country, but also for the ability of Bolsonaro to foster a process of identity-based realignment in the electorate around a retrospective evaluation of the PT, and the emphasis on demographic cleavages by gender, race, ethnicity, and religion.

And yet, any focus on the negative partisanship hypothesis would be incomplete without considering the politicization of racial resentment as one of the major trends in present-day American politics. More specifically, taking Abramowitz and McCoy (2019 , p. 139) into account, “the empowerment of new minority groups in the form of Barack Obama’s election reinforced a sense of loss and disempowerment by white working-class voters whose economic base was shifting in a globalized economy and whose previously dominant social status was being challenged”. The role of resentment has been found to act as a propeller of the so-called protest vote also outside the US ( Mudde; Kaltwasser, 2013 ; Rooduijn, 2014 ; Moffitt, 2016 ), and this line of research has been recently enriched by new perspectives on rural resentment ( Cramer, 2016 ; Hochschild, 2016 ).

In view of the foregoing, we posit the hypothesis that negative partisanship and resentment are not just two complementary explanations of the coming into power of populist leaders but that, under certain circumstances, resentment can act as a moderating force between hatred of opposing parties and the conservative vote. In other words, we assume that the strength of the effect of negative partisanship on voting behavior would change as a function of the level of resentment in a given society. This hypothesis is grounded on a socio-historical argument. In fact, both in the US (with the Obama administration) and in Brazil (from Lula’s administration onward), political parties that have confronted populist candidates in the most recent elections, have been recognized by the conservative electorate as the custodian of the sources of resentment (e.g., protections against discrimination, civil rights enforcement, affirmative action policies to increase participation of minorities, and so forth). Looking at the Brazilian case, this is a relative new avenue of research, which deserves closer attention. In our view, Brazilian society at the time of the 2018 presidential election was suffused with resentment at different levels. First, according to Baiocchi and Silva (2018) , a growing portion of well-off sectors might have developed resentment against social groups that benefited the most from income-transfer initiatives and the redistribution of wealth. Second, Amaral’s analysis (2020) found resentment from the bottom being driven by the Evangelical church, which is popular among the poorest Brazilians and is based on a moralistic rejection of values in favor of the rights of women and LGBTQI+ minorities. Third, Layton et al. (2021) , as well as Mayka and Smith (2021) , have framed the 2018 elections within a broader resurgence of demographic and social divisions in the country. Fourth, Chagas Bastos (2019 , p. 96) finds resentment among the military entourage to which Bolsonaro belongs, which was supported by “men who had to live thirty years of resentment for not taking part in the privileges of military power”.

Against this background, it seems reasonable to assume that resentment might have been one of the drivers of Bolsonaro’s ascent into the Brazilian presidency, specifically a type of resentment based on social class cleavages, or at least, that it might have functioned as a moderating factor between anti-PT sentiment and the conservative vote. As discussed in the following section, during its almost fifteen years of uninterrupted government, the PT took the lead in a process of inclusion of the most marginalized people in Brazil into society. However, the two years preceding the 2018 election, under President Michel Temer, represented a clear turnaround, with social spending being drastically reduced (i.e., the number of recipients of the Bolsa Família social welfare program was cut down), growing inequalities of income and extreme poverty increasing by 11 percent ( IBGE, 2018 ). As such, the figure of Bolsonaro might have been seen by the upper classes as an opportunity to extend and consolidate this turnaround, in defense of what Runciman (1972) called “differential advantages” against social groups “who are perceived as gaining rewards without having made sacrifices” ( Sennett and Cobb, 1972 , p. 137).

Populism in Brazil: the rise of authoritarianism in a context of selective modernization

The hypothesis of social class resentment addresses the challenge of how to extrapolate the main conclusions of the most recent literature on populism and transfer them to the case of Brazil, where, for instance, the very ideas of cultural backlash and negative partisanship can hardly be applied fully without considering the selective process of modernization in the country. Souza (2000) coined the term “selective modernization” to indicate a process of modernization structured around inequality and social exclusion. Against this background, a focus on the Brazilian history of slavery and underdevelopment is deemed crucial to build any analytical framework that could fit the Brazilian case ( Arretche, 2015 ; Souza, 2018 ). The democratic transition was in fact inaugurated on the basis of a huge polarization of incomes inherited from the period of the military dictatorship with more than eighty percent (80%) of the workforce having fewer than eight years of formal education, and the exclusion of the unemployed from the benefits of public health-care and social security ( Menezes‐Filho; Kirschbaum, 2015 ). Such a context of enduring inequalities might explain why prior analyses have identified some patterns of voters’ behavior showing that economic voting has always been of great relevance in Brazil, along with government approval, since the advent of democracy (Carreirão; Rennó, 2019).

