Polarization and ideology: exploring the contextual nature of democratic commitment

Polarização e ideologia: explorando a natureza contextual do compromisso democrático

ABSTRACT

Introduction:

Political polarization stands as a pivotal topic in contemporary Political Science. Existing literature indicates that ideological extremism, prevalent in polarized environments, fosters the erosion of democratic norms by political elites. This phenomenon has primarily been explored through the examination of isolated cases or limited comparisons among a handful of highly polarized countries. This article investigates the role of “guardians of democracy”, questioning whether specific ideological groups maintain their commitment to democratic principles even amidst profound polarization.

Materials and methods:

Our analysis draws on public opinion data from 57 countries, encompassing over 77,000 respondents, sourced from the most recent wave of the World Values Survey (2017-2020). We explore the interaction between contextual and individual factors, investigating the impact of national-level polarization on individuals with varying ideological orientations and how the government's dominant ideology shapes individual responses to political polarization.

Results:

There is a negative relationship between polarization and democratic legitimacy. However, the impact varies depending on the ideology of the voters and their alignment with the ideology of the ruling party.

Discussion:

Our analysis suggests that the interaction between contextual factors (polarization and government ideology) and individual factors (voter ideology) creates scenarios that foster different profiles of guardians of democracy. However, none of these scenarios involved right-wing individuals presenting themselves as such.

Keywords
ideology; political polarization; support for democracy; democratic backsliding; multilevel multivariate models

RESUMO

Introdução:

A polarização política é um tema crucial na Ciência Política contemporânea. A literatura sugere que o extremismo ideológico, prevalente em ambientes polarizados, propicia a erosão das normas democráticas pelas elites políticas. Esse fenômeno tem sido explorado predominantemente através do exame de casos singulares ou comparações limitadas entre poucos países, marcados por significativa polarização. O artigo investiga o papel dos “guardiões da democracia”, questionando se grupos ideológicos específicos mantêm seu compromisso com os princípios democráticos mesmo sob intensa polarização.

Materiais e métodos:

A análise inclui dados de opinião pública de 57 países abrangendo mais de 77.000 respondentes. Os dados são da última onda do World Values Survey (2017-2020). O estudo se debruça sobre a interação entre fatores contextuais e individuais, examinando como a polarização em nível nacional impacta pessoas com diferentes orientações ideológicas e como a ideologia dominante no governo influencia as respostas dos indivíduos à polarização política.

Resultados:

Há relação negativa entre polarização e legitimidade democrática, mas seu impacto depende da ideologia dos eleitores e de sua interação com a ideologia do partido governante.

Discussão:

Concluímos que a dinâmica entre fatores contextuais (polarização e ideologia do governo) e fatores individuais (ideologia do eleitor) gera cenários que estimulam perfis distintos de guardiões da democracia, mas, em nenhum deles, indivíduos de direita se apresentaram como tal.

Palavras-chave
ideologia; polarização política; apoio à; democracia; retrocesso democrático; modelos multivariados multiníveis

I. Introduction¹

Political polarization has been identified as one of the causes of the democracy erosion processes (McCoy et al., 2018; Levitsky & Ziblatt, 2018; Przeworski, 2021; Haggard & Kaufmann, 2021). One strand of this debate is the literature on the negative effects of polarization on democratic legitimacy. These studies present evidence that polarized contexts create favorable conditions for extremist voters to maintain support for their candidates even when they violate democratic norms. It would formerly be up to moderate voters to have the role of “guardians” of democracy (Svolik, 2019, 2020; Graham & Svolik, 2020; Gandhi & Ong, 2019; Torcal & Magalhães, 2022).

Concerned with the decline of democratic legitimacy, these studies have generally been developed, through case studies carried out in countries with extreme polarization (e.g., the USA, Hungary, Indonesia and Venezuela) (Svolik, 2020; Graham & Svolik, 2020; Abrams & Pope, 2022; Davis & Wood, 2021; Gandhi & Ong, 2019) or by comparing a few countries (Torcal & Magalhães, 2022). This limitation in the type and number of cases, due to a selection bias, does not allow us to conclude that the decline in democratic legitimacy is a more prominent phenomenon in polarized societies.

By comparing more or less polarized countries based on data from the World Values Survey (2017-2020), we seek to contribute to overcoming this limitation by testing, in a broader and more systematic way, the validity of the thesis of polarization has negative effects on support for democracy. Additionally, from this broader and more diverse empirical base, we tested the thesis that, in polarized contexts, moderates are the guardians of democracy, while ideologically oriented voters are less loyal to democracy.

