Views of democracy among brazilians: evolution and determinant factors¹

Visões de democracia entre os brasileiros: evolução e fatores determinantes

Abstracts

Abstract

This paper analyzes the perspective of Brazilian citizens on democracy and the individual determinant factors for each of these conceptions. This paper begins by mapping Brazilians' conceptions of democracy based on questions from the 2006, 2014, and 2018 World Values Survey, which allow us to differentiate to what extent citizens have more or less commitment to conceptions of liberal democracy, social democracy, and direct democracy, or accept certain authoritarian alternatives. Subsequently, the individual determinant factors of the different views of democracy are analyzed, including sociodemographic and attitudinal variables. The results show the stability of views of democracy among Brazilians between 2006 and 2018 and the prevalence of a liberal democracy view. Ideological positions of citizens are variously associated with different views of democracy.

Keywords:
Democracy; Views of democracy; Ideology; Political Attitudes

Resumo

Este artigo analisa quais visões os cidadãos brasileiros têm da democracia e quais os condicionantes individuais de cada uma dessas concepções. O trabalho começa por mapear as concepções dos brasileiros sobre a democracia a partir de questões do World Values Survey de 2006, 2014 e 2018 que permitem diferenciar em que medida os cidadãos têm maior ou menor compromisso com concepções de democracia liberal, social-democracia e democracia direta, ou aceitam certas alternativas autoritárias. A seguir são analisados os condicionantes individuais das diferentes visões da democracia, incluindo variáveis sociodemográficas e atitudinais. Os resultados mostram a estabilidade das visões de democracia entre os brasileiros entre 2006 e 2018 e a prevalência de uma visão de democracia liberal. Posições ideológicas dos cidadãos estão associadas de forma variada com as diferentes visões de democracia.

Palavras-chave:
Democracia; Visões da democracia; Ideologia; Atitudes Políticas

1. Introduction

Debates on political support and democratic legitimacy have undergone some important changes in recent years, especially related to two phenomena - on the one hand, analytical and methodological progress that has brought new possibilities for measuring and comparing levels of voter support for the democratic rule, its principles, institutions, and actors (Borba and Cardoso, 2021; Magalhães, 2018). On the other hand, modifications in the political environment have ensued intense new challenges to researchers. The new left-wing and right-wing populisms and their electoral expansion are important expressions of new patterns of political behavior, with voters increasingly willing to accept the violation of democratic norms by political elites in exchange for “economic growth” or “fighting corruption”, to name just two examples (Svolik, 2019).

The Brazilian case is a case in point- a third-wave democracy that, since 2013, has been experiencing a combination of political and economic crises and, in 2018, elected Jair Bolsonaro as President of the Republic, an extreme right-wing politician who has never omitted his appreciation for the military dictatorship, besides displaying a range of homophobic, racist, and sexist behaviors.

The results of the 2018 elections could indicate, at first sight, a majority rejection of democracy by the Brazilian electorate, but studies with time series data from 1989 to 2018 point out record support for the regime in the latter year (Borba and Ribeiro, 2021), which is corroborated by opinion research held on the day before the presidential elections (Datafolha, 2018). In this sense, a study by Cohen et al. (2021) identified that although Bolsonaro's campaign attracted the votes of voters who were skeptical of democracy, the support for the democratic regime expanded shortly after the election.

How to explain this apparent paradox of voters that, despite supporting democracy, elect openly antidemocratic candidates? A strategy used by literature is the use of experiments, in which it is possible to identify the people who support candidates who violate democratic principles/rules and under what conditions that happen. Results indicate that ideological or partisan extremists are the most favorable to follow these behavioral patterns when the political environment is polarized (Graham and Svolik, 2020; Svolik, 2019; Fossati et al., 2022; Carey et al., 2020).

Another aspect of the literature has resorted to a more inductive approach by interpreting democratic legitimacy as a product of the different views of democracy existing among the electorate (Ferrín and Kriesi, 2016). As for the paradox above, voting for a candidate who openly defends authoritarianism does not necessarily mean that his voters reject democracy itself, but that they hold conceptions/views of democracy compatible with this type of political speech, as is the case with majoritarian conceptions (Grossman et al., 2022; Albertus and Grossman, 2021).

From that context, this paper will examine the second aspect, seeking to identify the different views of democracy present in Brazilian voters, their variations over time (2006 to 2018), and their sociodemographic and attitudinal correlates. In the latter aspect, we dialogue with other two aspects of literature - the first one on ideological extremism (Svolik, 2019; Torcal and Magalhães, 2022), and the second one on multidimensional conceptions of ideology (Malka et al., 2017; Malka et al., 2020).

The paper is divided into three sections besides this one - in the first one, we discussed the literature on views of democracy, extremism, and multidimensional concepts of ideology. Throughout this section, we present the hypothesis of our work. The second section focuses on the methodological procedures used. Next, we introduce and discuss the results.

2. Views of democracy and democratic legitimacy

The concepts of legitimacy and political support began to be incorporated into the vocabulary of Political Science from the 1960s onwards, mainly influenced by the formulations of David Easton (1965). They went through major changes in the following years and are duly mapped and described by the literature (Borba and Cardoso, 2021; Magalhães, 2018).

This paper focused on one of these recent innovations, present in the study “How Europeans View and Evaluate Democracy” (2016), organized by Ferrín and Kriesi, which mobilized various European researchers and brought relevant contributions from theoretical, methodological, and empirical points of view.

Much of the previous research on how citizens view and evaluate democracy has been based on Easton's concept of political support, which has been used by many authors in different ways (Booth and Seligson, 2009; Dalton, 2004; Norris, 2011). The study by Ferrín and Kriesi (2016) is not linked to this tradition, thus it is not conceptually anchored in the distinction between diffuse support and specific support. The authors consider that the supporting objects (political community, regime, and authorities), and their types (specific or diffuse) are confusingly intertwined in the models deriving from Easton’s work, and a separation between them could provide a better understanding of this phenomenon. There is also confusion in the literature about how political support is defined and explained, what it is and what are its sources.

To simplify this discussion, Ferrín and Kriesi (2016) propose a conceptual alternative that would be analytically more precise. They distinguish between views, which relate to the normative ideal of democracy, and evaluations of democracy, which refer to judgments about evaluating how principles are implemented.