Under the 1988 Federal Constitution, a new system of social protection was introduced, starting what Arretche (2018) calls a process of inclusion of outsiders. Access to social rights such as health-care and education were universalized. Since then, social inequality and poverty have been tackled more effectively by state policies, this being due both to external factors (the commodities boom, for example) and the design of income-transfer policies (such as the Bolsa Família program). Beginning in 2002, the PT took the lead in this process, and social policies favoring outsiders were enhanced further.

However, more importantly as concerns this study, Singer (2012) argued that the PT’s policies have impacted on the class structure of Brazilian society, which in turn has produced changes in voting behavior, altering the party preferences of different sectors of the population and the country’s voting geography ( Soares; Terron, 2008 ). For instance, scholars have pointed to a political realignment in the country, so-called Lulismo , a reference to the PT’s former president Luiz Inácio Lula da Silva ( Singer, 2018 ), which was increasingly accompanied by antipetismo (opposition to the PT) or anti-Lulismo (opposition to Lula) ( Samuels; Zucco, 2018 ). As such, although the review of the literature reveals conflicting diagnoses regarding the role of ideology ( Singer, 2000 ; Carreirão, 2002 ; Rennó; Cabello, 2010 ) – and despite the Brazilian party system’s low level of institutionalization and extreme fragmentation – its role in structuring voting behavior has grown dramatically as a result of the two-party competition between the Partido da Social Democracia Brasileira (PSDB) and the PT in presidential elections between 1994 and 2014.

The most recent history of political confrontation in Brazil tells the story of the decline of the PT, which began with Lula handing the presidential office over to Dilma Rousseff in 2010 and continued until the protests of June 2013, in which millions took to the streets demanding an end to corruption and improvements to public services ( Tatagiba; Galvão, 2019 ). By the end of 2014, an enormous corruption scandal (Operation Lava Jato ) was added to this scenario, affecting a sizable proportion of the country’s political and economic elites, especially in the ranks of the PT. Lastly, the largest recession in the country’s economic history began in early 2015, generating unemployment rates of over thirteen percent (13%), and increasing inequality and social insecurity. In 2016, President Rousseff was impeached, corruption scandals continued to erupt as operation Lava Jato marched on, and the economy remained in recession. It was in this context that the 2018 presidential election took place.

Analytical framework

Data source

Data for this study were retrieved from LAPOP’s AmericasBarometer 2018/19 survey. Fieldwork in Brazil began on 29 January 2019, three months after the second round of the Brazilian general election, and ended on 3 March. Throughout this period, a total of 1,498 interviews were conducted. Missing data were handled with full information maximum likelihood, leading to a lower sample for this study due to the deletion of blank, null, and non-votes from the analysis ( n =586), and missing data on y variables in the different sets of hypothesis testing. As reported in the AmericasBarometer 2018/19: Brazil technical report, the sample is defined using a multi-stage probability design and stratified into Brazil’s five main geographical regions: North, Northeast, Midwest, Southeast, and South. The sample is representative of voting-age regional populations. A further sub-stratification is implemented to ensure representativeness at the urban and rural levels and by municipality size. As a result of this complex sample design, more than eight out of ten respondents in Brazil live in urban areas (86.4%) versus 13.6% of rural dwellers. Men represent 49.9% of the overall sample, as opposed to 50.1% of women. The mean age is 39.3 years, while the relative majority of interviewees self-identify as mulattos (43.9%), followed by White (27.7%) and Black (19%).

Research hypotheses and analytical approach

Building further on the input derived from the literature review, we have assumed that the arrival of Bolsonaro in the Brazilian presidency might have been the result of patterns previously depicted at the international level (i.e., negative partisanship, and cultural backlash), on the one hand, and a country-specific explanation (i.e., social class resentment), on the other. Specifically, the assumption is made that social class resentment might have influenced the vote for Bolsonaro directly, or rather functioned as a moderating factor between antipetismo and the dependent variable in our models. With the aim of testing the consistency of these hypotheses in the context of a relatively “young” democracy such as Brazil, and therefore offering a novel contribution outside the mainstream Western-centric literature, we ran binomial logistic regressions in which the endogenous variable determines whether a respondent voted for Bolsonaro or an alternative candidate. This information comes from the following question: “ Who did you vote for in the first round of the last presidential election of 2018? ” The nominal variable was divided in order to identify Bolsonaro’s voters as opposed to those who opted for another candidate.

