Corruption in the perception of Brazilian society: persistence and resilience

A corrupção na percepção da sociedade brasileira: persistência e resiliência

La corrupción según la percepción de la sociedad brasileña: persistencia y resiliencia

La corruption selon la perception de la société brésilienne : persistance et résilience

Political corruption is one of the most severe and complex problems facing new and old democracies. Fundamentally, it involves the abuse of public power for any kind of private benefit, including advantages to governing parties over their opposition. It thus defeats the principle of political equality inherent in democracy, as its protagonists can obtain or maintain power and political benefits disproportionate to what they would achieve through legitimate and forms of political competition. At the same time, it distorts the republican dimension of democracy because it makes public policies result, not from debate and public dispute between different projects, but from behind-the-scenes agreements that favor spurious interests. Despite the efficiency of classical methodological instruments such as opinion polls with structured questionnaires and qualitative research with focus groups, it is still possible to identify biases in the quality of the information collected on corruption. This article aims to address this issue by introducing two survey experiments, one conducted in 2014 and one in 2018. The 2014 survey indicated that there is a substantive difference between the direct question and the experimental question. The level of tolerance for corruption is much higher than the other methodological instruments point out. Comparison with the 2018 experiment allowed the validation and extrapolation of the results found by Moisés and Nunes de Oliveira (2018).

corruption; persistence; resilience; public opinion; bias; experimental design

Resumo

A corrupção política é um dos problemas mais graves e complexos enfrentados pelas novas e velhas democracias. Fundamentalmente, ela envolve o abuso de poder público para qualquer tipo de benefício privado, incluindo vantagens para partidos de governo. Desqualifica, assim, o princípio de igualdade política inerente à democracia, pois seus protagonistas podem obter ou manter poder e benefícios políticos desproporcionais ao que conseguiriam por meio de formas legítimas e legais de competir politicamente. Ao mesmo tempo, a corrupção distorce a dimensão republicana da democracia porque faz com que as políticas públicas resultem, não do debate e da disputa pública entre diferentes projetos, mas de acordos de bastidores que favorecem interesses espúrios. Apesar da eficiência de instrumentos metodológicos clássicos como pesquisas de opinião com questionário estruturado e pesquisas qualitativas com grupos focais, ainda é possível identificar vieses na qualidade das informações coletadas sobre corrupção. Este artigo visa abordar essa questão apresentando dois experimentos de pesquisa, um realizado em 2014 e outro em 2018. A pesquisa de 2014 indicou que há uma diferença substantiva entre a pergunta direta e a pergunta experimental. O nível de tolerância à corrupção é muito superior ao apontado por outros instrumentos metodológicos. A comparação com o experimento de 2018 permitiu a validação e extrapolação dos resultados encontrados por Moisés e Nunes de Oliveira (2018).

democracia; corrupção política; opinião pública

Resumen

La corrupción política es uno de los problemas más graves y complejos que enfrentan las nuevas y viejas democracias. Fundamentalmente, implica en el abuso del poder público para cualquier tipo de beneficio privado, incluidas las ventajas de los partidos de gobierno sobre su oposición. Derrota así el principio de igualdad política inherente a la democracia, ya que sus protagonistas pueden obtener o mantener poder y beneficios políticos desproporcionados a los que lograrían a través de formas legítimas y legales de competir políticamente. Al mismo tiempo, distorsiona la dimensión republicana de la democracia porque hace que las políticas públicas resulten no del debate y la disputa pública entre diferentes proyectos, sino de acuerdos entre bastidores que favorecen intereses espurios. A pesar de la eficiencia de los instrumentos metodológicos clásicos, como las encuestas de opinión con cuestionario estructurado y la investigación cualitativa con grupos focales, aún es posible identificar sesgos en la calidad de la información recolectada sobre corrupción. Este artículo tiene como objetivo abordar este problema mediante la introducción de dos experimentos de encuesta, uno realizado en 2014 y otro en 2018. La encuesta de 2014 indicó que existe una diferencia sustancial entre la pregunta directa y la pregunta experimental. El nivel de tolerancia a la corrupción es muy superior al que señalan los demás instrumentos metodológicos. La comparación con el experimento de 2018 permitió la validación y extrapolación de los resultados encontrados por Moisés y Nunes de Oliveira (2018).

democracia; corrupción política; opinión pública

Résumé

La corruption politique est l'un des problèmes les plus graves et les plus complexes auxquels sont confrontées les nouvelles et les anciennes démocraties. Fondamentalement, cela implique l'abus du pouvoir public pour tout type d'avantage privé, y compris les avantages des partis au pouvoir par rapport à leur opposition. Elle va ainsi à l'encontre du principe d'égalité politique inhérent à la démocratie, puisque ses protagonistes peuvent obtenir ou maintenir un pouvoir et des avantages politiques disproportionnés par rapport à ce qu'ils obtiendraient par des voies légitimes et légales de compétition politique. En même temps, elle dénature la dimension républicaine de la démocratie car, à cause d'elle, les politiques publiques résultent non pas de débats et de discussions publiques entre différents projets, mais d'accords en coulisses qui favorisent des intérêts fallacieux. Malgré l'efficacité des instruments méthodologiques classiques tels que les sondages d'opinion avec questionnaire structuré et la recherche qualitative avec focus group , il est encore possible d'identifier des biais dans la qualité des informations collectées sur la corruption. Cet article vise à résoudre ce problème en introduisant deux expériences d'enquête, l'une menée en 2014 et l'autre en 2018. L'enquête de 2014 a indiqué qu'il existe une différence substantielle entre la question directe et la question expérimentale. Le niveau de tolérance à l'égard de la corruption est beaucoup plus élevé que ne le signalent les autres instruments méthodologiques. La comparaison avec l'expérience de 2018 a permis la validation et l'extrapolation des résultats trouvés par Moisés et Nunes de Oliveira (2018).

démocratie; corruption politique; opinion publique

Introduction

Corruption is one of the main problems faced by democracies. The illegal appropriation of financial resources by public servants alters the balance of political forces and weakens the republican character of electoral disputes.

Due to the centrality of the theme to the quality of democracy, which justifies the increasing academic attention on corruption, there are still significant methodological gaps that need to be explored. Two points deserve to be highlighted: first, the real impact of corruption on the political and social behavior of individuals and second, the identification of variables that explain the prevalence of corruption in some regions of the world. If, toward the first dimension, researchers need to advance methodologically in the empirical verification of the impact of corruption on the quality of democracy, in the second point, they should aim to develop a more assertive understanding of the phenomenon itself.

In this article, the basic premise is that to deal with these two issues it is necessary to refine the instruments for collecting information on corruption. As will be shown in the next sections, the traditional data collection instruments on corruption - namely, perception surveys of either elites or the general population - have produced inadequate measures of the prevalence of corruption and individual attitudes towards the phenomenon. We argue that there is an important bias in these information collection instruments that needs to be considered. To demonstrate our point, we analyze individual perceptions of corruption in Brazil.

In 2021, Brazil completed 33 years of post-authoritarian democratic experience. Today, it is among the largest electoral democracies in the world, behind only India, the United States, and Russia, with nearly 150 million voters out of a total population of just over 212 million inhabitants. The Brazil's electoral cycles for government offices, suspended in April 1964 by the military regime which usurped power until 1985, take place according to constitutional rules that guarantee the alternation of power. After more than two decades of authoritarian rule, the military returned governance to civilian rule and resumed their professional duties answering to elected authorities - although still intervening in specific political conjunctures such as the 2018 presidential election. Under democratic rule, Brazil has enjoyed greater guarantees of: individual freedoms and citizen rights; freedom of expression for the press and media; and political participation of opposition parties without restrictions or threats to their freedoms. At the time of writing, three years ago the right-wing populist Jair Bolsonaro, who often expressed nostalgia for Brazil's authoritarian past, was democratically elected. After the promulgation of a democratic Constitution in 1988 and during the next 30 years, the principles of participation and contestation, which define democracy according to Dahl (1972) , were fairly guaranteed in Brazil. Nevertheless, these guarantees came under serious risk with the rise of the former army captain Jair Bolsonaro as President of the Republic; his government persistently threatened democratic institutions and pressured the military with inflammatory reactionary rhetoric to violate the constitutional principles that limit its activities.