The distinction between vision and evaluation becomes central to the debate about legitimacy because it is the comparison between democratic ideals and the functioning of democracy that allows voters to judge a democratic regime. In most analyses, legitimacy is conceived as a concept formed by a set of evaluations of the political system or a set of supporting indicators for different objects, and there is little effort to generate a direct measure of citizens' democratic beliefs (Ferrín and Kriesi, 2016).

Ferrín and Kriesi, as well as the other authors in the collection, consider that most citizens do not adopt a single theoretical model of democracy, but rather visions that are compatible with several models at the same time. Citizens' views on democracy are analyzed in the work considering three groups: liberal, social justice, and direct. The liberal type is divided between electoral and liberal components, with the indicators related to electoral competition capturing the essence of representative democracy by assessing the importance citizens give to free and fair elections. The liberal elements, on the other hand, encompass indicators related to liberal principles, which guarantee the basic civil rights of citizens and limit the power of the state.

The other two, social justice and direct democracy, correspond to two models of democracy that go beyond the basic liberal democracy model. The first measures the extent to which citizens consider it is important for democracies to reduce income inequalities, and how much they consider that governments should protect all citizens from poverty. The second one measures the relevance of citizen participation in the conduct of public affairs, beyond the representative strategies that have voting as their main modality.

An important distinction is a systematic differentiation between individual and contextual characteristics due to the premise that it is not possible to study the citizens’ behaviors concerning democracy regardless of the context into which they are inserted. Individual characteristics play a role to determine how citizens see democracy as an ideal, but these views are not entirely independent from the concrete situation such citizens experience (Ferrín and Kriesi, 2016; Kriesi and Morlino, 2016). Contextual and individual factors are mobilized in an interconnected way to explain views and evaluations of democracy. Democratic legitimacy, in turn, is produced in the relationship between views and evaluations of democracy.

Despite questioning the context, the various studies presented by Ferrín and Kriesi (2016) have very little to say about what its effects are on different visions and how the visions themselves change over time. On this last point, we resorted to studies on public opinion dynamics, and two perspectives are useful in sketching an attempt hypothesis on the relationship between context and views. These are “loser/winner consent” (Anderson et al., 2005; Cohen et al., 2021) and “thermostatic democratic support” (Claassen 2020; Soroka and Wlezien, 2010). When examining democratic support, one viewpoint looks at how winners and losers in a democratic competition perceive the relationship. Typically, winners exhibit greater trust in institutions and satisfaction with the regime's performance, while losers tend to react differently. An alternative perspective, the thermostatic model, posits that support for democracy is a result of an adversarial dynamic with the regime's performance. Therefore, increases in democratic levels could potentially elicit contrary responses in public opinion. Despite the differences between the approaches, both have in common the fact that they perceive democratic support to vary over time, following the logic of the loser/winner or thermostatic mood.

In this paper, we understand the views of democracy among Brazilians as a product of its context, but we do not refer to the cross-country variation, but rather its time dimension, which is how the views of democracy could vary at different points in time. To that purpose, the approaches above provide elements to hypothesize that (H1) the views of democracy tend to modify over time, especially influenced by cyclical aspects such as economic and political crises or, on the other hand, periods of political stability and economic growth.

It is worth mentioning that the consequences of views and evaluations of democracy were not entirely explored and remain open the causes and consequences of them for the electorate’s political behavior (Kriesi and Morlino, 2016). When it comes specifically to the causes, recent discussions on ideology (extremism and ideology multidimensionality) can provide promising paths.

We shall begin with the relationship between ideological extremism/moderation and support for democracy. The discussion about the incompatibility between ideological extremism and the values required for the functioning of liberal democracy has a long tradition, and has received fairly systematic empirical treatment in recent decades in studies that explore various dimensions of this relationship (Torcal and Magalhães, 2022).

Graham and Svolik’s recent works (Graham and Svolik, 2020; Svolik, 2019, 2020) have brought fairly innovative contributions through the use of candidate/choice experiments held in contexts of great political polarization and the emergence of populist leaders. Their results indicate that in these contexts, voters at the ideological or partisan extremes agree to give up democracy in order to prevent opposing groups from coming to power.

Svolik (2019) emphasizes that electoral competition confronts the voter with a conflicting choice between democratic principles and partisan interest. In polarized societies, people are for or against certain leaders and would only be democrats in the background. In this sense, deep social divisions, as well as political tension (polarization), undermine the public´s ability to prevent non-liberal inclinations in elected politicians. In other words, voters trade democracy for partisanship or ideology. On the other hand, voters located in the ideological center and with weak partisan preferences, or politically independent, are the least willing to give up on democratic principles. According to Svolik, centrists play the role of guardians of democracy, a profile that does not exist in polarized societies.

Torcal and Magalhães (2022) also examine this relation through a study carried out with survey data in eleven European countries, concluding that the more individuals move away from the average ideological center of their countries (measured by the mean of the ideological self-placement), the lower their democratic support.

Svolik (2019) and Torcal and Magalhães (2022)’s studies worry about the support of liberal-democratic principles and the rejection of authoritative forms of government, but they do not consider how extremist groups stand up to democratic models that are not liberal. This is especially important for those left-wing voters, for whom a vital part of the discourse is made up of ideals of direct or participatory forms of democracy with social content. The Brazilian case is especially relevant to explore this relationship, considering the great diffusion, after the 1988 Constitution, of participatory institutions, strongly driven by left-wing governments in municipalities (Avritzer, 2009).

Considering these postulations about the relationship between extremism and support for democracy, our H2.1 is that the more extremist the voter, the less he will have a liberal view of democracy and the more he will support non-liberal views. On the other hand, it is expected (H2.2) left-wing extremists support more strongly participative and social views of democracy.

A second way to analyze the relationship between ideology and views of democracy is by multidimensional approaches to ideology. There is an intense debate about the pros and cons regarding the use of citizen’s positioning on a left-right wing scale as a good indicator among the “mass audience”, in a way that, from such positioning, it would be possible to predict partisan alignment and votes. Such a debate takes place in a global scope (Powell Jr., 2000; Andeweg, 2011; Thomassen, 2012), as well as in Latin America (Otero-Felipe and Rodríguez-Zepeda, 2010; Zechmeister and Corral, 2013) and Brazil. In our country, even though some works are postulating the use and importance of using the left-right wing scale (L-R) (Singer, 2000; Holzhacker and Balbachevsky, 2007; Izumi, 2019), much of the literature points out limitations - to a greater or lesser degree - for this use, especially regarding electorates of low political sophistication or low education.