We first run Model a (see Table 2 ), including a composite index of social class resentment, which is made up of the averaged sum of three variables identifying the degree of agreement with statements referring to the need to increase public spending to “ help the poor ”, whether it would be “ okay for the rich to pay a lot in taxes but receive little in state services ”, and the idea that “ most of the unemployed could find a job if they wanted to ”. The original scale ranged from 1 (totally disagree) to 7 (totally agree), and therefore, we reverse-coded the first two items so that highest values in the index of social class resentment would equate to a negative attitude towards a fairer distribution of wealth. To normalize the proposed index to be between 0 (no resentment) and 1 (strong resentment), we used the following formula:

$z_i=\left(\left(x_i-min(x)\right)/[(max(x)-min(x)]\right)$

Table 2
: Predictors of the vote for Bolsonaro in Brazil, unstandardized coefficients (B)

Predictors	Model a	Model b	Model c	Model d
Social class resentment	.96(.47)*			.57(.56)
Anti-PT		3.41(.29)***		3.25(.31)***
Misogyny			.27(.41)	.10(.46)
Homophobia			.88(.28)**	.43(.34)
Abortion			.18(.21)	–.03(.21)
Gun for protection			1.07(.18)***	.81(.21)***
Death penalty			.37(.17)*	.26(.19)
Neglect of democracy			.45(31)	.21(.35)
Over 35	.31(.18)	.20(.20)	.39(.21)	.24(.22)
Female	–.45(.16)**	–.35(.17)*	–.15(.17)	–.15(.18)
White	.56(.18)**	.26(.20)	.52(.19)**	.25(.22)
Higher education	.06(.12)	.01(.13)	.25(.14)	.09(.15)
Evolution of income	.10(.21)	.09(.23)	.16(.24)	–.01(.27)
Evangelical Christianity	1.09(.21)***	1.06(.23)***	1.03(.23)***	.96(.26)***
Rural	–.33(.32)	–.05(.34)	–.57(.37)	–.29(.42)
Fear of crime	.48(.23)*	.59(.24)*	.31(.23)	.44(.26)
Perceived police response	–.93(.29)**	–1.19(.34)***	–1.10(.33)***	–1.26(.37)***
N	859	859	798	786

Source: Own elaboration.
• Note: Logistic regression where vote for Bolsonaro (0=no; 1=yes) is the dependent variable.
• *** p ≤ .001; ** p ≤ .01; *p ≤ .05; std. errors in parenthesis.

where: z_i is the normalized value, x_i is the original value in the dataset, min(x) identifies the minimum value across the seven-point Likert scale, and max(x) the maximum value in the same scale.

The consistency of the social resentment index was tested by means of a single-factor confirmatory factor analysis, which fit indices endorse the goodness of its operationalization (CFI/TLI>.95; RMSEA<.03).

Model b encompasses a mean composite score ( r =.692; p≤ .001) to postulate the hypothesis retrieved from Abramowitz and Webster’s (2016) notion of negative partisanship by addressing two questions exploring “ how much [respondents] like or dislike PT ” and their opinions about the right of PT candidates to run for presidential election, both measured on a ten-point scale. In this regard, the operationalization of the negative partisanship hypothesis is rooted in the assumption that, “when we talk about antipartisanship in Brazil, we are largely talking about antipetismo” ( Samuels and Zucco, 2018 , p. 28). As for the above, the two variables have been reverse-coded and rescaled prior to computing the averaged sum of the final score.

In Model c , a vote for Bolsonaro is postulated as being associated with conservative attitudes, in line with the conclusions of Norris and Inglehart (2019) . The operationalization of cultural backlash covered six variables linked with illiberal values, namely: (1) an index of homophobic attitudes, expressed as negative opinions about same-sex marriage and disapproval of homosexuals “ being permitted to run for public office ” ( r =.575; p≤ .001); (2) an index of misogyny, defined as the averaged combination of two variables focusing on the assumptions that “ men make better political leaders than women ” and that “ when a mother works outside the home, the children suffer ” ( r =.575; p≤ .001); (3) respondents’ opinions about abortion (e.g., “ It is justified to interrupt a pregnancy when the mother's health is in danger ”); (4) a variable establishing willingness to adhere to democracy, retrieved from the following question: “ Democracy may have problems, but it is better than any other form of government. To what extent do you agree or disagree with this statement? ”, with lower values indicating greater hostility to democracy; in addition to two variables, (5) one retrieved from a question asking whether respondents “ would have a gun for protection ” and (6) another allowing respondents to be identify as being “ in favor of capital punishment for those guilty of murder ”. While the inclusion of the last two measures might appear unorthodox compared to the original proposal by Norris and Inglehart (2019, p , p. 8), these authors recognize that “authoritarian values blended with populist rhetoric can be regarded as a dangerous combination fueling a cult of fear”. Acceptance of tough-on-crime policies and harsh punishment against those who violate the penal code (e.g., owning a gun for protection or supporting capital punishment) can therefore be regarded as an implicit component of the cultural backlash hypothesis, and more so in the context of Brazil, where rose-tinted views of the military dictatorship have been found to be strongly correlated with voting behaviors in Brazil during the 2018 elections ( Rennó, 2020) . All six variables encompassed in the operationalization of the cultural backlash hypothesis have been recoded or normalized to scale their original range to 0-1.