Democratization processes of the last 50 years in various parts of the world have shown that the institutionalization of democracy is not a linear or one-dimensional phenomenon, nor does it produce a single model of regime. Democratization can combine different constitutional arrangements, such as the well-known classification of Lijphart (1984) on the majoritarian and consensus models of democracy. These different arrangements affect the functioning of democracy, impacting the performance of institutions and influencing the State's relations with society, as in the Brazilian case. The country combines majority and consensual characteristics by offering a majoritarian system for Executive positions and for Senate, and a proportional system for the people's representatives in the House of Representatives, state Legislative Assemblies, and city Councils. The political system thus has a double source of legitimacy, and elections to the Executive branch do not condition elections to parliament, while the proportional representation of open-list candidates feeds a multiparty system that is combined with the presidentialism of imperial characteristics - this is made possible by the formation of large political coalitions. The multiparty fragmentation practically prevents presidents from obtaining a majority on their own, and the guarantee of governability is dependent on negotiations for the formation of governing coalitions, which in a way holds presidents' hostage to the large number of parties represented in parliament (Abranches, 1988).

Nevertheless, the mainstream of Brazilian political science maintains that this institutional structure ensures the necessary conditions for governments of different political hues to coordinate their forces in the National Congress and achieve success in their shared platforms. However, democracy is not fully consolidated in Brazil: the regime has important distortions that make Brazilian democracy unstable and compromise its quality. The separation of powers is not always respected, and the relationship between the Executive, Legislative, and Judiciary branches follows a conflictive pattern of negative consequences, such as the frequent judicialization of issues that, despite falling within the scope of parliament or more properly its relationship with the government, ultimately amounts to a Supreme Court decision. In addition, the country has gone through two impeachment processes of democratically elected presidents in less than 25 years. In both cases, the presidents were charged with abuse of power. The charges drove Brazil's National Congress and Supreme Court to pursue impeachment to the letter of the law. The results were traumatic, provoking conflicts between political parties and sectors of civil society, generating a chain of attitudes and behaviors of political disbelief, and raising doubts about the legitimacy of the regime, political intolerance, and institutional instability.

At the same time, the systemic nature of abuse of power practices, as is the case of corruption and disrespect for laws related to fiscal control, indicates that the rule of law is not fully consolidated in the country and that the system of checks and balances does not operate with the efficiency democracy requires. The scale of corruption practices and their continuity over time are signs that the interinstitutional accountability system is malfunctioning. The 1988 Constitution guaranteed important advances for the formation and consolidation of a Public Integrity System - with the improvement of legislation to combat corruption, autonomy of control institutions, and the combined efforts of the Federal Police, the Federal Public Prosecutor's Office, and the judicial system - allowing the so-called Operation Car Wash to unveil, in recent decades, a framework of actual sequestering of strategic spheres of the State by business sectors associated with leaders of political parties and bureaucrats of State-owned companies such as the oil company Petrobras.

This panorama had a profound influence on the results of the 2018 elections, since, in addition to signaling most voters' severe rejection of the political cycle initiated in the mid-1980s, it undermined the leading role of the political parties responsible for democratization, such as the MDB, PT, and PSDB, all of which were implicated in the criminal corruption scandal uncovered by Operation Car Wash. The crisis these parties experienced opened space for the emergence of a personalist populist leader who, apart from rejecting political pluralism, fed the moral opposition between the 'pure populace' and the 'corrupt elite', presenting itself as an alternative to the 'politically correct', supposedly associated with the rights of minorities considered morally unacceptable and politically dangerous, including for their alleged support for parties involved in corruption.

This has added to long-standing imbalances in electoral mechanisms marked by inequalities in the voting process. Voting is the main instrument through which citizens exercise their sovereignty, guarantee human rights, choose rulers, and defend their interests. However, this requires two conditions that differentiate democracy from its alternatives: first, it must provide the means whereby citizens are guaranteed rights that ensure their freedom and uphold the principle of self-governance; second, it must provide the governed one of the most effective instruments of control over those who govern, namely accountability mechanisms. Nevertheless, such guarantees depend on how well the electoral system transposes voters' wishes into policy. If these conditions are not met, people become frustrated with politics, lose their trust in institutions, and begin to doubt democracy, as is currently the case in Brazil.

In this sense, the country's electoral system has failings that compromise the function of representation. These shortcomings entail distortions in the process of choosing representatives as there are different ceilings on seats for each state in the Chamber of Deputies; votes in certain states have more weight than in others, which goes against the principle of equal representation. Furthermore, Brazil's open-list proportional representation system, with candidates chosen in districts that can reach up to 35 million voters, makes it difficult for voters to define their preferences in the face of an excessive number of candidates, stimulating intra-party competition among candidates, weakening the connection between representatives and their constituents and compromising the principle according to which citizens evaluate the performance of their elected officials.

Such factors, along with permissive party legislation, make the Brazilian case one of extreme party fragmentation. Faced with an ineffective barrier clause, at the time of writing, more than 30 parties are represented in the House of Representatives; 21 parties in Senate; and another 50 seeking registration, which is not to say that voters feel better represented by them. Parties have little meaning in terms of disputes over different programs and political conceptions; they number 13 (actual / "effective" number), which aggravates problems of governability and increases their irresponsibility ( Laakso; Taagepera, 1979 ). Another distortion was the system of party coalitions in proportional elections, a mechanism that defrauds voters who, by voting for a candidate on the basis of political-ideological affiliation, could help indirectly elect a candidate whose ideology or party is diametrically opposed to their partisan or ideological preferences. This practice was suspended by a partial political reform in 2017, but the majority of party leaders are now, at the time of writing, in favor or re-establishing the practice in the elections 2022.

The shortcomings in the system of representation also include the inequality of inclusion of different segments of the population in politics, starting with women, whose representation in parliament is less than 15% despite comprising more than 52% of the population. Indeed, the results of the 2018 elections showed that Brazil had the worst performance among South American countries in this regard. A law from the 1990s ensures female candidates represent 30% of the party lists, recognizing that women are under-represented. At 25%, the representation of African descendants is almost double that of women, but of these segments (55%) of the population are still far from achieving full participation. In 2018, an indigenous woman was elected to the Chamber of Deputies for the first time; nevertheless, the presence of indigenous people is less than 0.2%. In short, inequality of political participation and representation of different segments of the population is still prevalent in the country. These shortcomings are directly related to the financing model of electoral campaigns. The distribution of campaign funds - a clear factor of electoral success - has always been subordinated to the decision of the dominant oligarchies in the parties. The precarious internal democracy within the parties ensures the election of politicians linked to these oligarchies, whose dominance continues over time. In the case of women, their low levels of representation in the Chamber of Deputies is due, among other factors, to the limited funding of their candidacies and male resistance to women representatives being ratified according to party conventions.

Another critical factor in the weakening of Brazilian democracy refers to the asymmetries in relations between the Executive and the Legislative branches of government. The former concentrates many powers, such as the power to enact provisional measures ( medidas provisórias ) that immediately change the status quo, and the exclusive prerogative to propose both the federal budget and specific legislation; the parliament, for its part, has limited capacity to monitor, supervise, and control the Executive branch. This derives from the logic of coalitional presidentialism, despite the fact that the legitimacy of the Executive and Legislative branches is derived from different sources. The formation of governing majorities through the sharing of spheres of power in the Executive branch greatly limits the parliamentary function of oversight and control because it establishes an uncritical loyalty among the coalitions of congressmen. Thus this function is not performed in cases in which administrations fail to fulfill their duties or act against constitutional principles; obviously, this jeopardizes inter-institutional accountability.