In this debate, a significant set of studies (Treier and Hillygus, 2009; Ellis and Stimson, 2012; Baldassarri and Goldberg, 2014; Carmines and D’amico, 2015; Malka and Soto, 2015; Malka et al., 2017) has suggested that the ideology has a multidimensional nature, and, therefore, it would be necessary seek more adequate forms to measure it than just the usage of the left-right wing scale. We hereby develop mainly Malka et al. (2017)’s discussions that provide us guidance in our empirical analysis regarding some of the attitudinal variables as correlates to the views of democracy. The authors propose that a “right-left attitude organization” is not the most suitable to classify the different citizens of all countries, because many citizens are, at the same time, “right-winged” (conservative) in “cultural” terms, and “left-winged” in economic terms (they want a state that reduces inequality and protects their well-being and/or are favorable to governmental property of business and industries). An “attitudinal organization” that differentiates those seeking “protection” from those pursuing “freedom”, both in economic and cultural terms (protection-freedom attitude organization) could be better suited to encompass these sets of attitudes.

Using research data from the World Values Survey (WVS) with 229 national samples applied from 1989 to 2014, the authors corroborate their central hypothesis of the alignment between right-wing cultural views, and left-wing economic views. This pattern is more prominent in post-communist countries, less developed countries, and less committed individuals. The organization of right-left wing attitudes can be more suitable to characterize highly committed individuals and more developed and stable democracies.

In Brazil, Silva’s work is maybe the one that showed the greatest progress on ideological multidimensionality. According to Silva, “an individual is capable of being liberal when it comes to social customs and conservative regarding class economic premises” (Silva, 2017, p. 81-82). Such an author describes an operational ideology, measured from three dimensions - economic, social, and political participation. Analyzing the 1991 and 2014 WVS’s data related to Brazil, Silva concluded that “generally, people organize their political beliefs in separated scopes - and that is possibly the reason why they avoid classifying themselves as left-winged or right-winged or do that through disparate means” (Silva, 2017, p. 93).

One paper that performs a type of analysis that interests us more closely is that of Malka et al. (2020), which studies the relationship between citizens' ideology (from this typology based on their economic and cultural attitudes) and citizens' openness to authoritarianism. The analysis covers WVS data (1995-2014) for 14 Western democracies and recent Latin American Public Opinion Project (LAPOP) data for Canada and the US. The central results are as follows: in about half of the countries analyzed, the citizens most open to authoritarianism are those who mix “cultural conservatism” (of mores) and more statist positions in economic terms, i.e., voters who want the state to protect their moral values and protect them economically. In other countries, there are two groups with similar degrees of openness to authoritarianism: this same group above and those with a consistent package of right-wing attitudes, i.e., who want less state in the economy but more state protection of their conservative values.

These relationships between economic and cultural attitudes are quite complex and their analysis needs to consider context variation. Note, for example, that the demands for economic protectionism are quite different when we consider country variations in terms of socioeconomic inequalities. In the Brazilian case, one of the unequal societies in the world (Arretche, 2018), attitudes for economic protectionism (Arretche and Araujo, 2017; Mendez, 2015) can be linked to demands for democratization of political power (Holston, 2013).

It is based on these studies, especially the last one, that we formulate our H3: citizens who want the state to protect their well-being and their moral values tend to be more likely to accept non-liberal conceptions of democracy than citizens who are more “individualistic” in terms of a lower expectation of protection of their well-being by the state and more liberal in terms of moral values.²

3. Data and Methodology

In order to apply Ferrín and Kriese (2016)’s model to the Brazilian case, we focused exclusively on mapping views, identifying their evolution over a short historical data series, and emphasizing some main determinants. It is worth highlighting that Cardoso’s (2021) thesis carried out an initial analysis of the Brazilian context, stressing the influence of violence on the views of democracy, as well as the predominance of the liberal conception while associated with democracy. The study of connections in views and evaluations among Brazilians is on our agenda for future research.

The empirical material used here was provided by the World Values Survey (WVS) and was collected in the years 2006, 2014, and 2018.³ It is important to begin detailing the methodological procedures adopted through the description of the employed variables for the composition of different views of democracy, as well as the recodifications adopted. As Ferrín and Kriesi (2016) point out, most citizens do not adopt a single theoretical model of democracy; on that basis, they advocate a broad conception of democracy that includes a range of attributes. Agreeing with this general principle, instead of trying to frame national respondents as “liberals”, “social democrats”, “participatory”, or even “illiberal” by employing clustering techniques, we chose to create scales on which all individuals can score. That is, we are interested in evaluating how intensely Brazilians adhere to distinct visions of democracy, without categorizing them into groups based on arbitrary cutoffs.

With this in mind, in order to measure the so-called “Social” view, we used two variables of a battery that catalogs a series of potential features of a democratic regime and asks the respondents to indicate, on a 10 points-scale, whether or not they consider them to be fundamental features of such a regime. The first selected feature was “the government taxes rich and gives poor people money”, and the second one was “people receive unemployment compensation from the government”, both with answers varying from 1 (= it is not a fundamental feature), and 10 (= it is a fundamental feature). The answers were summed and, next, recodified to form a 0-10 scale. As for the “Liberal” view, we used three other items from the same battery - “people choose their leaders in free elections”, “citizen’s rights protect people’s liberty against oppression”, and “women have the same rights as men”. The answers were also combined through a sum procedure and were subsequently patterned on a 0-10 scale.

To compose the measure of adherence to the “Participatory” view, we used two of the well-known batteries designed by Inglehart (1990) to measure post-materialist values. In these batteries, respondents are presented with four items, and among these, they must choose the one they consider a priority as the nation's collective goal, and then indicate what their second choice would be. Each battery is presented to respondents twice so that they can made their first and second choices. In the first one, there is the item of “increasing people's participation in the decisions that are made in their work and their communities” and in the second, there is the item of “increasing people's participation in important government decisions”. Thus, we proceeded with a recoding that assigned two points for interviewees who chose the item “participatory” as a priority in these two batteries, and one point for those who indicated such items as a second option. All other choices were scored as zero. The scores for the four recoded variables (two first choices and two second choices) were added together and finally changed to the range 0 to 10.