In each of these models, the strength of our hypotheses was controlled by a set of socio-demographic, and socio-environmental variables. Specifically, nine variables are included:

• age, which is measured as a dichotomous variable splitting the sample into younger cohorts aged 16–34 years – that is, born after the end of the military dictatorship in 1985 – and a residual group aged 35 or above;

• sex (male/female);

• ethnicity (White as opposed to a residual cluster of other ethnicities);

• religion (Evangelical Christianity versus other religions);

• educational attainment, obtained by recoding the number of years enrolled in formal education into a dichotomous variable of ensino superior (higher education) against lower levels of educational attainment;

• perceived evolution of household income over the past two years (rescaled to 0-1);

• place of residence, as binary indicator classifying the address of the respondents as falling into an urban or rural area;

• the fear of crime in the neighborhood, asking, e.g., Speaking of the neighborhood where you live, and thinking of the possibility of being assaulted or robbed, do you feel very safe, somewhat safe, somewhat unsafe, or very unsafe? (rescaled to 0-1);

• and the perception of police responsiveness, asking, e.g., Suppose someone enters your home to burgle it and you call the police. How long do you think it would take the police to arrive at your house on a typical day around noon? (rescaled to 0-1).

These two latter variables are conceived as proxy measures of the deterioration of public security, which the previous literature has acknowledged as a reason for voting for Bolsonaro ( Hunter; Power, 2019 ). In addition, the perception of police responsiveness has been associated with socio‐economic background ( Brown; Benedict, 2002 ; Telles, 2014 ) in the framework linking a negative relationship of trust in the police with deprivation, thus identifying a facet of marginalization in one’s place of residence.

As a final step in our analysis, we ran a model with all the predictors ( Model d ). The models were run using Mplus software to estimate the coefficients (estimator=mlr), accounting for design effect (stratification, clustering, and unequal weighting). A description of the variables included in the full model is provided in Table 1:

Table 1
: Variables’ description

Variable	Mean	Variance
Voted for Bolsonaro (0=no; 1=yes)	.54	.49
Social class resentment (0=no resentment; 1=strong resentment)	.41	.03
Anti-PT (0= support to PT; 1=strong antipetismo)	.65	.12
Misogyny (0=in favor of women’s rights; 1= strongly against women’s rights)	.41	.05
Homophobia (0=in favor of homosexuals’ rights; 1=strongly against homosexuals’ rights)	.40	.12
Abortion (0=yes, if mother’s health is in danger; 1=never justified)	.25	.08
Gun for protection (0=against; 1=in favor)	.43	.24
Death penalty (0=against; 1=in favor)	.53	.24
Neglect of democracy ( 0=full adherence to democracy; 1=democracy is no better than any other form of government )	.34	.08
Age (0=aged 16-34; 1=over 35)	.57	.24
Sex (0=males; 1=females)	.45	.24
Ethnicity (0=non-White; 1=White)	.30	.21
Educational attainment (0=primary or middle education; 1=higher education)	.37	.12
Evolution of income ( 0=decreased; 1=increased )	.50	.12
Religion (0=other; 1=Evangelical Christianity)	.21	.17
Place of residence (0=urban; 1=rural)	.11	.10
Fear of crime ( 0=very safe; 1=very unsafe )	.51	.12
Perceived police response ( 0=less than 10 minutes; 1=police would never show )	.41	.06

Source: Own elaboration based on AmericasBarometer 2018/19.

Results

The results of the logistic regressions are presented in Table 2 . Models a and b provide support to the hypotheses of the social class resentment (B=.96; p≤ .05) and the rejection of the PT (B=3.41; p≤ .001) respectively, even after controlling for socio-demographic and socio-environmental variables. Findings in Model a point to a growing social resentment which may have turned into actual aporophobia (i.e., hostility towards the poor) and negative views with regard to redistribution (i.e., the idea that the rich should pay more taxes, or that the government should spend more on the poor). They therefore corroborate the existence of a statistically significant correlation with the likelihood of voting for Bolsonaro. On the other hand, concealing anti-PT feelings proves to be a predictor of an electoral choice in favor of Bolsonaro, which is line with the analyses of Amaral (2020) and Rennó (2020) .

The positions against homosexuals’ rights (B=.88; p≤ .01), in favor of carrying a gun for protection (B=1.07; p≤ .001), or supporting capital punishment (B=.37; p≤ .05), were found to be significant predictors of the response variable in Model c . As such, conservative positions and illiberal values share some connection with the likelihood of voting for Bolsonaro, emphasizing a macho-aggressive attitude endorsing armed self-defense, which was one of the pillars of Bolsonaro’s campaign. The influence of misogyny and neglect for democracy is not statistically significant, although the signs of the coefficients go in the expected direction, showing a sort of ambivalence about democracy and attitudes denigrating women within Bolsonaro’s electorate.