This set of observations does not put into question the existence of democracy in the country, but rather problematizes its qualities. Many democratic advances in recent decades have been inspired by the principles established in the 1988 Constitution, which consolidated a set of fundamental social and political rights and concomitantly established the legal premises for the adoption of important universal social policies. In this sense, relevant economic changes adopted by social democratic parties and left-wing parties from the 1990s onwards improved the living conditions of poorer segments of the population, with the social issue acquiring great relevance in the initial years of the current century. Even without changing the structure of income distribution, this has positively affected inequality rates and the likelihood of social mobility for the lower echelons of society.

The adoption of these policies by progressive parties helps to explain the role of identity politics in the Brazilian case. Unlike the United States and some European countries, Brazil has not seen the same magnitude of mobilization in favor of policies sensitive to issues of identity as its peer countries. Nevertheless, feminist movements, the defense of the rights of Afro-descendants, and the LGBT community are very active and, even if they drive some fragmentation within the democratic forces, together with other social movements, they permanently pressure the political system to expand and consolidate both universal and specific rights. In this sense, they played an extremely important role before and during the 1987/88 Constituent Assembly, presenting proposals and pressuring constituents in favor of embracing very inclusive principles, which led to the adoption of significant universal public policies, as well as some specific ones. In general, this pressure challenged the parties across the political-ideological spectrum in the sense of the enlargement of citizenship, aiming at integrating the action from inside and outside these parties.

Examples that contextualize the singularities of the Brazilian case include: I) article 5 of the federal Constitution which establishes fundamental rights and advances the explicit equality of men and women; II) in the case of Afro-descendants, the law treats racism as a non-bailable crime, with no statute of limitations. Based on these principles, in 2010 the National Congress approved the Statute of Racial Equality, as proposed by a left-wing parliamentarian; III) subsequent years saw the establishment of a quota system of affirmative action guaranteeing access of Black or Afro-descendant peoples to Brazilian universities; IV) this affirmative action policy later extended this access to public education to indigenous and economically disadvantaged students and applied these quotas to governmental positions. Article 68 of the transitional provisions of the Constitution and article 215 of the main text are also important: the former ensured descendants of slaves enjoyed the right to ownership of the lands they and their peers occupied; the latter guaranteed the preservation and enjoyment of the material and immaterial cultural heritage of Black and indigenous peoples and popular sectors. Also based on article 5, the Supreme Court recently recognized citizens' right to same-sex unions.

In the case of women, other advances have occurred since the 2000s, with the creation of governmental agencies such as the specialized police stations to deal with abuse or ill-treatment and crimes of violence and conferences in charge of defining public policies, both initiatives led by progressive parties. To raise awareness and combat violence against women, in addition to the systems put in place to report threats and aggression, the Maria da Penha Law was sanctioned in 2006, the criminal classification of rape in 2009, and the definition of femicide in 2015, addressing different types of sexual crimes. These advances are reinforced by resources now provided through Brazil's Unified Health System to care for women victims of sexual, physical, or psychological violence; the Unified Health System reports more than 147,000 cases per year (i.e., 405 cases per day or one case every four minutes). Deaths of women resulting from male aggression between 1980 and 2013 rose more than 100%, from 2.3 to 4.8 per 100,000. In 2016, the justice system registered more than 1 million cases of domestic violence against women - that is, one (1) case for every 100 Brazilian women. Laws and procedures in solidarity and defense of women require the State to be effective in law enforcement and justice.

As for the political participation of women, the Brazilian legislation has ensured that, since 1997, each party must advance, in proportional elections, a minimum of 30% and a maximum of 70% of candidatures of each sex. In 2018, the Federal Supreme Court recognized the right of female candidates to access the resources of the Special Fund of Electoral Campaign Financing. At least 30% of these resources must be distributed to women candidates. This has facilitated an increase in participation and recognition of the demands of women as their voices are channeled and echoed through political parties sympathetic to women's movements. However, following the example of Afro-descendant and the LGBT movements, many feminist leaders are complaining that rates and means of inclusion remain insufficient and inequitable; they maintain that mobilization and political participation should occur independently of parties and directly pressure the structures of the State. Feminists argue that the presence of women in institutional politics is not enough to guarantee the incorporation of specific guidelines of female interest, such as the decriminalization of abortion, facing inequalities related to the sexual division of labor, unequal remuneration, etc. This reinforces the fragmentation of democratic forces without amounting to an alternative to the crisis of representative democracy, but it is a reaction to the ineffectiveness of democratic politics.

The virtuous cycle represented by post-authoritarian democratic achievements was interrupted in the second decade of the 21^st century with the leftist government's adoption of a set of economic policies that produced recession, loss of income, and soaring unemployment rates (exceeding 14 million unemployed workers). Moreover, the revelations of the extent and volume of corruption practices have completely discredited the leaders among the traditional elites. The extreme aggravation of criminality and public safety problems - in light of reports of more than 60,000 victims of violent crimes each year and of the Brazilian police forces recognized as one of the most violent in the world - Brazilian police are responsible for the deaths of more than 5,000 people per year, the majority of whom are young blacks between the ages of 16 and 26. These factors have greatly deteriorated the country's political environment into one of precarity.

It was in this troubled context that Jair Bolsonaro saw an increased opportunity to assert himself as a populist leader. Supported by men from the richest regions of the country, especially in the 45-64 age bracket and with high levels of education, he adopted a crude and coarse rhetoric and occupied the political vacuum left by the forces of the left and the moderate center. He presented himself as a champion of crime prevention and law enforcement, a radical fighter against corruption, and, like right-wing populist leaders from other countries, he explored the contrast between the supposed moral values of evangelical Christianity and the conservatism of the traditional Brazilian family on the one hand and the behaviors of the LGBT community and other minority groups considered as immoral on the other hand. In addition to defending the military dictatorship and torture, Bolsonaro embraced the use of social networks and fake news to publicize his criticisms of the equal rights of women, as well as the rights, recognized by National Congress and the Federal Supreme Court, of Afro-descendant and indigenous peoples to occupy territorial reserves and ancestral lands. Bolsonaro attacked his opponents with vulgar or offensive language and by calling for protests against the forces of political pluralism and challenging not only the evidence of global warming, but also the scientific data that account for the enormous growth of deforestation in the Amazon. Bolsonaro even adopted a stance of denial in relation to the Covid-19 pandemic, refusing to adopt the measures advocated by the World Health Organization and his own government's Ministry of Health, instead sponsoring policies supporting the unscientific thesis of 'flock contamination'. His government, in this context, proved to be a strategic actor in detriment of the country's democratic advances over the last 33 years.

In the face of the complex political implications of Bolsonaro's rise, an inevitable meta-theoretical line of inquiry has emerged in Brazil. The country is experiencing a wave of diffuse support for anti-democratic political behavior, which is contrary to the process of consolidating the democratic regime in the country. How accurate are the traditional instruments for collecting and processing empirical data related to this question? To put it differently, to what factors can we attribute the presence of authoritarian traits in Brazilian society? Do these traits reflect a qualitative change in the population's perception of the behavior of political leaders' in light of serious denunciations of corruption? Or, as an alternative hypothesis, are we witnessing the resilience of authoritarian cognitive traits that were not being captured by means of the usual opinion polls?