The last view, named “illiberal”, had its adherence measured only by an item of the same battery that we used on the first two views of the fundamental features (or not) of democracy. The item is “armed forces take over the government when it is not qualified.” There was no need for recodification because the original scale was already compatible with that adopted in the other views measures.

These four measures are the objects of longitudinal analysis to identify tendencies, and they are also dependent variables in multivariate models that seek to verify their individual determinants.

As we anticipated when formulating the second hypothesis, the main independent variable in the proposed models concerns Brazilian ideological positioning. Initially, we used the classic variable of positioning on the left-right scale but to test more directly the hypothesis that relates voter extremism to illiberalism we recoded this original measure. To distinguish moderates from extremists, we recoded the scores 1 and 2 from the scale as “extreme left”, 9 and 10 as “extreme right”, and all others as “moderate”.⁴ For each vision of democracy two models were proposed, each of them using a codification of the ideological position of the interviewees. Although the average on the original scale remains stable at 5.4 throughout the period covered by the WVS surveys, the group distribution (Graph 1) indicates an expressive reduction of the moderate field, with just over 10% between the beginning and the end of the short historical series, followed by almost identical gains in the two poles.

Graph 1
Ideological Extremism, Brazil, 2006-2018. | Source: World Values Survey Association, 2023.

Two other measures of ideology were proposed based on the use of variables related to the attitudes on economic and cultural topics. Although Malka et al. (2020) operationalizes these two attitudinal groups in a single measure, we opted for the separation in economic and cultural positions because we understand that the isolated treatment could contribute to an eventual identification and comparison of the distinct effects of these predictors. The economic measure reflects, on a 10-point scale, the respondents' position between two statements: 1) the government should take more responsibility to guarantee good living conditions for everyone; and 2) people should take more responsibility for themselves. Thus, increases in scores reflect more “individualistic” attitudes (in the sense defined above).

The cultural measure was constructed by combining respondents' answers to four items in a battery about the justification of individual practices, with the following statement: “Please indicate for each of these actions whether you think they are never justified, always justified, or somewhere in between. The items selected were 1) homosexuality, 2) prostitution, 3) abortion, and 4) divorce. As the responses varied on a scale from 1 (never justified) to 10 (always justified), we added up the scores for the four items, recoded so that the combined measure took on a range of 0 to 10. Thus, higher scores reflect more liberal stances.

Graph 2 shows the evolution of means of these two measures, indicating a subtle reduction in individualistic economic attitudes and a subtle rise in the cultural measure toward more liberal stances.

Graph 2
Economic and Cultural Attitudes (mean), Brazil, 2006-2018. | Source: World Values Survey Association, 2023.

Socio-demographic (gender, age, education, employment status, and subjective social status) and attitudinal (interpersonal trust, trust in institutions, and interest in politics) predictors were also used as controls. Ceka and Magalhães (2016) argue that social status determines the version of democracy that people endorse. In this way, individuals who take over privileged positions in society tend to defend a conception of democracy more consistent with the political status quo. Furthermore, attitudinal variables employed are recurrently used in studies about democratic legitimacy, highlighting the relevance of interest in politics, interpersonal and institutional trust for strengthening democracy (Booth e Seligson, 2009).

4. Results

4.1. Views over time

We begin the presentation of the results with the evolution of the four measures of views of democracy over this short historical series. The data indicate the remarkable stability of these views, despite the troubled national political and economic trajectory between 2006 and 2018, which contradicts our first working hypothesis. Part of this stability may be due to the absence of intermediate points between the first survey and the second, separated by eight years.

Recognizing these data limitations, we can verify that there is very little variation in all measures, pointing out the stability in the views of democracy. The Liberal view stands out from the others with average scores above 8 and variation of just 0,20 (Graph 3). The Social view is in the second position, with an even lower variation in the analyzed period. Next, there is the measure of adhering to the so-called “illiberal” view, which also varies insignificantly. The measure that indicates the adherence of Brazilians to the “Participatory” view is the one presenting the lowest averages and highest variation in the period (positive) but still by just over half a point on a 10-score scale. Thus, stability is the main feature of these measures, indicating that individual characteristics play a more relevant role than attributes in political and economic contexts.

In any case, it is worth emphasizing the longitudinal predominance of the liberal view over the other conceptions of democracy among Brazilians, which is similar to the result observed among Europeans. Moreover, we would point out that adhesion to the participatory view is limited among Brazilians, and, at the same time, presents similar values to those of the illiberal conception.

Graph 3 – Evolution of measures of views of democracy, Brazil, 2006-2018 (Means) | Source: World Values Survey Association, 2023.

It is essential to contrast this permanence with the data on the turbulent reality of these years in the economic and political level - depicted, by way of illustration, in Graphs 4 and 5, where we see significant increases in unemployment and declines in the Liberal Democracy Index in 2018 compared to 2006 and 2014. In addition to the political “earthquake” of 2018, the years 2006 and 2014 were also politically tense, the former due to the scandal that became known as Mensalão (Portuguese for “big monthly bribe”) in 2005, while the latter was affected by the repercussions of the 2013 protests and the Petrolão (Portuguese for “big oil”) scandal in the context of Operação Lava Jato (Portuguese for “Car Wash Operation”)

Graph 4 – Economic Indicators (annual %), Brazil, 2001-2020. | Source: Databank – World Bank, 2023.

Graph 5 – Democracy Indices (VDem), Brazil, 2000-2021 | Source: Varieties of Democracy, 2023.

4.2. Determinants

The results of the linear models for each of the views of democracy are shown in the graphs bellow. In each of them, there are three lines for each independent variable, corresponding to the years covered by WVS.⁵

Starting with the social view (Graph 6), we can determine that there are few controls with relevant effects.

Graph 6 – Models for Social View | Source: World Values Survey Association, 2023.

In 2006, political trust had a small positive effect and education had a small negative one. In 2014, education, subjective social status, and interest in politics had a negative effect. It is important to highlight that subjective social status only has a negative effect on social view, which reinforces the argument that individuals with higher social status tend to include social justice less frequently as a way of understanding democracy (Ceka and Magalhães, 2016).

Only in 2018 did interpersonal trust present a positive significant effect. Turning to our independent variables, the measure of ideological extremism became statistically relevant only after 2014, the year in which extremist groups began to grow in percentage terms, as we analyzed earlier. In 2014, left and right extremists were more committed than the moderates to this social vision of democracy, with a slight advantage for the latter group. In 2018, however, the difference between right extremists and center disappears, with only the positive effect of being a left extremist remaining.⁶ The economic attitudes measure also proved relevant in 2018, indicating that individualists score fewer points on this measure of adherence to the social vision.