Finally, Model d reaffirms the disproportionate weight of negative partisanship as a driver of the authoritarian vote in the 2018 Brazilian elections (B=3.25; p≤ .001). As for the cultural backlash hypothesis, its influence is reduced after controlling for anti-PT, and only positions in favor of the right to keep and bear arms significantly correlate with the decision to vote for Bolsonaro (B=.81; p≤ .001) in Model d . In this regard, the results replicate the conclusions of Setzler (2020) .

The index of social class resentment is no longer significant in the full model. We looked at interaction effects to further investigate the role of the index in the full model, based on the hypothesis that social class resentment would increase the influence of anti-PT on voting for Bolsonaro. The interaction between anti-PT sentiment and the index of social class resentment was found to be statistically significant in the prediction of the vote in favor of Bolsonaro at the value of p≤.05. The test for moderation reported in Table 3 revealed that with the moderator at values approaching 1, the conditional effect of anti- petismo (anti-PT) on the outcome variable increases up to B=5.21; p≤.001. Once the interaction term is added, the meaning of the other correlates within the model remains unaltered.

Table 3
: Interaction effect between anti-PT sentiment and social class resentment

Interaction terms	B
Anti-PT*Social class resentment	3.30(1.63)* Test for moderation: $a+b^{\ast }(1)=5.21(1.06{)}^{\ast \ast \ast }$ $a+b\ast (.4)=3.23(.31{)}^{\ast \ast \ast }$ $a+b\ast (0)=1.90(.69{)}^{\ast \ast }$

Source: Own elaboration.
• Note: *** p ≤ .001; ** p ≤ .01; *p ≤ .05; std. errors in parenthesis.

As for socio-demographic and socio-environmental explanatory variables of the vote for Bolsonaro in the first round of the presidential elections in October 2018, there are a few consistencies across the four models. Being White and male is linked with a greater likelihood of voting for Bolsonaro. Evangelical Christianity is confirmed to have played a key role in 2018 elections as well ( Amaral, 2020 ). The negative correlation between the place of residence and the outcome variable indicates that urban dwellers were more prone to choose Bolsonaro in 2018, although this coefficient was not found to be statistically significant. Variables associated with the fear of crime and the perception of police responsiveness also arise as meaningful independent variables. Interpretation of the coefficients suggests that those with higher levels of a fear of crime might have seen in Bolsonaro the strong leader that would have dealt with the sources of criminal insecurity in the country. On the other hand, respondents who rely on a prompt response from the police in case of need are also more likely to have voted for Bolsonaro. As anticipated above, the perception of police responsiveness could be seen as a proxy of wealth (living in affluent areas), and the relationship between socio‐economic background and the vote seems also supported by the positive sign of the variables of educational attainment and the perception of the evolution of the household income.

Conclusions

This article has explored possible explanatory factors behind decisions to vote for Jair Bolsonaro during the general election of 2018, thus directly fitting into a growing literature with similar aims ( Bacelar da Silva; Larkins, 2019 ; Amaral, 2020 ; Davis; Straubhaar, 2020 ; Rennó, 2020 ; Setzler, 2020 ; Layton et al., 2021 ; Mayka; Smith, 2021 ). Nevertheless, our analysis adds to the current literature in two ways. First, by providing initial support for a country-specific hypothesis linked to social class resentment and negative attitudes towards a fairer distribution of wealth. We found that social class resentment was one of the drivers of the populist and authoritarian vote in Brazil in that it might have acted as a moderating factor in the relationship between anti-PT sentiments and the vote for Bolsonaro. Second, it also introduces a nuance to the cultural backlash hypothesis, which explains the authoritarian vote in Europe and the US as a reaction against the establishment, including the elites in society. However, the case of Brazil seems to indicate that elites (i.e., men and White people, reporting increased incomes, living in affluent areas) may have played a major role in electing Bolsonaro by fueling a sense of social class resentment and positioning themselves against the redistribution of wealth. To some extent, then, voting for Bolsonaro represented a statement by particular social groups of a desire to maintain the status quo and retain their previously acquired privileges. Future analyses are needed to explore further the influence of resentment in Brazilian politics and, specifically, to shed light on the possible role of mutual resentment. In this regard, a recent contribution by Layton et al. (2021) should be welcomed as it provides evidence on the effect of demographic cleavages on presidential vote choice in Brazil.