Survey instruments are an ideal tool to test the accuracy of measurements of complex issues like the degree of diffuse adherence to democratic regimes. We chose to approach the perception of corruption as a proxy for the degree of consolidation of the democratic regime for two main reasons, the first with methodological implications and the second with political significance. The present article aims to address this issue by introducing two survey experiments conducted by the authors in 2014 and 2018. The 2014 survey indicated that there is a substantive difference between responses of interviewees to explicit questions on corruption and the indirect experimental questions about it. Based on the first survey, the level of tolerance for corruption among the population seems to be much higher than what is revealed by the usual methodological instruments. At the same time, comparison with the 2018 experiment allowed the validation and extrapolation of the results found by Moisés and Nunes de Oliveira (2018) in the first case. Thus, the statistical procedures allowed us to observe more accurately the degrees of democratic adherence in the country at the present time.

This article is divided into five sections. In the first, we discuss how difficult it is to measure corruption as a social phenomenon. We argue that the multifactorial and multidimensional nature of corruption poses an additional challenge to studies in the area. How the specialized literature methodologically outlines this problem is the focus of our literature review. In the second section, we present the dynamics of corruption in Brazil and the population's perception of the phenomenon. Then, we present the experiments of the list, introduced in two national opinion surveys, to understand the degree of congruence between the direct questions about the importance of corruption and the measurement of the phenomenon by more sophisticated methodological instruments. Finally, in the last section, there is a gap between the answers to the direct questions and the results of the survey experiment, indicating a more pronounced degree of tolerance towards corruption in Brazil than opinion polls and other indicators suggest.

On the concept of corruption

The conceptual definition of corruption is related to multidimensional characteristics. The term describes a vast myriad of legal and illegal behaviors. The search for private benefits derived from interactions with public officials is one of the central points of different studies on the theme. Offenders, whether elected or appointed for bureaucratic functions, divert public resources to reap private benefits. Even when corruption is related to the illegal financing of electoral campaigns or political parties, the practices of corruption involved in the process generate a complex network that financially benefits the private agents involved. As a result, the practice of corruption distorts the republican dimension of democracy because it withdraws resources from public policies whose decisions are the result of spurious agreements ( Kunicová, 2006 ).

Corruption also compromises the perception of the comparative advantages of democracy, making common people believe that it is constitutive of politics as such, whether democratic or authoritarian. In addition, corruption also impacts the functioning of state bureaucracy, particularly in relation to the Integrity System. This is so because it disqualifies the specific mechanisms for controlling abuse of power by disabling accountability mechanisms (vertical and horizontal accountability). In sum, corruption affects the effectiveness of public policies, impacts the quality of democracy by compromising the rule of law and consequently compromises the legitimacy of the regime. This is the reason we take corruption as an ideal proxy for carrying out a methodological discussion on the determinants of democratic adherence in Brazil.

Most of the literature on the matter turns its attention to the relationship between corruption at the individual level, countries' economic development, and its interaction with institutions. However, there is still little research about the effects of corruption on the quality of the democratic regime and the effects of cognitive and cultural variables to explain the dynamics of corruption. One of the rare exceptions is Brazilian scholar Meneguello (2011) . The author aims to explain how the perceptions and individual judgments about corruption make up a meaningful construct of support for democracy. According to Meneguello, there is a degree of independence between perceptions about corruption and adherence to the democratic regime, indicating that corruption scandals have a marginal effect on adherence to democracy in Brazil. Following the same line of argument and searching for explanations about why corrupt politicians survive democratic elections, Pavão (2018) points out that issues related to corruption are subsumed under other electoral determinants of the vote. As voters believe that corrupt practices are constant among Brazilian politicians, apparently there are no rational reasons to punish politicians involved in acts of corruption. To a certain extent, the evidence presented by Pavão corroborates the findings of Rennó (2007) . According to the latter, the empirical data do not allow the association between corruption scandals and the exercise of vertical accounting by the Brazilian voter. In other words, voters' ability to monitor and punish the reported cases of corruption is obliterated by other voting constraints, such as the retrospective assessment of government's economic performance and sympathy for or rejection of the political party in power.

Although not entirely in disagreement with Meneguello (2011) , Rennó (2007) , and Pavão's (2018) respective interpretations, Moisés (2009) adds nuance to the understanding of the phenomenon of corruption by introducing the influence of political culture. According to the author, whatever the cause of voters' decisions, it is a consequence, alongside other determinants, of traits of the political culture. Thus, it implies that acts of corruption affect the public's perceptions of democracy in the country. For the author, social and political behavior is caused by a variety of factors - institutional, economic, etc. - but also by diverse features of the political culture, whose influence affects not only the behavior of public and private agents but also society as a whole as well as the functioning of institutions such as political parties. New knowledge on this subject is gained from an explanation of how voters come to evaluate corruption in their countries and how much these perceptions and convictions affect their support for governments, public institutions, and democracy. Therefore, political culture is seen not as the only causal determinant of political behavior, but as an intervening factor in the functioning of democracy due to the strong and persistent presence of the phenomenon of corruption in the country's public life.

This article proposes to advance the agenda introduced in previous works by Moisés (2013 , 2017 , 2018) and departs from the hypothesis that the perception of corruption is conditioned by the domain of political values; moreover, people's cognitive traits are influenced by the behavior of public agents and social actors. In other words, political culture is seen as influencing the contextual space where institutions are formed and how they respond to voters' expectations. In turn, public and individual agents are supposed to form their preferences and make their political decisions considering both the institutional design and the influence of cultural and political values. In this sense, data collected by Moisés in the first decade of this century have shown that the effects of social acceptance of corruption impact on the quality of democracy in important ways: reducing adhesion to democracy and, more importantly, encouraging the acceptance of authoritarian alternatives which might seek to replace it in times of crisis. Although the likelihood of the latter diminishes with time, the risk it poses may increase if the effects of acceptance of corruption combine with civic mistrust of democratic institutions. In another study, Moisés and Carneiro (2008) show the influence of this perception of the pervasiveness of corruption on Brazilian voters' preference for anti-institutional models of democracy. These works confirm the risk that authoritarian alternatives to democracy gain popular support when corruption is perceived as a generalized pattern of political life.

This article intends to integrate different analytical dimensions to explain this political behavior, such as socioeconomic and institutional approaches proper to modernization theory. According to this model, economic development is a determinant of the formation of individual attitudes and behavior. Important contributions to these studies address, on the one hand, the negative systemic consequences of corruption, such as clientelism, nepotism, and political illegitimacy, among others ( Banfield, 1958 ; Johnston, 1979; Etzioni-Halevy, 1985 ; Seligson, 2002 ). On the other hand, scholarly contributions point to corruption's supposed positive implications, such as political stability and the so-called 'greasing' of rigid bureaucratic structures ( Huntington, 1968 ; Waterbury, 1976 ; Becquart-Leclerq, 1989; Heidenheimer; Johnston; Levine, 1990). The literature also addresses the implications of the phenomenon for investment initiatives by the government and companies that affect development ( Mauro, 1998 ; Dreher; Herzfeld, 2005 ) . Regardless of their levels of development, a large number of democratic nations must deal with corruption, albeit to varying degrees.

In fact, in addition to corruption being part of the scenario of recently democratized nations ( Huntington, 1993 ), the existing evidence shows that different forms of "misuse of public resources" undoubtedly are part of the scenario of economically and politically developed nations ( Klitgaard, 1988 , 1998 ). Democratic political systems capable of ensuring effective political competition will facilitate greater public scrutiny of the actions of government officials and political leaders. As a result, their levels of corruption will be lower as the polyarchic structure will favor higher levels of transparency in public choice, providing constituencies stronger possibilities of control and oversight of the behaviors of politicians and publicly appointed officials.

In contrast, the institutionalist theoretical models start from the premise that the occurrence of cases of corruption is a phenomenon independent of the social or economic structure. The causes of the consolidation of anti-social and anti-republican practices are directly associated with the design and performance of control institutions. Starting from the axiom that rational agents aim to maximize their interests with the least possible effort, acts of corruption generally occur under conditions in which their benefits outweigh their costs. Regardless of the socioeconomic standard, institutional design must consider the costs of anti-social and anti-republican practices. The main hypotheses linked to this approach refer to the existence of institutional structures capable of guaranteeing civil and political rights, freedom of the press, and competitive and open electoral systems as a condition that accountability mechanisms can be effectively activated by inspection and control institutions and by voters to curb corruption and prevent those responsible from going unpunished.