In addition to comparing more or less polarized contexts, this study also analyzed the effect of polarization on voters’ the democratic commitment, depending on the possible combinations between their ideological orientation and that of the party in power. In other words, we understand that defining aspect of the impact of the polarized environment on support for democracy is the ideology of the party in power, with possible heterogeneous effects between aligned and opposition voters (Anderson et al., 2005). We argue that the effect of polarization on the electorate´s democratic commitment varies according to the combination between the ideology of the government and that of the voter.

Our general expectation, based on the literature on democratic erosion, is that there is a negative relationship between polarization and democratic legitimacy. Furthermore, we expect that the effect of polarization depends on the ideology of the individual. Furthermore, we expect the effect of polarization to depend on the individual's ideology, as postulated by the “rigidity of extremes” and “rigidity of the right” theories (Greenberg & Jonas, 2003; de Leeuw et al., 2021). Due to personality traits, such as “intolerance, overconfidence, distress, dogmatism, and simplicity” extremist voters, especially those on the right, are more likely to renounce their commitment to democracy in polarized environments. Finally, we expect the negative effect of polarization on democratic commitment to vary depending on the convergence/divergence between the ideology of the voter and that of the party in power (Anderson et al., 2005).

The argument underlying this set of expectations is that polarization does not produce an actor who is the universal “guardian” of democracy (Svolik, 2020; Graham & Svolik, 2019). We suggest that a theory that combines aspects of theses based on the ideology of the voter, the moderating role of the center, and the reaction to the threats presented by the political opponent contributes to a more comprehensive understanding of the phenomenon.

The results indicated that each of these theories contributes to explaining the dynamics surrounding polarization and democratic legitimacy. Regardless of the context, individual traits seem to configure the most static attitudinal pattern of individuals situated on the right wing of the ideological scale. They almost always show less commitment to democracy, with the opposite occurring with left-wing individuals. In turn, the attitudes of this second ideological group depend more on the context, and when there is a polarized right-wing government, its loyalty to democracy grows, thus following the expectations of the theory of democratic reaction. The center only offers itself as the guardian of democracy in polarized left-wing governments. Therefore, the evidence we had did not allow us to conclude that there is a universal guardian of democracy nor a unified theory capable of explaining on its own, the complexity of the relationship between political polarization and democratic legitimacy.

II. Polarization, context, and ideology

Political polarization has been one of the central topics of recent debate in Political Science. While its types and causes are well documented (Iyengar et al., 2012; Iyengar et al., 2019; Mason, 2018; Fiorina et al., 2010; Webster & Abramowitz, 2017), its consequences are less obvious and continue to be the subject of new studies.

On the one hand, the recent debate on democratic backsliding has pointed to polarization as one of its leading causes. There are several reasons for this. Based on a relational definition of polarization, McCoy et al. (2018) perceived in it destructive potentials for democracy that may occur in the form of collapse caused by old or new elites or through violation of democratic norms. Similarly, Levitsky & Ziblatt (2018, p. 115) indicated that “polarization can destroy democratic norms” by undermining the foundations of political “tolerance” required by democracy. Also, according to the authors, a certain level of polarization is healthy and even necessary for the quality of the regime; however, its excess renders “politicians increasingly tempted to abandon the institutional reservation and try to win at any cost”. At this moment, “democracy is in trouble” (Levitsky & Ziblatt, 2018, p. 115). In turn, Przeworski (2021, p. 33) pointed out that polarization makes it impossible to “find measures acceptable for all political forces”, thus creating the conditions for democratic erosion. Haggard & Kauffmann (2021) also gave centrality to polarization in their explanatory model of democratic backsliding. Among the various reasons for this is the fact that it “reduces support for centrist political forces and thereby opens the door to autocratic electoral appeals” (Haggard & Kauffmann, 2021, p. 6).

At the electorate level, although some studies believe there is no relationship between polarization and democratic support (Albertus & Grosman, 2021; Grossman et al., 2022), the most widely accepted argument in the literature is that polarized environments cause voters to develop more intense political preferences (partisan or ideological) and accept the violation of democratic principles (Svolik, 2020; Svolik, 2019; Torcal & Magalhães, 2022; Fossati et al., 2022; Gandhi & Ong, 2019; Carey et al., 2022).

The set of papers developed by Milan Svolik & Mathew Graham (Graham & Svolik, 2020; Svolik, 2019) established the core of the argument and is the main reference point of this literature. Through experiments, the treatments of which are candidates who violate democratic principles, the authors showed that, in polarized environments, ideological or partisan extremist voters, who become the majority, accept to renounce democracy to prevent opposing groups from coming to power.