The influence of antipetismo on the decision to vote for Bolsonaro comes as no surprise, given the key role of the PT in fostering a process of social inclusion throughout its almost fifteen years of uninterrupted government. All in all, the negative partisanship hypothesis is reaffirmed by our test. In Brazil, this means that anti-PT sentiments are likely to have played a major role in opting for Bolsonaro as the future president. Yet, it remains unclear if it was antipetismo or a more tailored anti-Lulismo that motivated Bolsonaro’s base. Also, from an international perspective, it is possible that negative partisanship as such is meaningful in two-party systems (such as the US, and de facto in Brazil), but less so in multi-party systems like those in Europe. As reported in Model d , the inclusion of the composite score of negative partisanship in the model clearly attenuates the effect of other variables, including the index of social class resentment. However, after testing for moderation, we found that resentment worked as a multiplier that amplified the effect of antipetismo on the outcome variable, and this stands as one of the main contributions of this article. Future research should look closer to possible differences between racial and social class resentments, bearing in mind the idea of anti-Blackness proposed by Bacelar da Silva and Larkins (2019) .

The results reveal the influence of hyper-punitive attitudes (i.e., owning a gun for self-protection and, to a lesser extent, supporting capital punishment) on the likelihood of voting for Bolsonaro, a finding that coincides with the conclusions of Norris and Inglehart (2019) . More generally, some of the foundational arguments of the cultural backlash hypothesis are corroborated as well, for instance, showing that homophobic views went along with an increased likelihood of opting for Bolsonaro. This aligns with Mayka and Smith (2021) , and the social change advocated by the grassroots right. Security-related variables in our model point to contradictory findings, showing that Bolsonaro’s voters report higher levels of fear of crime, but also that they were more optimistic with regard to police responsiveness in case of need. We can venture two preliminary explanations for these findings. First, that Bolsonaro’s campaign was able to attract both people from vulnerable (and allegedly more unsafe) communities and better-off electors (i.e., those living in areas where the response of the police is quicker). Second, they could be seen as a consequence of the mismatch between subjective and objective safety (on the topic of the fear of crime paradox, see Hale, 1996 ; Ditton; Farrall, 2000 ; Lee, 2007 ), thus suggesting that one might feel unsafe even if one lives in a place with a reliable mechanism(s) of formal control.

As for the limitations of our study, the retrospective nature of our analysis of Brazilians’ electoral choices, together with the fact that we were only able to analyze electoral preferences during the first round of the elections, are the most evident. Panel data and ad hoc electoral surveys would probably have suited our purposes best. Nevertheless, two nuances justify our analytical strategy. First, LAPOP’s AmericasBarometer survey addresses a broader set of social issues, unlike other surveys. Second, the mismatch in time that existed between the survey’s fieldwork (January-March 2019) and the second round of the elections (October 28, 2018) was relatively short – thus allowing reliable measures of party competition and ideological alignment.

Therefore, is Bolsonaro ultimately a tropical Trump, or was the Brazilian’s rise due to structural and conjunctural factors that are inherent in and specific to Brazil’s socio-political history? Indeed, Bolsonaro’s mandate constitutes a political laboratory with its own characteristics and with an uncertain outcome, especially in the context of a highly polarized country like contemporary Brazil. We can nevertheless propose two arguments in favor of the uniqueness of the Brazilian socio-political scenario. First, it is worth recalling the circumstances in which the electoral campaign took place. After Bolsonaro was stabbed at the beginning of September 2018, his campaign was in effect cancelled. Yet, he kept appearing on mainstream media in violation of government regulations controlling candidates’ access to TV and radio, which definitely helped him consolidate his position in the political arena ( Hunter; Power, 2019 ). More generally, Bolsonaro was a political opportunist who appeared on the stage at the right moment. As Hunter and Power (2019 , p. 70) argue, “in a country in which one out of three members of Congress was under either indictment or investigation for criminal activity, Bolsonaro’s previous political insignificance proved a boon”, and he clearly took advantage of that. Second, and more directly connected to the empirical findings presented in this article, our preliminary insight is that, even though Bolsonaro’s rise is not unconnected with similar processes described internationally, it nevertheless has specific features that were not entirely revealed by previous analyses. Our results emphasize how the main difference between Bolsonaro and other populist leaders might rest in his ability to capitalize a wave of resentment “from the top” and to capture those electors who were disposed to subordinate civic and human rights to the defense of acquired privileges.