As for the liberal view (Graph 7), the effect of education presents a higher consistency of effect, covering 2006 and 2014 with positive effects. Apart from this control, only the interpersonal confidence showed a statistically significant positive impact in 2014, and the political confidence a negative impact in 2018.

Graph 7 – Models for the Liberal View | Source: World Values Survey Association, 2023.

Ideological measures present more frequent effects, with the right-wing extremists scoring higher than moderates in all three years. In 2018, an identical effect began to separate extreme left-wing individuals from moderates.⁷ Economic and cultural attitudes were also relevant in these models but with inverse effects. In 2018 and, to a lesser degree, in 2014, the ‘individualist’ economic condition caused the score for adherence to liberal view decrease, while in 2018 the progressive cultural condition presented a higher score.

The consistent effects of higher education are repeated in the model for the participatory view (Graph 8), as it extends over all three years, producing increases in scores. Another solid effect over time is produced by interest in politics, which raises adherence to this view in all three years. Gender, taking women as a reference, becomes relevant in 2006 and 2018, both with reducing effects, meaning that in those years men adhered less to the participatory view than women. Interpersonal confidence presents a positive impact in 2014. Right-wing ideological extremism has a negative effect in 2006 and 2018, with an upward trend in this last year, while no significant differences are registered between moderates and extreme left-wing.⁸ The progressive status also presents statistically significant but positive effects in 2006 and 2018.

Graph 8 – Models for the Participatory View | Source: World Values Survey Association, 2023.

Finally, the model for the illiberal view (Graph 9) confirms the relevance of education, which has significant negative effects (all above 1 point) in all three years. Thus, it is possible to highlight the relevance of education in support of the liberal view of democracy, and as a counterweight to the illiberal conception, in the same way as observed by Ceka and Magalhães (2016).

Graph 9 – Models for the Illiberal View | Source: World Values Survey Association, 2023.

In addition, the attitudinal controls of trust and interest were also important. Interpersonal trust causes a reduction in 2014, and the measure of political trust increases the score in 2006. Political interest reduces adherence to an illiberal view in 2006 and 2014.⁹ On the ideological front, left-wing extremism showed a negative effect, always taking moderates as a reference, in 2006 and 2018. The same reducing effect was observed in the “progressive” condition, but only in 2018. The “individualist” status, on the other hand, presented a small positive effect in 2006.

We also estimated a model that combines the liberalism and illiberalism scales in a dichotomous dependent measure that classifies individuals as staunch liberal or not. Staunch liberals were defined as citizens with scores equal to or higher than 8 on the liberal measure and equal to or lower than 2 on the illiberal one (full model in the Appendix, available at https://github.com/ednaldorip/democracyviews). Individuals who meet this double condition make up the group of “convinced” and show a percentage evolution, albeit subtle in the period: 19.5 in 2006, 22.8 in 2014, and 21.7 in 2018.

By including it in logistic models (Graph 10), the results confirm the relevance of education and, to a lesser extent, interest in politics, which increase the chance of being a convinced liberal in the three years, although these positive effects are reduced over the series.

Graph 10 – Models for the Convinced Liberal View | Source: World Values Survey Association, 2023.

Only in 2014, interpersonal trust has positive effects. Trust in the institutions has a negative effect both in 2006 and 2014. In extremism, the left wing has a higher probability to become convinced liberals in 2006 and 2018, whereas there is no difference between right-wing and moderates.¹⁰ In 2014 and 2018, there is a reduction in the chance of the individualist status becoming staunch liberals, while the progressive positively affects such a probability both in 2014 and 2018.

5. Discussion

The results presented above allow us to make some considerations related to the hypothesis test and its consequences for the debate about Brazilian democratic legitimacy, as well as the research agenda around this topic.

Starting with the hypothesis, in H1, contrary to what we expected, views of democracy are very stable over time and little affected by the conjuncture. It is relevant data, especially when confronted with measures of democracy support traditionally used by the literature. In a previous study, Borba and Ribeiro (2021) analyzed a temporal series (1989-2018) of the Churchillian measures and the support to the non-partisan democracy (illiberalism proxy), finding quite significant variations over time, especially in the period after 2005. For one, while in 2005, democracy support was only 55%, in 2018 it was 77%. How can we explain these differences concerning data variation when we modify the measures? It is an interesting topic to be studied in future research, considering that the work by Ferrín and Kriesi (2016) only analyzes the cross-country variation because the research that originated the book arises from a sole survey applied in various European countries. For the moment, we can predict that the attitudes related to the views of democracy are much more stable than those linked to the distinction between specific and diffuse support.

As for the determinants of views, our H2.1 was also rejected in a sense that we consider entirely counterintuitive: in our models, extremists are more liberal than moderates, which contradicts the most recent literature on the topic (Svolik, 2019; Graham and Svolik, 2020; Torcal and Magalhães, 2022). There are two possible explanations for these results: first, the problems of the left-right scale as a measure of voters' ideological positioning are recognized (Pereira, 2020; Silva, 2017). Since it is a foundation for our extremism measures, it has all the problems pointed out in the literature, of which we highlighted the wide number of non-responses and the overrepresentation of the most sophisticated voters, therefore having a selection bias. As a consequence, it is possible to suppose that such a bias reflects on the distribution of the answers, for the most sophisticated voters are those with the best ability to position themselves in the left-right scale (Almeida, 2001; Carreirão, 2007; Pereira, 2020), as well as being the most committed to the liberal-democratic values, an aspect also recorded in the literature (Reis and Castro, 1992; Moisés, 1995, 2010; Borba and Ribeiro, 2021).¹¹

A second point is that, as we mentioned in the methodology section, our measures of views of democracy were not made from the clustering of respondents, so we may have voters who are simultaneously located at the highest point in both liberal and illiberal views, for example. In other words, the views do not necessarily express an internal consistency of the voter in the form of a structured “belief system.” One way to solve this problem would be to create a measure of “democratic consistency,” made from comparing the responses to the different views. For example, a “convinced” liberal view should manifest a higher score on the items related to this measure, while rejecting the illiberal alternative. We ran a test along these lines, with the data presented in Graph 10. The results are relatively different from those presented so far, but they do not invalidate the previous finding that extremists are more liberal, although this is true only for those on the left and in the years 2006 and 2018. Right-wing extremists and moderates, on the other hand, did not differ in either year. We believe that the use of consistency measures is a promising analytical alternative and that it would fit for further exploration, but we are unable to do so in this space.