Nevertheless, without underestimating the importance of the social dynamics and the relevance of institutional variables, the present article starts from the hypothesis that these components are related to spurious causality. Both are explained together with cognitive factors. According to the cognitive theoretical model, the presence of corruption is explained by the prevalence of certain socio-subjective characteristics of the population ( Inglehart; Welzel, 2005 ; Diamond, 1999 ; Moisés, 1995, 2008). To test the empirical validity of the cognitive hypotheses, two methodological strategies were adopted: the first part of the article is dedicated to a comparative analysis of the perception of corruption among Latin American countries. Using data from the Latinobarometro Consortium⁴, the comparative exercise discussed here aims to shed light on the dynamics of corruption in the region and the relationship between corruption dynamics, institutional features, and socioeconomic factors. This general structural approach will be followed up by a more structured discussion on the interactions between the phenomenon of corruption, institutional and socioeconomic factors, and cognitive variables.

Analysis of empirical results

The effect of the role of corruption, or the co-optation of public resources for private gain of any type, including advantages for government parties at the expense of the opposition, has been widely discussed in the specialized literature. While at the theoretical level the relationship between corruption, institutional design, and economic development is based on solid foundations, the discussion still needs to advance at the empirical level. This gap can be partly credited to the difficulty in measuring the phenomenon. As it refers to a phenomenon with expressive latitude in a conceptual, legal, and social sense, operationalization in a comparative perspective is not linear. Hamilton and Hammer (2018) present a remarkable methodological contribution by producing a comparative meta-analysis of different international corruption indicators. According to them, there are two main indicators of corruption: subjective and objective. The former indicator measures, through opinion surveys with specialists, elites, and/or the population, the experience and perception with corruption in a country. According to the authors, the main subjective indexes are Transparency International's Corruption Perception Index (CPI) and the World Bank's Control of Corruption Index (CCI). Based on the advantages and limits of each of these forms of measuring corruption, the meta-analysis carried out by Hamilton and Hammer (2018) points to a high degree of correlation, around 98%, between the CPI and the CCI.

Treisman (2007) , however, considers that subjective indices of corruption are not a direct measurement of the phenomenon. While such indices measure the perceptions of the dynamics of corruption, the author argues that such perception may not be based on direct knowledge of its dynamics, and it may produce a bias in the data, making comparability between countries imprecise. This is particularly problematic because researchers have used the data to make arguments about the relationships between social trust, inequality, and democracy, on the one hand, and corruption, on the other ( Treisman, 2007 ). However, Hamilton and Hammer (2018) responded that the use of multiple sources of information on corruption, as the possibility to capture the interaction of different actors involved in corruption, among other advantages, implies that subjective indicators are methodologically more suitable for conducting research on corruption when compared to data such as objective indicators. The contributions of Treisman (2007) , Hamilton and Hammer (2018) and Kaufmann, Kraay and Mastruzzi (2007) have led to subjective indicators being widely accepted in studies on the matter, yet even the authors caution about the degree of reliability of these indicators. According to Kaufman, Kraay and Mastruzzi (2007), objective corruption indices are not empirically operational since the legal definition of corruption differs between countries. Even the indicators that use instrumental variables are volatile due to the absence of a factor exclusively correlated to the phenomenon. According to the authors, the methodological procedures of subjective indicators are efficient.

In order to contribute to a better understanding of the phenomenon of corruption in Brazil, the authors conducted a series of public opinion surveys at the Center for Public Policy Research (NUPPs) of the University of São Paulo in partnership with the Center for Public Opinion Studies (CESOP) of the State University of Campinas over the last decades. These were nationwide opinion polls with a probabilistic sample⁵. In the opinion polls, two survey experiments were introduced for internal validation of data on the perception of corruption by the Brazilian population. The procedure provided an external data source for the validation of international corruption indicators and allowed for meta-analysis to be carried out on the degree of accuracy of the indicators of perceived corruption.

Latin America is a region of the world where the perception of the prevalence of corruption is strongly accentuated. In comparison with other regions, the average for countries in the region, based on Transparency International's Corruption Index, is 63.0⁶. The index ranges from 12.0, Denmark's index, considered a country with low corruption prevalence, to 87.0, Syria's index, considered the country with the highest prevalence of corruption. The perception of corruption prevalence in Latin America is lower than in the African region (mean 69.0), and Eurasia, mean (66.0).

Figure 1
: Corruption Perception Index (CPI) by region, in 2018

When one looks more closely at the dynamics of corruption in Latin America, two groups of countries are visible. For most countries in the region, Transparency International indicates the prevalence of corruption: of the 18 countries in the region in which the information from Transparency International is available, 15 countries scored greater than 60 points on Transparency International's Corruption Index. Only in Costa Rica, Chile, and Argentina is the perception of the prevalence of corruption low among the countries' elites (see Figure 2 ):

Figure 2
: Corruption Perception Index (CPI), Latina America, in 2018

At the same time, the perception of the prevalence of corruption measured based on both Transparency International's Index and the World Bank's Corruption Index shows that the Brazilian population is supposed to have a high rate of intolerance to the phenomenon. Latinobarometro (2018)⁷ has asked respondents if some corruption is tolerable if it helps implement solutions to national problems. In 2018, most of the Brazilian population considered acts of corruption unacceptable: while 30.5% of the sample strongly or partially agreed that corruption was a reasonable price to pay to solve national problems, 63% of Brazilians disagreed totally or partially of the leniency position in relation to corruption ( Figure 3 ). The corruption intolerance rate in Brazil in the period is almost 10 points higher than the regional average, placing the country among the least tolerant of corruption in Latin America, alongside Colombia, Costa Rica, and Uruguay.

Figure 3
: Degree of tolerance for corruption in order to solve national problems

Empirical data show that the Brazilian population perceives corruption as a persistent and resistant problem in the country and tends to present a more critical position as compared to other Latin American countries. For instance, Latin Americans were asked what their subjective position is when they are aware of corruption cases - that is, whether their attitude is one of prudence or of direct involvement in combating the crime of corruption. Brazil was divided on the issue: while 47.5% of the population revealed an attitude of fear in the face of the knowledge of acts of corruption, 50.1% disapproved of an attitude of apprehension ( Figure 4 ). El Salvador, Honduras, and Nicaragua were the countries in the region with the highest rates of intolerance or apprehension regarding knowledge of corruption.

Figure 4
: Expected attitude towards knowledge of acts of corruption

This division of Brazilian society vanishes when respondents are asked about the legal implications of the knowledge of acts of corruption. Most of the country's population considers that not reporting cases of knowledge of acts of corruption is a crime of complicity. More than 80% of the Brazilian population believes that cases of corruption should be reported. Most of the Latin America population considers, on the contrary, that denunciation or reporting is not a legal or political duty.

Figure 5
: Principles of the population on the legal need to report corruption cases

According to Latinobarometro, in 2018, the Brazilian population believed that the phenomenon of corruption was worsening. The prevailing view was that corruption crimes increased between 2017 and 2018. For 56.5% of the interviewees the phenomenon of corruption "increased a lot" in the previous 12-month period, and for 11.9%, corruption simply "increased". More than 67% of interviewees revealed their perception of an increase in corruption versus 8.2% of the population which reported a decline of cases. This perception of corruption increased in the country was among the highest rates in the region, only surpassed by Mexico, Venezuela, and Nicaragua.