Bibliographic references

• Abramowitz, A. The great alignment. Race, party transformation, and the rise of Donald Trump. New Haven and London: Yale University Press, 2018.
• Abramowitz, A.; McCoy, J. “United States: racial resentment, negative partisanship, and polarization in Trump’s America". The ANNALS of the American Academy of Political and Social Science, vol. 681, n^o1, p. 137-156, 2019.
• Abramowitz, A.; Webster, S. “The rise of negative partisanship and the nationalization of U.S. elections in the 21st century". Electoral Studies, vol. 41, p. 12-22, 2016.
• Abramowitz, A.; Webster, S. “Negative partisanship: why Americans dislike parties but behave like rabid partisans". Advances in Political Psychology, vol. 39, n^oS1, p. 119-135, 2018.
• Amaral, O. “The victory of Jair Bolsonaro according to the Brazilian electoral study of 2018”. Brazilian Political Science Review, vol. 14, n^o1, p. 1-13, 2020.
• Arretche, M. Trajetórias das desigualdades: como o Brasil mudou nos últimos 50 anos. São Paulo: Editora da Unesp/CEM-USP, 2015.
• Arretche, M. “Democracia e redução da desigualdade econômica no Brasil: a inclusão dos outsiders”. Revista Brasileira de Ciências Sociais, vol. 33, n^o96, p. 395-417, 2018.
• Arzheimer, K.; Evans, J.; Lewis-Beck, M. S. The SAGE handbook of electoral behaviour. London: Sage, 2017.
• Bacelar da Silva, A. J.; Larkins, E. R. “The Bolsonaro election, antiblackness, and changing race relations in Brazil". The Journal of Latin American and Caribbean Anthropology, vol. 24, n^o4, p. 893-913, 2019.
• Bafumi, J.; Shapiro, R. Y. "A new partisan voter". The Journal of Politics, vol. 71, n^o1, p. 1-24, 2009.
• Baiocchi, G.; Silva, M. K. Who supports Brazil's new strongman? [online]. Boston Review, 12 October 2018. Available at: <http://bostonreview.net/world/gianpaolo-baiocchi-marcelo-k-silva-who-supports-brazils-new-strongman> Accessed on: April 19, 2022.
  » http://bostonreview.net/world/gianpaolo-baiocchi-marcelo-k-silva-who-supports-brazils-new-strongman>
• Barbalet, J. M. Emotion, social theory, and social structure. A macrosociological approach. Cambridge, UK: Cambridge University Press, 2004.
• Brown, B.; Benedict, W. “Perceptions of the police: past findings, methodological issues, conceptual issues, and policy implications”. Policing: An International Journal of Police Strategies & Management, vol. 25, n^o3, p. 543‐580, 2002.
• Carreirão, Y. A decisão do voto nas eleições presidenciais brasileiras. Rio de Janeiro: Fundação Getúlio Vargas, 2002.
• Carreirão, Y.; Rennó, L. Presidential voting: partisanship, economy and ideology. In: Ames, B. (ed.). Routledge Handbook of Brazilian Politics. New York: Routledge, p. 216-235, 2018.
• Chagas Bastos, F. “Political realignment in Brazil: Jair Bolsonaro and the right turn". Revista de Estudios Sociales, vol. 69, p. 92-100, 2019.
• Cramer, K. J. The Politics of Resentment. Rural consciousness in Wisconsin and the rise of Scott Walker. Chicago: University of Chicago Press, 2016.
• Davis, S.; Straubhaar, J. "Producing antipetismo: media activism and the rise of the radical, nationalist right in contemporary Brazil". The International Communication Gazette, vol. 82, n^o1, p. 82-100, 2020.
• Ditton, J.; Farrall, S. The fear of crime. Burlington, VT: Ashgate/Dartmouth, 2000.
• Fiorina, M.; Abans, S. "Political polarization in the American public". Annual Review of Political Science, vol. 11, p. 563-588, 2008.
• Hale, C. "Fear of crime: a review of the literature". International Review of Victimology, vol. 4, n^o2, p. 79-150, 1996.
• Hill, S. J.; Tausanovitch, C. "A disconnect in representation? Comparison of trends in congressional and public polarization". The Journal of Politics, vol. 77, n^o4, p. 1.058-1.075, 2015.
• Hochschild, A. R. Strangers in their own land. Anger and mourning on the American right. New York: The New Press, 2016.
• Hunter, W.; Power, T. “Bolsonaro and Brazil's illiberal backlash”. Journal of Democracy, vol. 30, n^o1, p. 68-82, 2019.
• Ibge. Pesquisa Nacional por Amostra de Domicílios Contínua. IBGE, 2018 Available at: <https://www.ibge.gov.br/estatisticas/sociais/trabalho.html>. Accessed on: September 14, 2022.
  » https://www.ibge.gov.br/estatisticas/sociais/trabalho.html
• Inglehart, R.; Welzel, C. Modernization, cultural change, and democracy. The human development sequence. Cambridge: Cambridge University Press, 2005.
• Jucá, I.; Melo, M. A.; Rennó, L. "The political cost of corruption: scandals, campaign finance, and reelection in the Brazilian chamber of deputies". Journal of Politics in Latin America, vol. 8, n^o2, p. 3-36, 2016.