The empirical studies developed to reach these conclusions analyzed several countries, such as Venezuela, United States and Turkey (Graham & Svolik, 2020; Svolik, 2019). In all of them, voters with greater partisan or ideological intensity accept renouncing democratic principles in order to defeat the opposing candidate/party. And this is true even for voters who responded positively to the classic questions about preference/support for democracy. In the words of Svolik (2019, p. 12), “voters are reluctant to punish politicians who disrespect democratic principles when doing so requires abandoning their party or favorite policies”. A second finding is that voters in the ideological center and/or with weak party preferences are the least willing to renounce these principles. According to Svolik, centrists would play the role of guardians of democracy, which is a scarce resource in polarized societies.

In summary, the main argument is that political polarization mainly affects voters situated at the ideological or partisan extremes (who become the majority in such contexts), pushing them into a political game in which democratic rules are relativized to prevent opposing groups from coming to power.

The major problem with these papers is that they only investigated polarized contexts (Svolik, 2020; Graham & Svolik, 2020; Fossati et al., 2022; Clayton et al., 2021; Gandhi & Ong, 2019). Therefore, their results do not allow for comparison with non-polarized countries. Consequently, it is not known whether polarization is the factor responsible for the decline in support for democracy in the electorate.

Torcal & Magalhães (2022) tried to solve this problem by expanding the number of cases and considering a more significant variation in the degree of political polarization. The analysis focused on the relationship between ideological extremism, the perception of polarization, and support for democracy. The authors concluded that the further individuals stray from the average ideological center of their countries, the less support they have for democracy. A second finding of the study was that support for democracy is higher among voters who perceive polarization as moderate.

Although Torcal & Magalhães (2022) provide a broader perspective than previous studies by significantly expanding the number of cases with the inclusion of eleven European countries in the analysis, the study suffers from similar problems as the literature mentioned above, since the context does not vary in their analysis, only the individual perception of polarization. To cover this limitation, in this article, we have carried out an analysis with a large number of cases and with considerable political-institutional, economic, and cultural variation. In this way, we can confirm (or not) whether polarization actually reduces the democratic commitment of citizens.

We then tested whether the ideology of the voter and that of the government condition the relationship between polarization and democratic commitment. We considered the validity of a set of theories, including the approach centered on individual attributes and those that highlight the political context. The classic thesis of the first approach is that of the “rigidity-of-the-right model” (de Leeuw et al., 2021), which postulates that individuals on the right wing of the political spectrum would have some specific personality traits, such as rigidity of thinking, already identified in the seminal study by Adorno et al. (1950) and confirmed by later studies (Eysench, 1954; McClosky, 1958; Stenner, 2005). Such an internal structure would be associated with less commitment to democratic forms of government (Jost, 2017). More accepted, however, has been the “rigidity-of-the-extremes model” (idem), which, in different versions, identifies, in ideological or partisan extremists, the bases of negative attitudes toward democracy (Fossati et al., 2022; Clayton et al., 2020; Gandhi & Ong, 2019; Torcal & Magalhães, 2022). The reason why extremists would be closer to authoritarianism is that they have certain psychological traits, such as “intolerance, overconfidence, distress, dogmatism, and simplicity” (de Leeuw et al., 2021, p. 416), which would be incompatible with democracy.

In turn, the contextualist approach, present in the studies reviewed earlier, emphasizes the political environment in which the actors form their opinions and make their decisions. Thus, for Svolik (2020), Graham & Svolik (2020), Fossati et al. (2022), and Torcal & Magalhães (2022), especially in polarized environments, extremists are primarily responsible for putting democracy at risk, while centrists and the moderate context appear as the guardians of democracy.

Our study also explored an aspect of the context that had not received attention from studies on polarization and support for democracy: the combination between the ideology of the voter and that of the party in government. Previous studies on the attitudinal effects of political polarization only considered the ideology of the voter (Torcal & Magalhães, 2022) without associating it with that of the governing party. This precluded testing the hypothesis that the combination of voter and government ideology explains, at least in part, democratic commitment, especially given the reaction of government opponents in defense of democracy.

III. Methodology

As our main objective was to study the effects of polarization on different individual ideological profiles in distinct national political contexts, we combined individual and national level data to conduct multivariate multilevel models.

The latest wave of the World Values Survey, conducted from 2017 to 2020, provided information at the micro level, totalizing 77,370 cases useful for the analyses, distributed across 57 national units (see the complete list in the Appendix, Box A)². Our dependent variable measures at this individual level the intensity of Democratic Commitment and was constructed using the attitudes of the respondents relative to two hypothetical political settings: a system governed by a strong leader (variable E114) and a military regime (variable E116) (the complete statements of these variables are shown in Box B of the Appendix). Since the possible answers are organized into evaluative scales that start from “very good” and end in “very bad”, we recoded the positive evaluations as “0” and the negative evaluations as “1”, i.e., we excluded from the “democratic” condition all respondents who evaluated these two settings positively (no matter if as “good” or “very good”). Those who expressed negative attitudes on both measures were obviously classified as “democratic” and received code “1”. The result of this operation was 52.15% of respondents committed to democracy.