As for H2.2., data partially confirm our expectation that the left wing was more adept at participatory and social views. In the first one, left-wing extremists and moderates do not differ between themselves, but from right-wing extremists in 2006 and 2018, when the latter presented lower chances to share such democratic views. Therefore, there is a significant increase of right-wing extremists in the rejection of the participatory conception of democracy in 2018. When it comes to the social view, the difference appears between moderates and extremists in 2014, favoring the latter. In 2018, the model operates according to our expectations with left-wing extremists presenting themselves as different from the other groups, favoring a social view. In this case, the increased environment of political polarization in 2018 may have contributed to the differentiation between voters' views regarding their attachment to social content. It is worth noting, however, that the same data limitations highlighted in H2.1 are present here.

Regarding the H3 test, the data also partially confirmed our expectations, as cultural progressives consistently presented a liberal view and rejected the illiberal one, but that only occurs in 2018, although in 2014, it happens when it comes to the “Convinced liberal” variable created complementarily. As for the measure of economic protectionism/individualism, it was less consistent, as the most individualistic ones were negatively associated with the liberal view (2014 and 2018) and positively with the illiberal one (2006). Furthermore, in 2014 and 2018, there is a reduction in the chance of the economically individualistic citizens becoming staunch liberals. These results provide some elements that merit further study in relation to the political values of Brazilian liberalism. If, at the elite level, it has already been widely observed that economic liberalism has elective affinities with political authoritarianism (Trindade, 1985; Santos, 1978), this has not yet been properly thematized for the mass public. Our results provide some evidence to this direction.

6. Final Remarks

The recent discussion on democratic legitimacy in Political Science has sought to analytically face the contradiction problem of voters avowedly “democrats” who elect governors willing to violate democratic principles in a self-destruction process. Some explanations for that resort to the existence of trade-offs, which would cause voters behave contradictorily to their declared values and attitudes in order to reach their most important political interest in politically polarized environments (Svolik, 2019). An alternative is understanding the multiplicity of meanings assigned to democracy by voters who may be more or less close to the liberal ideal.

We used this second analytical strategy with an analysis of the Brazilian case. We relate this to discussions around extremism and multidimensional approaches to ideology, both of which have been used by the literature to explain political support for democracy or authoritarianism.

Despite the data limitation and the preliminary characters of the results herein presented, we found some evidence that the views of democracy are more stable than other attitudes towards democracy (such as measures of specific or diffuse support) and are related to some socio-demographic attributes or political attitudes. In the latter aspect, it is worth mentioning the proximity between participatory and social views and left-wing extremism, as well as the liberal view with “cultural progressiveness”, links that become clearer when we reach 2018 when political polarization heightens in the country. It is also worth highlighting the association found between a more individualistic economic conception - in the sense of perceiving the individuals’ well-being more as their own problems than that of the state’s - and the acceptance of an illiberal view of democracy.

We also highlight the fact that the liberal view of democracy is the predominant one among the Brazilian electorate. This is a surprising fact, considering the country's brutal economic deficiencies and inequalities, where one would expect the elements that characterize a social democracy to stand out. Despite the economic differences in relation to European countries, the liberal view also predominates in the Brazilian context.

As the next step in this research, we intend to follow the model of Ferrín and Kriesi (2016), confronting views and evaluations of democracy in the Brazilian electorate.

• 1
  A pesquisa que deu origem a esse artigo foi financiada com recursos do CNPq, através de Editais de Apoio a Pesquisa e Bolsas de Produtividade em Pesquisa. Os autores agradecem aos comentários, críticas e sugestões recebidas através dos pareceristas anônimos da Revista.
• 2
  We will use the expression "individualistic" for citizens who, in the socioeconomic dimension, rely less on state action than on themselves to ensure their social well-being. In the cultural dimension, the term "liberals" (or “progressives") refers to citizens who value individual freedom more than the intervention of the state to guarantee their moral values, while the term "conservatives" refers to those with an opposite view.
• 3
  The databases and codes necessary to replicate the results are available at https://github.com/ednaldorip/democracyviews, accessed on 10/30/2023.
• 4
  The conventional ideological positioning scale was also used in alternative models, as a strategy to test the robustness of the results. Tables with the complete results of these models can also be consulted in the Appendix mentioned in a previous note.
• 5
  The complete models, in table form, are also in the Appendix indicated above.
• 6
  This effect of the ideological extremism variable in 2014 disappears when we operationalize ideology as a continuous variable (instead of the way it was initially operationalized, as described in the methodological section).
• 7
  When we operationalize ideology as a continuous variable, only in 2014 does a statistically significant positive effect remain: when we move from left to right, adherence to the liberal vision of democracy increases.
• 8
  Only in 2018 does a statistically significant effect remain if ideology is operationalized as a continuous variable. This effect is negative, meaning that as we move from left to right, adherence to a participatory vision of democracy decreases.
• 9
  Using ideology as a continuous variable, these effects remain in 2006 and 2018: as we move from left to right, adherence to the illiberal vision of democracy increases.
• 10
  Operationalizing ideology as a continuous variable, these effects remain in 2006 and (to a lesser extent) in 2018: as we move from left to right, the probability of a person being a convinced liberal decreases.
• 11
  As an additional robustness test, we replicated all models by inserting the interaction between education and schooling, both measured continuously. The results, as shown in the Appendix tables, indicate no relevant effects.
• DOI: 10.1590/3811040/2023.