The phenomenon of corruption is therefore quite complex in Brazil. Considering the main international subjective indicators, the country is strongly marked by the aforementioned phenomenon. The discovery in recent years of a systemic scheme of corruption in the country - unveiled through Operação Lava Jato (Operation Car Wash) which culminated in the impeachment of President Dilma Rousseff, the return of 2.3 billion dollars to the nation's coffers, the imprisonment of 293 people (among them, a former President of the Republic and the head of the Chamber of Federal Deputies) - was not enough to change the social scenario of corruption in the country, but it had a significant and impressive impact on the social perception of the phenomenon. In this sense, Latinobarometro data show that, compared to their peers in neighboring countries, Brazilians are intolerant towards corruption, aware of its reality, and willing to legally intervene to eliminate it, as demonstrated in the 2018 elections. In addition to electing an outsider of the political mainstream, President Bolsonaro, voters also promoted strong reforms in the Chamber of Deputies, with the non-reelection of approximately 52% of the deputies. The question, then, is how should we interpret all this evidence. Is it the resilience of the phenomenon, with the return of corruption as a gravitational center of the Brazilian population's cognitive traits? Or, conversely, is the persistence of corruption in the country associated with the main institutional variables, among other factors?

To deal with those questions, we ran two opinion polls designed to verify the population's assessment of the seriousness of corruption in the country, its electoral impact, the popular adherence to the democratic regime and the population's assessment and support for the main institutions responsible for fighting corruption. In addition to the traditional variables, both the 2014 and the 2018 surveys introduced an experiment that allows for an in-depth investigation into the sublinear dimensions that make up individuals' cognitive assessments of corruption.

Before analyzing the results of the two experiments, we shall explore the disentanglement of the main explanations for the tolerance of corruption. The variable chosen is the perception of the gravity of corruption in the country. Figure 6 shows the distribution of this perceived seriousness of the phenomenon. In 2014, 80% of the population in Brazil considered corruption a "very serious problem", 18% a "serious problem", and 3% "non-serious". In 2018, the percentage of the population who considered it very serious rose to 87%, while the percentage that considered it a serious problem dropped to 10%, and those who did not view it as a serious problem amounted to 4%. Empirical data from these two opinion polls align with and support the results produced by the World Bank's Corruption Index as well as Transparency International and the Latinobarometro. Few Brazilians do not consider corruption to be a serious issue in the country.

Figure 6
: Perception of the seriousness of corruption in Brazil

The effect of social structure on corruption occupies an important space in the literature about the subject. One concern is related to gender. Improving women's rights can reduce corruption ( Lambsdorff, 2006 ). The international literature has pointed out that the presence of women in the main political positions is negatively associated with the prevalence of corruption ( Swamy et al., 2001) . Therefore, at the microsocial level, women are expected to be more concerned about corruption than men. Data presented in Figure 7 describe the degree of concern with the topic among men and women in Brazil. There is a small gender difference regarding the perception of the seriousness of corruption in the country. While more than 81% of the male population believed that corruption was very serious in the country in 2014, 78% of women concurred that same year; however, four years later, the level of perception of seriousness of the phenomenon by women increased almost 10 percentage points, and the male sample reached nearly the same level as females in 2018, making the perception of seriousness more homogeneous between genders.

Figure 7
: Perception of the seriousness of corruption in Brazil, by gender

Another important feature of corruption is the role played by the educational structure. When the discussion refers to the comparison between countries, the data demonstrate a negative correlation between education levels and the occurrence of corruption. As Treisman (2006) points out, the high degree of correlation between the economic variables and education levels is well known. But this evidence makes it difficult to measure the independent effect of formal education on tolerance to corruption. Therefore, the direction of the correlation at the microsocial level is not clear. Research on antisocial behavior indicates that there is a tendency toward specialization among career criminals. Economically motivated crimes such as occasional petty theft, kidnapping, illegal trade, and trafficking in illicit goods are committed by individuals with low education levels. The most complex economically motivated crimes, such as defrauding the financial system, international trafficking of illicit goods, and corruption are committed by more educated individuals. At the core of the explanation for the prevalence of crime, among more educated individuals is the degree of social acceptance of criminal behavior ( Braithwaite, 1989 ). The data presented in Figure 8 indicate that individuals with intermediate and higher education levels were most concerned about the prevalence of corruption in the country in 2014. In the 2018 survey, the greatest concern over the phenomenon was expressed by respondents with intermediate education levels.

Figure 8
: Perception of the seriousness of corruption in Brazil by education level

Ideological and electoral debates in the recent period in Brazil have been strongly shaped by issues related to corruption. As already highlighted, corruption scandals had a strong impact on the political system and the 2018 election. Latinobarometro data revealed that in that year, when asked about the presence of acts of corruption in the executive branch, approximately 27% of Brazilians believed that none or only some presidents were involved in acts of corruption. Conversely, 35% believed that all presidents or all recent executive administrators in the country were involved with corruption and the identification rate of the phenomenon in the Executive branch was almost ten percentage points higher than the regional average, placing the country among those Latin American nations with the highest rates, alongside Venezuela, Paraguay, Mexico, and Nicaragua.

Recent presidential elections in the country were decided between a center-left candidate from the Workers' Party and a moderate social democrat candidate in 2014 and between a right-wing and a left-wing candidate in 2018. This makes apparent the impact of ideological and electoral choices on perceptions of corruption. In the 2014 election, those center-left candidate Dilma Rousseff's constituents in the Workers' Party were most concerned with corruption, but in the 2018 election, it was the right-wing candidate Jair Bolsonaro's constituents who were most concerned, demonstrating the volatility of preoccupation or concern with the theme in recent years.

Finally, one of the central focuses of the discussion about the causes and consequences of corruption is the relationship of the phenomenon with the functioning of the democratic regime. A number of articles have explored the effect of democratic institutions on perceived corruption. Some have argued that the effects of democracy are gradual and only a long or sustained experience with popular governance is needed to diminish the impact of corruption. Likewise, others have suggested that the relationship is non-linear: democratization can increase corruption in the short term, even if corruption is reduced as democracy deepens. Finally, scholars have examined the effects of particular institutions within democracies - the type of electoral rules, whether the regime is parliamentary or presidential, and the degree of freedom of the press - as well as the effect of decentralized government ( Treisman, 2007 ). On the other hand, the effect of perceived corruption on the degree of adherence to the democratic regime remains a little explored theme in the scholarly production on the subject.

Latinobarometro data indicated that approximately 35% of the population of the region had an antidemocratic sentiment in 2018. In Brazil, this rate was 44% - that is, 9% greater than the average for the region. The impact of diffuse adherence to the democratic regime on the tolerance for corruption goes in the opposite direction of what is theoretically predicted. In the region, among individuals who consider that democracy is always the best political regime, 45.5% have some degree of tolerance for corruption. Among individuals with authoritarian traits, tolerance for corruption drops to 34.8%. In Brazil, the impact of adherence to the democratic regime on tolerance for corruption is smaller. Between individuals with strong adherence to democracy, 9% did not consider corruption very serious. Among individuals with authoritarian traits, 22% did not consider corruption very serious, according to data from Latinobarometro (2018). NUPPs (2014) and CESOP (2018) survey data on the Brazilian case point to a different scenario. In the two years surveyed, individuals with authoritarian traits had a lower perception of the seriousness of corruption than individuals with clear democratic inclinations (see Figure 9 ):

Figure 9
: Perception of the seriousness of corruption in Brazil, by adherence to the democratic regime

Survey experiment

Data on the perception of the seriousness of corruption - revealed by international indices and Latinobarometro data - indicate a complex relationship between the tolerance for the phenomenon, individual socioeconomic variables, political behavior and diffuse adherence to the democratic regime. Part of the inaccuracies found can be explained by the disassociation between common preferences revealed in the surveys and the cognitive values of individuals. To provide new information on the question, the opinion polls carried out by NUPPs and CESOP introduced two statistical experiments.