Before providing the descriptions of the other predictors, it is important to recognize the limits of the dependent variable compared to the proposal by Graham & Svolik (2020) and Svolik (2020). In these papers, the authors developed experiments based on the manipulation of candidacies with platforms that attack fundamental democratic principles, thereby avoiding the cons involved in operationalizations based on classic questions about the individuals ‘normative adherence to democracy as a form of government, which is strongly contaminated by the socially desirable (Fuks et al., 2019). Given the lack of questions about democratic principles in the surveys currently applied to representative samples from a large number of countries, we opted for the operationalization described before to avoid the traditional questions that use the term “democracy” and, at the same time, enable a test that takes into account the contextual variation that does not exist in the experimental papers mentioned.

There is also a justification for using a binary measure of democratic commitment. Studies such as Wuttke et al. (2022, 2023) and Claassen et al. (2021) work with variables similar to ours, but operationalized in terms of scales. Our theoretical understanding of this measure is anchored in the idea that commitment to democracy means rejecting alternative forms of political organization other than democracy under all circumstances. In other words, it is a more demanding measure and is not intended to be comparable to those used in the two studies above.

The percentages by country (Table A in the Appendix) indicate a considerably varied picture, with 27 countries not reaching 50% of committed respondents. Only Chile escapes this group among Latin American countries yet registering a modest 52.8%. Brazil, Argentina, Bolivia, Colombia, Mexico, and Peru ranged from 20% to 40%. European countries such as Portugal, Romania, and Georgia are also part of this group, as well as several other African and Asian nations. Among the countries with the largest contingent of respondents committed to democracy, all with more than 80%, are Denmark, Sweden, Estonia, Austria, Finland, Norway, Iceland, and Greece.

These data illustrate how problematic it is to use the classic measures of support for democracy, as was evident in the case of United States, which achieved more than 80% of positive evaluations in the WVS question on “a democratic government” but registered 60% of committed respondents in the measure we proposed. The overestimation of support for democracy is even more significant in the case of Brazil, which registered 70% of positive responses with the evaluation measure using the socially desired word “democracy”, while only 27% of respondents were “committed” to the regime.

Among the individual predictors, we mobilized some demographic measures: age, (coded in four ranges <20, 20-40, 41-60, and >60), education (binary with 1 = “higher” education), income (10-point scale), sex (binary with 1 = male) and religion (catholic = 1, protestant/evangelical = 2, and others = 3). Noticeably, because of our research interests, we also used a measure of ideological positioning from the scores of respondents to the question “When it comes to politics, people talk about ‘left’ and ‘right’. In general, where would you place your view on this scale?”. The original ten-point scale was changed so that the extremes of 1 and 10 represented left- and right-wing extremists, respectively. The left and right wings, in turn, corresponded to the combination of points 2-4 and 7-9. Finally, the center gathered the respondents located in points 5 and 6. With this operationalization, we intended to test both hypotheses of the individual-level model emphasizing extremism (Fossati et al., 2022; Clayton et al., 2021; Gandhi & Ong, 2020; Torcal & Magalhães, 2022) and those inspired by the model that emphasizes the rigidity of the right wing (de Leeuw et al., 2021).

The distribution of commitment among the ideological groups was as follows (Graph 1), and, considering the contingent of respondents resulting from the combination of representative samples from the 57 countries, it indicates important differences among the ideological groups (the percentage per country may be found in Table A of the Appendix). There was a balance between committed and uncommitted respondents at the center and the extreme left, while democrats predominated in the “moderate” left and right wings. In turn, the extreme right registered a majority of uncommitted respondents. These simple distributions, decontextualized and without any control by other factors, provide clues for a more in-depth analysis, as they question the models that attribute less democratic positions to the extremes without distinguishing the ideological matrix and seem to corroborate the hypotheses that emphasize the peculiarities of the right wing in its extreme version.

Graph 1
Distribution of Democratic Commitment by Ideological Group (%)

At the macro level, we used the variable Political Polarization from the Varieties of Democracy (V Dem) project, which measures how much political differences affect social relations beyond the political arena on a five-point scale. According to this initiative, highly polarized societies are those in which supporters of one political camp are reluctant to establish friendly interactions with supporters of a divergent camp (see descriptions of this variable in the Appendix, Table A).