Bibliography

• ALBERTUS, Michael; GROSSMAN, Guy. (2021), “The Americas: when do voters support power grabs?”. Journal of Democracy, 32, 2:116-131. DOI: https://doi.org/10.1353/jod.2021.0023
  » https://doi.org/10.1353/jod.2021.0023
• ALMEIDA, Alberto. (2001), “À esquerda dos números, à direita dos fatos”. Insight Inteligência, 15:113-128.
• ANDEWEG, Rudy. (2011), “Approaching perfect policy congruence: Measurement, development, and relevance for political representation”, in M. Rosema; B. Denters; K. Aarts. (eds.), How Democracy Works: Political representation and policy congruence in modern societies. Amsterdam, Pallas Publications.
• ANDERSON, Christopher; BLAIS, André; BOWLER; Shaun; DONOVAN, Todd; LISTHAUG; Ola. (2005), Losers' consent: Elections and democratic legitimacy Oxford, Oxford University Press.
• ARRETCHE, Marta. (2018). “Democracia e redução da desigualdade econômica no Brasil: a inclusão dos outsiders”. Revista Brasileira De Ciências Sociais, 33, 96:1-23. DOI: 10.17666/339613/2018.
  » https://doi.org/10.17666/339613/2018
• ARRETCHE, Marta; ARAÚJO, Victor. (2017), “O Brasil tornou-se mais conservador? Apoio à redistribuição e à taxação no Brasil”. Novos Estudos CEBRAP, volume Dinâmicas da Crise, 15-22. DOI: http://dx.doi.org/10.25091/S01013300201700040002
  » http://dx.doi.org/10.25091/S01013300201700040002
• AVRITZER, Leonardo. (2009). Participatory Institutions in Democratic Brazil Baltimore, Johns Hopkins University Press.
• BALDASSARRI, Delia; GOLDBERG, Amir. (2014), “Neither ideologues nor agnostics: Alternative voters’ belief system in an age of partisan politics”, American Journal of Sociology, 120, 1:45-95. DOI: https://doi.org/10.1086/676042
  » https://doi.org/10.1086/676042
• BOOTH, John; SELIGSON, Mitchell. (2009), The Legitimacy puzzle in Latin America 1. ed. Cambridge, Cambridge University Press.
• BORBA, Julian; CARDOSO, Gabriela. (2021), “Legitimidade democrática e apoio político: Inovações recentes no debate internacional”, Opinião Pública, 27, 2:333–359. DOI: http://doi.org/10.1590/1807-01912021272333
  » http://doi.org/10.1590/1807-01912021272333
• BORBA. Julian; RIBEIRO, Ednaldo. (2021), “Adesão à democracia e educação escolar no Brasil (1989–2018): considerações a partir das teorias da legitimidade política”, Educação e Sociedade, 42, e240374:1-22. DOI: https://doi.org/10.1590/ES.240374
  » https://doi.org/10.1590/ES.240374
• CARDOSO, Gabriela Ribeiro. (2021), Vitimização, medo do crime e legitimidade democrática na América Latina: mecanismos causais e efeitos contextuais em perspectiva Tese de doutorado. Universidade Federal de Santa Catarina, Florianópolis.
• CARMINES, Edward; D'AMICO, Nicholas. (2015), “The new look in political ideology research”, Annual Review of Political Science 18, 205-216. DOI: https://doi.org/10.1146/annurev-polisci-060314-115422.
  » https://doi.org/ https://doi.org/10.1146/annurev-polisci-060314-115422
• CARREIRÃO, Yan. (2007), “Identificação ideológica, partidos e voto na eleição presidencial de 2006”, Opinião Pública, 13, 2: 307-339.
• CAREY, John; CLAYTON, Katherine; HELMKE, Gretchen; NYHAN, Brendan; SANDERS, Mitchell; STOKES, Susan. (2020), “Who will defend democracy? Evaluating tradeoffs in candidate support among partisan donors and voters”, Journal of Elections, Public Opinion and Parties, 32, 1:230-245, DOI: 10.1080/17457289.2020.1790577.
  » https://doi.org/10.1080/17457289.2020.1790577
• CEKA, Besir; MAGALHÃES, Pedro. (2016), “How people understand democracy: A social dominancy approach”, in M. Ferrín; H. Kriesi. (eds.), How europeans view and evaluate democracy New York, Oxford University Press.
• CLAASSEN, Christopher. (2020), “In the mood for democracy? Democratic support as thermostatic opinion”, American Political Science Review, 114, 1:36–53, DOI: 10.1017/S0003055419000558.
  » https://doi.org/10.1017/S0003055419000558
• COHEN, Mollie; SMITH, Amy Erica; MOSELEY, Mason; LAYTON, Matthew. (2021), “Winners’ consent? Citizen commitment to democracy when illiberal candidates win elections”, American Journal of Political Science: 67, 2:261-276. DOI: https://doi.org/10.1111/ajps.12690
  » https://doi.org/10.1111/ajps.12690
• DALTON, Russell. (2004), Democratic challenges democratic choices New York, Oxford University Press.
• DATAFOLHA. (2018), “Apoio a democracia é recorde na véspera de eleição”. Available at https://datafolha.folha.uol.com.br/eleicoes/2018/10/1982948-apoio-a-democracia-e-recorde-na-vespera-de-eleicao.shtml, accessed on 04/01/2022.
  » https://datafolha.folha.uol.com.br/eleicoes/2018/10/1982948-apoio-a-democracia-e-recorde-na-vespera-de-eleicao.shtml
• EASTON, David. (1965), A systems analysis of political life. 1. ed. New York, John Wiley & Sons.
• ELLIS, Christopher; STIMSON, James. (2012), Ideology in America Cambridge, Cambridge University Press.
• FERRÍN, Monica; KRIESI, Hanspeter. (2016), “Introduction: Democracy - the European verdict”, in M. Ferrín; H. Kriesi. (eds.), How europeans view and evaluate democracy 1. ed. New York, Oxford University Press.
• FOSSATI, Diego; MUHTADI, Burhaunddin; WARBURTON, Eve. (2022), “Why democrats abandon democracy: Evidence from four survey experiments”. Party Politics; 28, 3:554-566. DOI: https://doi.org/10.1177/1354068821992
  » https://doi.org/10.1177/1354068821992
• GRAHAM, Matthew; SVOLIK, Milan. (2020), “Democracy in America? Partisanship, polarization, and the robustness of support for democracy in the United States”. American Political Science Review, 114, 2:392–409. DOI: https://doi.org/10.1017/S0003055420000052
  » https://doi.org/10.1017/S0003055420000052
• GROSSMAN, Guy; HOLZHACKER, Denilde Oliveira; BALBACHEVSKY, Elizabeth. (2022), “The majoritarian threat to liberal democracy”. Journal of Experimental Political Science, 9, 1: 36-45.