• Layton, M. L., et al. “Demographic polarization and the rise of the far right: Brazil’s 2018 presidential election". Research and Politics, vol. 8, n^o1, 2021.
• Lee, M. Inventing fear of crime. Criminology and the politics of anxiety. Portland, OR: Willan, 2007.
• Lewis-Beck, M.; Costa Lobo, M. Economic voting in ordinary and extraordinary times. In: Arzheimer, K., et al. (eds.). Sage handbook of electoral behaviour. London: Sage, p. 609-629, 2017.
• Mason, L. Uncivil agreement. How politics became our identity. Chicago: University of Chicago Press, 2018.
• Mayer, S. J. “How negative partisanship affects voting behavior in Europe: evidence from an analysis of 17 European multi-party systems with proportional voting". Research & Politics, vol. 4, n^o1, p. 1-7, 2017.
• Mayka, L.; Smith, A. E. “The grassroots right in Latin America: patterns, causes, and consequences". Latin American Politics and Society, vol. 63, n^o3, p. 1-20, 2021.
• Mcgregor, R. M.; Caruana, N.; Stephenson, L. "Negative partisanship in a multi-party system: the case of Canada". Journal of Elections, Public Opinion and Parties, vol. 25, n^o3, p. 300-316, 2015.
• Medeiros, M.; Noël, A. "The forgotten side of partisanship: negative party identification in four Anglo-American democracies". Comparative Political Studies, vol. 47, n^o7, p. 1.022-1.046, 2014.
• Menezes‐Filho, N.; Kirschbaum, C. Educação e desigualdade no Brasil. In: Arretche, A. (ed.). Trajetórias das desigualdades: como o Brasil mudou nos últimos 50 anos. São Paulo: Unesp/CEM-USP, p. 109-131, 2015.
• Moffitt, B. The global rise of populism. Performance, political style, and representation. Stanford: Stanford University Press, 2016.
• Mudde, C.; Kaltwasser, C. R. "Exclusionary vs. inclusionary populism: comparing contemporary Europe and Latin America". Government and opposition, vol. 48, n^o2, p. 147-74, 2013.
• Norris, P.; Inglehart, R. Cultural backlash. Trump, Brexit, and authoritarian populism. New York: Cambridge University Press, 2019.
• Reiljan, A. “Fear and loathing across party lines (also) in Europe: affective polarisation in European party systems". European Journal of Political Research, vol. 59, n^o2, p. 376-396, 2019.
• Rennó, L. "The Bolsonaro voter: issue positions and vote choice in the 2018 Brazilian presidential elections". Latin America Politics and Society, vol. 62, n^o4, p. 1-23, 2020.
• Rennó, L.; Cabello, A. F. “The basis of Lulismo: the return of personalism. Ideological realignment or non-alignment?”. Revista Brasileira de Ciências Sociais, vol. 25, n^o74, p. 39-60, 2010.
• Rooduijn, M. “Vox populismus: a populist radical right attitude among the public?". Nations and Nationalism, vol. 20, n^o1, p. 80-92, 2014.
• Runciman, W. G. Relative deprivation and social justice. A study of attitudes to social inequality in twentieth century England. Harmondsworth: Penguin, 1972.
• Samuels, D.; Zucco, C. Partisans, antipartisans, and nonpartisans. Voting behavior in Brazil. Cambridge: Cambridge University Press, 2018.
• Sennett, R.; Cobb, J. The hidden injuries of class. New York: Vintage, 1972.
• Setzler, M. “Did Brazilians vote for Jair Bolsonaro because they share his most controversial views?". Brazilian Political Science Review, vol. 15, n^o1, 2020.
• Singer, A. V. Esquerda e direita no eleitorado brasileiro. São Paulo: Edusp, 2000.
• Singer, A. V. O Lulismo em crise: um quebra-cabeça do período Dilma (2011-2016). São Paulo: Companhia das Letras, 2018.
• Singer, A. V. Os sentidos do Lulismo: reforma gradual e pacto conservador. São Paulo: Companhia das Letras, 2012.
• Soares, G.; Terron, S. L. “Dois Lulas: a geografia eleitoral da reeleição. Explorando conceitos, métodos e técnicas de análise geoespacial”. Opinião Pública, vol. 14, n^o2, p. 269-301, 2008.
• Souza, J. A modernização seletiva. Uma reinterpretação do dilema brasileiro. Brasília: UNB, 2000.
• Souza, P. Uma história das desigualdades. Concentração de renda entre os ricos no Brasil. São Paulo: Hucitec, 2018.
• Tatagiba, L.; Galvão, A. “Os protestos no Brasil em tempos de crise (2011-2016)”. Opinão Pública, vol. 25, n^o1, p. 63-96, 2019.
• Telles, E. Pigmentocracies. Ethnicity, race, and color in Latin America. Chapel Hill: The University of North Carolina Press, 2014.

Publication Dates

• Publication in this collection
  26 May 2023
• Date of issue
  Jan-Apr 2023

History

• Received
  14 Dec 2021
• Accepted
  26 Jan 2023