A preliminary bivariate test suggested the relevance of the theses associating polarization and erosion of democratic support. Below (Graph 2), one may observe a statistically significant and negative correlation (R = -0.498) between these measures, considering the 57 countries included in the analysis.

Graph 2
Correlation between Democratic Commitment (%) and Polarization

To classify national governments ideologically, we used the dataset “Identifying Ideologues: A Global Dataset on Chief Executives, 1945-2020” produced by Sebastian Herre (Herre, 2022)³. Taking as references papers on party ideology, this initiative identifies the orientation of the parties that head national governments relative to economic policy, classifying them into “right-wing”, “left-wing”, and “center” (see the distribution by groups in the Appendix, Box C). Governments about which there is insufficient information for the classification and those whose parties do not fit the criteria do not enter the classification. In addition to these absences, governments without executive branches are not covered.

The distribution of the percentages of committed respondents by ideological groups in the governments indicated a more favorable position of the center (Graph 3), which presented a higher concentration of cases and a higher median. The left-wing and right-wing groups obtained closer positions but with a slight advantage of the latter. A similar distribution involving the measure of polarization (Graph 4) suggested a more homogeneous picture, with practically identical medians, although the concentration of cases placed the centrist governments in a more favorable position, i.e., less polarized.

Graph 3
Distribution of Democratic Commitment by Government Ideology

Graph 4
Distribution of Polarization by Government Ideology

As a Level 2 economic control, we included GDP per capita adjusted for purchasing power parity⁴ (V13 in V-Dem Dataset) and Gini index. We also added the age of democracy of the countries as a control (Boix et al. 2013), taking into account the year in which the WVS data was collected.

Due to this hierarchical structure of the data and the level of measurement of the dependent variable, we used a multilevel logistic regression model to mainly understand the effects of triple interactions between levels involving polarization, government ideology, and individual ideology (Bryk & Raudenbush, 1992).

IV. Results

A null model, without any predictor, indicated a odds ratio of 1.07 for Democratic Commitment in all countries and a variance of 1.14 among national units. The intraclass correlation coefficient (ICC) (Hox et al., 2017) was calculated as 0.258, indicating that just over 1/4 of the variation in the probability of democratic commitment is explained by national-level factors, justifying the two-level modeling. This conclusion is confirmed by the significant variation in the conditional modes displayed in Graph 5.

Graph 5
Conditional modes for Democratic commitment, 2017-2020

Graph 6 shows results from a first model (details in Table B of the Appendix) containing all individual and national-level predictors, still without interactions, indicating that polarization, despite showing a negative effect on a democratic commitment (about 7%), does not reach a statistically significant level (0.7). The fragility of the evidence prevents a conclusive analysis of the thesis recurrent in the literature, that polarization leads to the erosion of democratic legitimacy. With regard to ideology, commitment tends to be a more prevalent attribute, with the center as a reference, in the moderate left and less frequent in the right, whether moderate or extremist.

Graph 6
Individual and contextual predictors of democratic commitment, 2017-2020

The model results, presented in Graph 7, show the interaction of national polarization with individual ideology and indicate changes in the probabilities of commitment for right groups (details in Table C of the Appendix). The probability of an individual identified with the left (extreme or moderated) is not affected by the polarization of the national political context (p-value of 0.4 and 0.09). The effects on the right groups are negative, indicating that as the context polarizes, the probability of democratic commitment becomes lower. This analysis can be complemented with the predicted values for each category of ideology along the polarization scale (Graph 8). We can observe that the center is minimally affected by the advancement on the polarization scale, and the far left sees an increase in its commitment probability, although without statistical significance as we stated earlier. The moderate left also experiences a reduction in commitment, but its effect is not statistically significant either. The notable decrease in probability occurs for the right-wing groups.

Graph 7
Effects of interactions between polarization and ideology on democratic commitment, 2017-2020

Graph 8
Predicted probabilities for ideology on the polarization scale, 2017-2020

Graph 9 shows the results of the three-way interaction between polarization, the ideology of national governments, and individual ideology. We found that, in right-wing governments, the increase in polarization caused the probability of democratic commitment to increase in the two left-wing groups, even though the effect of the interaction is only statistically significant for the extreme position (p-value = 0.06). In this same political context, the probability of both right-wing groups decreases, but, just like before, with statistical significance only at the extreme position. In left-wing governments, polarization did not affect the probability of democratic commitment of the left-wing groups compared to the center, but it did negatively affect the two right-wing groups. In other words, when the government is left-wing, as polarization increases, right-wing groups have their chance of commitment to democracy reduced compared to the center.