The samples of the two surveys were divided into two groups balanced by their socioeconomic characteristics. In one (experimental group), a card was provided with 4 statements and respondents were asked to indicate with how many of them they agreed. The statements were:

1. In Brazil, rich or poor, black or white, male or female, all people are represented equally in politics";

2. Media coverage of political affairs is always fair and rarely sensationalist";

3. "The large number of political parties in Brazil makes Brazilian politics more complicated";

4. Despite being undesirable, it is understandable that politicians sometimes steal when they do politics in Brazil".

Evidently, the first three sentences are not related to perceptions of corruption. The main idea of the experiment was to make the interviewees reveal their view of the phenomenon without rational mediation. This is based on the presumption that it is easier for common people to reveal their true preferences when they are not stimulated to rationalize, verbalize, or formally register their preferences. On the other hand, the fourth statement - which indirectly refers to corruption - seeks to capture the degree of tolerance with the phenomenon. The design of the experiment proceeded with the construction of the above-mentioned control groups. In the surveys, one card - for the control group - had only the first three sentences versus the experimental group which saw the fourth statement on corruption. The difference between the mean of the experimental group (card with four sentences) and the control group (card with three sentences) should demonstrate the degree of acceptance of corruption not usually revealed on non-experimental surveys. The test of mean difference indicates whether the means of the experimental group and of the control group are statistically different from zero. In other words, whether the fourth sentence of the experimental group would be selected by an expressive number of respondents.

As can be seen in Tables 1 and 2, the tests of mean difference allowed the adjustment of the hypothesis to the empirical reality. In other words, the mean agreement with the statements between the experimental group and the control group is statistically different from zero. This demonstrated that unconsciously part of the individuals in the experimental group tends to accept corruption. In fact, in 2014 the mean of agreement with the sentences was 1.84 for the 875 respondents in the experimental group. The confidence interval of the true mean of the experimental group was between 1.79 and 1.90. The mean for the control group was 1.63, with a confidence interval of 1.59 and 1.68. The difference between the means was 0.21, with a confidence interval of 0.13 and 0.28. The difference between the means indicates that there is a difference in means and that of the experimental group is higher than that of the control group, with a 95% degree of certainty.

Table 1
: Differences between means test for survey experiment, in 2014

Variable	N	Mean	SE	SD	IC
Experimental group	875	1.84	0.03	0.9	1.78	1.90
Control group	1000	1.63	0.02	0.79	1.59	1.68
Combined	1875	1.73	0.02	0.85	1.69	1.77
Difference		0.21	0.04		0.13	0.28
Student's T test= 5,3**

Source: NUPPs, 2014; ESEB, 2018.
• Note: **The mean difference is significant at the 0.1 level.

On the other hand, in 2018 the means of agreement with the sentences was 1.61 for the 1219 respondents in the experimental group. The confidence interval of the true mean of the experimental group was between 1.55 and 1.67. The mean for the control group was 1.41, with a confidence interval of 1.36 and 1.46. The difference between the means was 0.20, with a confidence interval of 0.14 and 0.24. The difference between the tests of means indicates that the means are different and that the mean of the experimental group is higher than the control group with a degree of certainty of 95%.

Table 2
: Differences between means test for survey experiment, in 2018

Variable	N	Mean	SE	SD	IC
Experimental group	1219	1.61	0.03	1.03	1.55	1.67
Control group	1064	1.41	0.02	0.83	1.36	1.46
Combined	2283	1.47	0.02	0.97
Difference		0.20	0.04		0.14	0.24
Student's T test = 5,3**

Source: NUPPs, 2014; ESEB, 2018.
• Note: **The mean difference is significant at the 0.1 level.

Given the statistical validity of the experiment, it was possible to divide individuals into three distinct groups: Corruption-tolerant individuals, who agreed with all four statements; corruption-critical individuals, who did not agree with any statement; and undefined individuals, who agreed with one, two, or three of the statements. Figure 10 shows the results of the reclassification of individuals in Brazil. Comparing the 2014 results to those of 2018, it is possible to notice a significant increase in corruption-critical individuals, from 2.8% of the sample to 11.2%, and in tolerant individuals, from 2.2% of the sample to 5.8%.

Figure 10
: Corruption tolerance profile, according to the results of the experiment

The impact of this empirical evidence is significant. It means that we must analyze data of corruption surveys with some caution. The degree of popular tolerance to corruption in Brazil is greater than what is captured by the usual opinion polls. Cognitive factors obliterate the perception of the phenomenon, even if individuals are aware of its impact. As shown in Figure 11 , both in 2014 and 2018 the percentage of individuals who considered corruption a very serious problem was 10% higher among critical citizens.

Figure 11
: Perception of the seriousness of corruption in Brazil, by corruption tolerance profile

The impact is also significant when the investigation refers to adherence to the democratic regime. Among critical citizens, the range of individuals with authoritarian traits runs between 13% and 12%, in 2014 and 2018. While for the corruption-tolerant citizens it runs between 17% and 21%. This evidence indicates that the degree of adherence to the democratic regime is related to some extent with tolerance for corruption. But the direction of the correlation is not the same found in the Latinobarometro data. According to this survey, individuals with authoritarian traits in Brazil were less tolerant of corruption than individuals strongly adherent to the democratic regime. However, when capturing the perception of individuals through the experiment, the relationship was inverted. Individuals with authoritarian traits were more likely to be lenient in relation to the phenomenon.

Figure 12
: Adherence to the democratic regime, by corruption tolerance profile

Summary

The scenario that emerges from the evidence gathered in this article is complex and challenging. On the one hand, subjective international corruption indices appear to be the most robust and methodologically adequate to research the phenomenon. On the other hand, at least for the Brazilian case, the usual indices and surveys seem not to be fully capturing individuals' perceptions of corruption. The presented statistical experiments have demonstrated that a significant part of the Brazilian population responds differently when asked directly about the occurrence of corruption and, alternatively, when investigated in a sublinear way. The impact of this evidence is clearly important for democracy.

As explored in Figure 6 , only 0.3% of Brazilians considered corruption not to be a problem in 2014. The data from the experiment show that the prevalence rate with corruption was 2.2%, or 7.3% higher than what is revealed by other surveys. In addition, 96.3% of the individuals classified by the experiment as tolerant of the phenomenon considered corruption to be a serious problem in the country in 2014. In 2018, the survey showed a disassociation between the direct question and the evidence revealed in the experiment, which was smaller, but still significant. Survey data in that year revealed that 1% did not consider corruption to be a serious problem. The data from the experiment revealed the prevalence of tolerance for corruption was 5.8% of the population, 52.6% higher than what appeared in the survey. As in 2014, about 90% of individuals who considered corruption serious in Brazil were classified as tolerant of corruption in the 2018 experiment.

From a theoretical point of view, the data analyzed here suggest a clear counterpoint to research that indicates that the perception of the phenomenon of corruption has a residual impact on the adherence to democracy and electoral behavior. The degree of perception about the seriousness of corruption, both measured by direct questioning and experimentally, is strongly correlated with the diffuse attitude of adherence to the democratic regime and with electoral behavior, and more than pointed out by Pavão (2018) , Meneguello (2011) and Rennó (2007) . Also, when capturing the perception of corruption sublinearly, the relationship is the opposite of what is revealed by data from Latinobarometro. The data analyzed here confirm our hypotheses about the association between tolerance with corruption and antidemocratic values and weakening of adherence to democracy ( Moisés, 2010 ).

AGRADECIMENTOS

This article presents results of the Thematic Projects "Brazil, 25 years of democracy: a critical survey, public policy, institutions, civil society and civic culture (1988-2013)" (FAPESP n° 2011/50771-8) and "The Quality of Brazilian Democracy: political representation, parties and the Integrity System" (FAPESP n° 2018/02738-0). It was presented at the 26th IPSA World Congress of Political Science held by the International Political Science Association (IPSA) on 10-15 July 2021. The authors are grateful for the IPSA commentators' suggestions and to the comments of two anonymous reviewers.