• HOLZHACKER, Denilde; BALBACHEVSKY, Elizabeth. (2007), “Classe, ideologia e política: uma interpretação dos resultados das eleições de 2002 e 2006”. Opinião Pública, 13, 2:283-306. DOI: https://doi.org/10.1590/S0104-62762007000200003
  » https://doi.org/10.1590/S0104-62762007000200003
• HOLSTON, James. (2013). Disjunções da democracia e da modernidade no Brasil São Paulo, Cia das Letras.
• INGLEHART, Ronald. (1990), Culture shift in advanced industrial society. Princeton, Princeton University Press.
• IZUMI, Mauricio. (2019), “Ideologia, sofisticação política e voto no Brasil”. Opinião Pública, 25, 1:29-62. DOI: https://doi.org/10.1590/1807-0191201925129
  » https://doi.org/10.1590/1807-0191201925129
• KRIESI, Hanspeter; MORLINO, Leonardo. (2016), “Conclusion - What we have learnt, and where do we go from here?”, in M. Ferrín; H. Kriesi. (eds.), How Europeans View and Evaluate Democracy New York, Oxford University Press.
• MAGALHÃES, Pedro. (2018), “Regime support”, in J. Fisher. (eds.), The Routledge Handbook of Elections, Voting Behavior and Public Opinion. London, Routledge.
• MALKA, Ariel; LELKES, Yphtach; SOTO, Christopher. (2017), “Are cultural and economic conservatism positively correlated? A large-scale cross-national test”. British Journal of Political Science, 49, 3:1045-1069. DOI: https://doi.org/10.1017/S0007123417000072
  » https://doi.org/10.1017/S0007123417000072
• MALKA, Ariel; LELKES, Yphtach; BAKKER, Bert; SPIVACK, Eliyahu. (2020), “Who is open to authoritarian governance within Western democracies?” Perspectives on Politics, 20, 3:1-20. DOI: https://doi.org/10.1017/S1537592720002091
  » https://doi.org/10.1017/S1537592720002091
• MALKA, Ariel; SOTO, Cristopher. (2015), “Rigidity of the Economic right? Menu-Independent and Menu-Dependent Influences of Psychological Dispositions on Political Attitudes”. Current Direc- tions in Psychological Science, 24, 2:137–42. DOI: https://doi.org/10.1177/0963721414556340
  » https://doi.org/10.1177/0963721414556340
• MENDEZ, Yasmin. (2015), Ensaios sobre preferências por redistribuição na América Latina e no Brasil. Tese Doutorado, Universidade Federal Fluminense.
• MOISÉS, José. (1995), Os brasileiros e a democracia: bases sociopolíticas da legitimidade democrática. São Paulo, Ática.
• MOISÉS, José. (2010), “Os significados da democracia segundo os brasileiros”. Opinião Pública, 16, 2:269-309. DOI: https://doi.org/10.1590/S0104-62762010000200001
  » https://doi.org/10.1590/S0104-62762010000200001
• NORRIS, Pippa. (2011), Democratic deficit: critical citizens revisited 1. ed. New York, Cambridge University Press.
• OTERO-FELIPE, Patricia; RODRIGUEZ-ZEPEDA, Juan. (2010), Measuring representation in Latin America: A study of ideological congruence between parties and voters. 106th Annual Meeting of the American Political Science Association, Washington, September 2 to 5. Available at https://ssrn.com/abstract=1642532, accessed on 10/30/2023.
  » https://ssrn.com/abstract=1642532
• PEREIRA, Frederico. (2020), “Non causa pro causa O voto de direita e esquerda no Brasil”. Opinião Pública, 26, 2:154-179. DOI: https://doi.org/10.1590/1807-01912020262154.
  » https://doi.org/ https://doi.org/10.1590/1807-01912020262154
• POWELL JR., Bingham. (2000), Elections as instruments of democracy: Majoritarian and proportional visions New Haven, Yale University Press.
• REIS, Fábio; CASTRO, Mônica. (1992), “Regiões, classe e ideologia no processo eleitoral brasileiro”. Lua Nova, 26, 3:81-131. DOI: https://doi.org/10.1590/S0102-64451992000200004
  » https://doi.org/10.1590/S0102-64451992000200004
• SANTOS, Wanderley Guilherme dos. (1978), Ordem burguesa e liberalismo político Rio de Janeiro, Duas Cidades.
• SILVA, Thiago. (2017), Para além de esquerda e direita: a multidimensionalidade das crenças no Brasil contemporâneo (1989-2014) Tese de Doutorado, Universidade de Brasília, Brasília.
• SINGER, André. (2000), Esquerda e direita no eleitorado brasileiro São Paulo, Edusp.
• SOROKA, Stuart; WLEZIEN, Christopher. (2010), Degrees of democracy: politics, public opinion, and policy New York, Cambridge University Press.
• SVOLIK, Milan. (2019), “Polarization versus democracy”. Journal of Democracy, 30, 3:20-32.
• SVOLIK, Milan. (2020), “When polarization trumps civic virtue: Partisan conflict and the subversion of democracy by incumbents”. Quarterly Journal of Political Science, 15, 1:3–31. DOI: 10.1561/100.00018132.
  » https://doi.org/10.1561/100.00018132
• THOMASSEN, Jacques. (2012), “The blind corner of political representation”. Representation, 48, 1:13-27. DOI: https://doi.org/10.1080/00344893.2012.653229
  » https://doi.org/10.1080/00344893.2012.653229
• TORCAL, Mariano; MAGALHÃES, Pedro. (2022), “Ideological extremism, perceived party system polarization, and support for democracy”. European Political Science Review, 14, 2:188-205.
• TREIER, Shawn; HILLYGUS, Sunshine. (2009), “The nature of political ideology in the contemporary electorate”. Public Opinion Quarterly, 73, 4:679-703.
• TRINDADE, Hélgio. (1985), “Bases da democracia brasileira: lógica liberal e práxis autoritária (1822/1945)”, in A. Rouquié; B. Lamounier; J. Schvarzer. (eds.), Como renascem as democracias. São Paulo, Brasiliense.
• ZECHMEISTER, Elizabeth; CORRAL, Margarita. (2013), “Individual and contextual constraints on ideological labels in Latin America”. Comparative Political Studies, 46, 6: 675-701. DOI: https://doi.org/10.1177/00104140124638
  » https://doi.org/10.1177/00104140124638

Publication Dates

• Publication in this collection
  08 Dec 2023
• Date of issue
  2023

History

• Received
  21 Dec 2022
• Accepted
  17 Oct 2023