Graph 10 displays the predicted probabilities of democratic commitment for ideological groups along the polarization scale in three government contexts and confirms these different trends (details in Table D of the Appendix). In the context of centrist governments, all groups lose commitment except for the far-right, which starts at considerably lower levels than the others. In the context of right-wing governments, increased polarization reduces the commitment of all groups, with the exception of the extreme left, whose probability increases along the scale. In left contexts, both right and extreme right lose commitment, while the other three groups, including the center, gain.

Graph 9
Effects of interactions among polarization, government ideology, and individual ideology on democratic commitment, 2017-2020

Graph 10
Predicted probabilities for ideology on the polarization scale, 2017-2020

In turn, the hypothesis of democratic reaction, in polarized contexts, by ideological groups that are in opposition was partially proven, specifically when the government is right-wing. In this case, compared to the center, there was a decline in the loyalty of the right wing to the democratic regime, while the commitment of the extreme left wing increased. It is important to highlight the considerable reverse effect in the latter context, since the reaction of the right when the government is left-wing is to withdraw commitment as polarization advances. Last but not least is the fact that in polarized contexts with centrist governments, all ideological groups lose their democratic commitment, reaching similar positions when there is extreme polarization.

V. Discussion

What can we say about the current theses on the relationship between polarization and democratic legitimacy? To address the limitations of previous studies, we tested the thesis of the negative effects of polarization on democratic legitimacy with a more significant number of countries and with more or less polarized contexts. If, on the one hand, the direction of the results was in line with the literature, indicating that polarization negatively affects democratic commitment, on the other hand, they were not statistically significant. What can be stated with greater conviction is that the opposite does not occur: the polarization of a country does not promote support for democracy.

Regarding the thesis that moderate voters would be the guardians of democracy, since when there is polarization, the ideological extremes would be more willing to renounce democratic commitment in order to guarantee the electoral victory of their party, the evidence presented here showed something different. The results pointed to the left wing as the group that remained most faithful to democracy in more polarized contexts, while the right wing was the group whose commitment to democracy declined the most. The lesson here is that the left and right wings present unalike profiles and attitudinal dynamics, as suggested by studies in the field of political psychology (Eysench, 1954; McClosky, 1958; Stenner, 2005; Jost, 2017; Clayton et al., 2021; Gandhi & Ong, 2019; de Leeuw et al., 2021), which has not yet been considered in the literature on polarization and democratic legitimacy.

As for the expectation of the existence of heterogeneous effects depending on the alignment (or not) between ideologies of individual and government, the results did not show a clear pattern. Our initial bet was that, considering this “fit” of the ideology of the two levels of analysis (the individual and the party in power), the guardian of democracy would no longer have a universal profile, but would vary according to the political context. The substantive phenomenon underlying such a dynamic would be the democratic reaction on the part of the group ideologically opposed to the party in power, while the group ideologically aligned with the party would be less emphatic in its defense of democratic regime because it has more to gain from the government.

In right-wing polarized governments, we have in fact observed an attitudinal pattern compatible with this model: an increase in the democratic commitment of the left wing and a decrease in the democratic commitment of the right wing. However, in left-wing polarized governments, it was the center that remained most faithful to democracy.

In general, we can say that the debate about polarization and democratic legitimacy wins when it begins to consider ideology, both in terms of its heterogeneous effects and the political context. The combination of the theories of the guardian of democracy, the rigidity of the right wing, and the democratic reaction offered a model that better portrayed reality than each theory alone.

VI. Final considerations

Our study showed that the attitudinal consequences of polarization are more complex than we thought. Comparing dozens of countries at different times and with varying degrees of polarization, not even the effect most highlighted by the literature - the erosion of the foundations of democratic legitimacy - was evident. We also did not find a pattern indicating an ideological group exercising, in a polarized environment, the universal role of guardian of democracy. Furthermore, a theory of democratic reaction according to which, in a polarized context, the defense of democracy would fall to the opposition found only partial support in our study.

Given the empirical results presented, it is worth briefly reflecting on some of the analytical consequences of the study, noticeably recognizing its exploratory nature.

First of all, this paper draws attention to the dimension of extremism, especially right-wing extremism. We know this group has the least significant commitment to democracy, but it is still unclear how essentially individual aspects (such as personality traits) are articulated with the context. In order to elucidate this gap, it is essential that future comparative surveys include in their questionnaires psychometric batteries dedicated to identifying these traits linked to dispositions favorable to democratic principles. Only in this way will it be possible to establish more consistent relationships among ideology, personality, and democratic commitment, including testing hypotheses of mediation and interaction of the effects of the first two on the latter.