Bibliographic references

• Abranches, S. “Presidencialismo de coalizão: o dilema institucional brasileiro”. Revista Brasileira de Ciências Sociais, vol. 31, nº 1, p. 5-14, 1998.
• Banfield, E. The moral basis of a backward society. Illinois: The Free Press, 1958.
• Becquart-Leclercq, J. "Paradoxes of Political Corruption: A French View". In: Heidenheimer, A.; Johnston, M.; Levine, V. T. Political corruption: a handbook. Edition 2ª, New Brunswick: Transaction Publishers, 1990.
• Braithwaite, J. Crime, shame and reintegration . New York: Cambridge University Press, 1989.
• Dahl, R. Polyarchy: participation and opposition. New Haven and London: Yale University Press, 1972.
• Diamond, L. Developing democracy: toward consolidation. Maryland: Johns Hopkins University Press Books, 1999.
• Dreher, A.; Herzfeld, T. “The economic costs of corruption: a survey and new evidence”. Available at: < http://dx.doi.org/10.2139/ssrn.734184>. Accessed on: Aug. 15, 2023.
  » http://dx.doi.org/10.2139/ssrn.734184
• Etzioni-Halevy, E. Bureaucracy and democracy: a political dilemma. New York: Routledge & Kegan Paul Press, 1985.
• Hamilton, A.; Hammer, C. Can we measure the power of the grabbing hand? A comparative analysis of different indicators of corruption. OKR, 2018. Available at: < https://openknowledge.worldbank.org/handle/10986/29162>. Accessed on: May, 04, 2023.
  » https://openknowledge.worldbank.org/handle/10986/29162
• Heidenheimer, A.; Johnston, M.; Levine, V. T. Political corruption: a handbook. Edition 2ª, New Brunswick: Transaction Publishers, 1990.
• Huntington, S. Political order in changing societies. New Haven: Yale University Press, 1968.
• Huntington, S. The third wave: democratization in the late 20th century. Oklahoma: University of Oklahoma Press, 1993.
• Inglehart, R.; Welzel, C. Modernization, cultural change, and democracy: the human development sequence. Boston: Cambridge University Press, 2005.
• Johnston, M. “The political consequences of corruption: a reassessment”. Comparative Politics, vol. 18, nº 4, 1986.
• Kaufmann, D.; Kraay, A.; Mastruzzi, M. The worldwide governance indicators: methodology and analytical issues. OKR, 2007. Available at: < https://openknowledge.worldbank.org/handle/10986/3913>. Accessed on: May, 04, 2023.
  » https://openknowledge.worldbank.org/handle/10986/3913
• Klitgaard, R. Controlling corruption. Berkeley: University of California Press, 1988.
• Klitgaard, R. “International cooperation against corruption”. Finance & Development, vol. 35, nº 1, p. 3-6, 1998.
• Kunicová, J. Democratic institutions and corruption: incentives and constraints in politics. In: Rose-Ackerman, S. (ed.). International handbook on the economics of corruption. Cheltenham: Edward Elgar Publishing Limited, 2006.
• Laakso, M.; Taagepera, R. “Effective number of parties: a measure with application to West Europe”. Comparative Political Studies, vol. 12, nº 1, p. 3-27, 1979.
• Lambsdorff, J. Causes and consequences of corruption: what do we know from a cross-section of countries?. In: Rose-Ackerman, S. (ed.). International handbook on the economics of corruption. Cheltenham: Edward Elgar Publishing Limited, 2006.
• Lijphart, A. Democracies: patterns of majoritarian and consensus government in twenty-one countries. New Haven: Yale University, 1984.
• Mauro, P. “Corruption and the composition of government expenditure”. Journal of Public Economics, vol. 69, p. 263-279, 1998.
• Meneguello, R. O lugar da corrupção no mapa de referência dos brasileiros: aspectos da relação entre corrupção e democracia. In: Avritzer, L.; Filgueiras. F. Corrupção e Sistema Político no Brasil. São Paulo: Record/Civilização Brasileira, 2011.
• Moisés, J. A. “Political corruption and democracy in contemporary Brazil”. Revista Latino-Americana de Opinión Pública, vol. 1, nº. 0. Buenos Aires: Dedalus Editores, p. 103-124, 2010.
• Moisés, J. A. Corrupção política e democracia no Brasil contemporâneo. In: Moisés, J. A.; Meneguello, R. (eds.). A desconfiança política e seus impactos na qualidade da democracia. São Paulo: Edusp, p. 201-236, 2013.
• Moisés, J. A. Corrupção e democracia: os efeitos de uma sobre a outra (online). Instituto Millenium, 28 ago. 2017. Available at: < https://institutomillenium.org.br/corrupo-democracia-os-efeitos-de-uma-sobre-outra/>. Accessed on: Aug. 15, 2023.
  » https://institutomillenium.org.br/corrupo-democracia-os-efeitos-de-uma-sobre-outra/
• Moisés, J. A.; Nunes de Oliveira, E. “A corrupção na visão da sociedade brasileira”. Trabalho apresentado no XI Encontro da Associação Brasileira de Ciência Política. Universidade Federal do Paraná, Curitiba, 2018.
• Pavão, N. “Corruption as the only option: the limits to electoral accountability”. The Journal of Politics University of Chicago Press, vol. 80, nº 3, p. 996-1010, 2018.
• Rennó, L. “Escândalos e voto: as eleições presidenciais brasileiras de 2006”. Opinião Pública, Campinas, vol. 13, nº 2, p. 260-282, 2007.
• Seligson, M. "The impact of corruption on regime legitimacy: a comparative study of four Latin American countries". Journal of Politics, vol. 64, nº 2, 2002.
• Swamy, A., et al. “Gender and corruption”. Journal of Development Economics, vol. 64, n° 1, p. 25-55, 2001.
• Treisman, D. “What have we learned about the causes of corruption from ten years of cross-national empirical research?”. Annual Review of Political Science, vol. 10, p. 211-244, 2007.
• Waterbury, J. “Corruption, political stability and development: comparative evidence from Egypt and Morocco”. Government and Opposition, vol. 11, nº 4, 1976.

• 4
  Available at: < https://www.latinobarometro.org/latContents.jsp >. Accessed on: July 24, 2020.
• 5
  The 2014 opinion poll was conducted by NUPPs within the scope of the Thematic Project "Brazil, 25 years of democracy: a critical survey, public policy, institutions, civil society and civic culture (1988-2013)", funded by FAPESP (nº 2011/50771-8). The survey sample is 1,875 respondents, its scope is national and it was fielded in the period following the second round of the 2014 presidential election. The 2018 opinion poll is the Brazilian Electoral Survey ( ESEB) , undertaken by CESOP under the coordination of Professor Rachel Meneguello and funded by FAPESP (nº 2018/02738-0, Thematic Project "The Quality of Brazilian Democracy: political representation, parties and the Integrity System"). The survey sample was 2,506, its coverage is nationwide and was fielded in the period following the second round of the 2018 presidential election. ESEB is available at CESOP Archives: < https://www.cesop.unicamp.br/eng/banco_de_dados/v/4538 >.
• 6
  Since 1995, Transparency International has published the annual Corruption Perceptions Index (CPI), which ranks countries according to the degree to which corruption is perceived to exist among public officials and politicians. CPI aggregates data from different sources that provide perceptions of the private sector and of experts on the level of corruption in the public sector. The index assigns score s on a scale between 0 and 100. The higher the score, the greater the country's perception of integrity. For a discussion of the methodological construction of the CPI, see Hamilton and Hammer (2018) .
• 7
  Available at: < https://www.latinobarometro.org/latContents.jsp >. Accessed on: July 24, 2020.

Publication Dates

• Publication in this collection
  23 Oct 2023
• Date of issue
  May-Aug 2023

History

• Received
  28 June 2022
• Accepted
  16 Mar 2023