The second aspect refers to the very definition of context and which of its characteristics are relevant to offer significant explanations about the support of voters for democracy. We know that context is important, but there are still many questions about why and how it influences voter behavior. We intend to present a first effort in this direction, but it is necessary to advance in the identification of causal mechanisms based on case studies, process tracing of a set of exemplary cases, and experimental approaches. The results presented here may even serve as a guide for selecting cases to implement these analytical strategies.

By questioning the thesis of the homogeneous effects of political polarization, this text suggested paths for an eminently comparative agenda aimed at understanding this increasingly important phenomenon for the near future of democracy on a global scale.

• 1
  We would like to thank the anonymous referees for their suggestions and criticisms. We would also like to thank CNPq for funding the research that led to this article.
• 2
  The weighting of observations was done including the weight variable provided by the World Values Survey under the code wght_eq1000.
• 3
  “The dataset identifies the economic ideologies and political parties of heads of government in 178 countries and of political leaders (as coded by Archigos) in 178 countries from 1945 or independence to 2020. The dataset distinguishes between chief executives with leftist, centrist, rightist, and no discernible economic ideology”.
• 4
  To make it easier to visualize the results, this variable was recorded by dividing it by 1000.

References

• Abramowitz, A.I. (2010) The disappearing center: engaged citizens, polarization, and american democracy New Haven: Yale University Press.
• Abramowitz, A.I. & Saunders, K.L. (2008) Is polarization a myth? Journal of Politics, 70(2), pp. 542-555. DOI
  » DOI
• Adorno, T., Frenkel-Brunswik, E., Levinson, D. & Sanford, R. (1950) The authoritarian personality New York: Harper & Collins.
• Albertus, M. & Grossman, G. (2021) The Americas: when do voters support power grabs? Journal of Democracy, 32(2), pp. 116-31. DOI
  » DOI
• Anderson, C.J., Blais, A., Bowler, S., Donovan, T. & Listhaug, O. (2005) Loser's consent: elections and democratic legitimacy. New York: Oxford University Press.
• Boix, C., Miller, M.K. & Rosato, S. (2013) A complete data set of political regimes, 1800-2007. Comparative Political Studies, 46(12), pp. 1523-1554. DOI
  » DOI
• Bryk, A.S. & Raudenbush, S.W. (1992) Hierarchical linear models (applications and data analysis methods) New York: Sage Publications.
• Carey, J., Clayton, K., Helmke, G., Nyhlan, B., Sanders, M. & Stokes, S. (2022) Who will defend democracy? Evaluating tradeoffs in candidate support among partisan donors and voters. Journal of Elections, Public Opinion and Parties, 32(1), pp. 230-245. DOI
  » DOI
• Claassen, C., Ackermann, K., Bertsou, E., Borba, L., Carlin, R. (2021) Conceptualizing and measuring support for democracy: a new approach. SSRN. New York: Elsevier, pp. 1-30. Available at: <https://ssrn.com/abstract=4622645>. DOI
  » https://ssrn.com/abstract=4622645» DOI
• Clayton, K., Davis, N.T. & Wood, T. (2021) Elite rhetoric can undermine democratic norms. Proc. Natl. Acad. Sci, 118(23), p. e2024125118. DOI
  » DOI
• Coppedge, M., Gerring, J., Knutsen, C.H., Lindberg, S.I. & Teorell, J. (2024) V-Dem Dataset v14. Varieties of Democracy (V-Dem). Avaiable at: <https://www.v-dem.net/data/dataset-archive/>. Accessed on: May 3, 2024. DOI
  » https://www.v-dem.net/data/dataset-archive/» DOI
• de Leeuw, S. E., Rekker, R., Azrout, R. & van Spanje, J.H.P. (2021) Are would-be authoritarians right? Democratic support and citizens' left-right self-placement in former left-and right-authoritarian countries. Democratization, 28(2), pp. 414-433. DOI
  » DOI
• EVS/WVS (2021) European values study and world values survey: joint EVS/WVS 2017-2021 Dataset (Joint EVS/WVS) JD Systems Institute & WVSA. Dataset Version 1.1.0. DOI
  » DOI
• Eysenck, E. J. (1954) The psychology of politics London: Routledge & Kegan Paul.
• Fiorina, M. P., Abrams, S.J. & Pope, J.C. (2010). Cultural war? The myth of polarized America 3a ed. New York: Pearson Longman.
• Fossati, D., Muhtadi, B. & Warburton, E. (2022) Why democrats abandon democracy: evidence from four survey experiments. Party Politics, 28(3), pp. 554-566. DOI
  » DOI
• Fuks, M., Casalecchi, G. & Ribeiro, E. (2019) Determinantes contextuais da coesão do sistema de crenças democrático: evidências a partir da América Latina. Revista Brasileira de Ciência Política, 28, pp. 7-32. DOI
  » DOI
• Gandhi, J. & Ong, E. (2019) Committed or conditional democrats? Opposition dynamics in electoral autocracies. American Journal of Political Science, 63(4), pp. 948-63. DOI
  » DOI
• Graham, M.H. & Svolik, M.W. (2020) Democracy in America? Partisanship, polarization, and the robustness of support for democracy in the United States. American Political Science Review, 114(2), pp. 392-409. DOI
  » DOI
• Greenberg, J. & Jonas, E. (2003) Psychological motives and political orientation - The left, the right, and the rigid: comment on Jost et al. Psychological Bulletin, 129(3), pp. 376-382. DOI
  » DOI
• Grossman, G., Kronick, D., Levendusky, M. & Meredith, M. (2022) The majoritarian threat to liberal democracy. Journal of Experimental Political Science, 9(1), pp. 36-45. DOI
  » DOI
• Haggard, S. & Kaufmann, R. (2021) Backsliding: democratic regress in the contemporary world (Elements in political economy) Cambridge: Cambridge University Press.
• Herre, B. (2022) Identifying ideologues: aglobal dataset on chief executives, 1945-2020 Avaiable at: <https://www.bastianherre.com/data>. Accessed on: Maiy 2, 2024.
  » https://www.bastianherre.com/data
• Hox, J., Moerbeek, M. & Van de Schoot, R. (2017) Multilevel analysis: techniques and applications. 3a ed. New York: Routledge.
• Iyengar, S., Lelkes, Y., Levendusky, M., Malhotra, N. & Westwood, S.J. (2019) The origins and consequences of affective polarization in the United States. Annual Review of Political Science, 22, pp. 129-46. DOI
  » DOI
• Iyengar, S., Sood, G. & Lelkes, Y. (2012) Affect, not ideology: a social identity perspective on polarization. Public Opinion Quarterly, 76(3), pp. 405-431. DOI
  » DOI
• Jost, J.T. (2017) Ideological Asymmetries and the essence of political psychology. Political Psychology, 38(2), pp. 167-208. DOI
  » DOI
• Levitsky, S. & Ziblatt, D. (2018) How democracies die. New York: Broadway Books.
• Mason, L. (2018) Uncivil agreement: how politics became our identity Chicago: University of Chicago Press.
• McClosky, H. (1958) Conservatism and personality. American Political Science Review, 52(1), pp. 27-45. DOI
  » DOI
• McCoy, J., Rahman, T. & Somer, M. (2018) Polarization and the global crisis of democracy: common patterns, dynamics, and pernicious consequences for democratic polities. American Behavioral Scientist, 62(1), pp. 16-42. DOI
  » DOI
• Przeworski, A. (2021) Crises da democracia São Paulo: Cia das Letras.
• Stenner, K. (2005) The authoritarian dynamic Cambridge: Cambridge University Press. DOI
  » DOI
• Svolik, M.W. (2019) Polarization vs. democracy. Journal of Democracy, 30(3), pp. 20-32.
• Svolik, M.W. (2020) When polarization trumps civic virtue: Partisan conflict and the subversion of democracy by incumbents. Quarterly Journal of Political Science, 15(1), pp. 3-31. DOI
  » DOI
• Torcal, M. & Magalhães, P. (2022) Ideological extremism, perceived party system polarization, and support for democracy. European Political Science Review, 14(2), pp. 188-205. DOI
  » DOI
• Webster, S.W. & Abramowitz, A.I. (2017) The ideological foundations of affective polarization in the U.S. electorate. American Politics Research, 45(4), pp. 621-647. DOI
  » DOI
• Wuttke A., Gavras, K. & Schoen, H. (2022) Have Europeans grown tired of democracy? New evidence from eighteen consolidated democracies, 1981-2018. British Journal of Political Science, 52(1), pp. 416-428. DOI
  » DOI
• Wuttke, A., Schimpf, C. & Schoen, H. (2023) Populist citizens in four European countries: widespread dissatisfaction goes with contradictory but pro-democratic regime preferences. Swiss Political Science Review, 29(2), pp. 246-257. DOI
  » DOI

Appendix

The next tables are part of the Appendix:

Box A
Countries, samples, and years of collection

Country	n	Year
Albania	1435	2018
Germany	3698	2017/18
Argentina	1003	2017
Armenia	1500	2018
Australia	1813	2018
Austria	1644	2018
Azerbaijan	1800	2018
Bangladesh	1200	2019
Belarus	1548	2018
olivia	2067	2017
Bosnia	1724	2017
Brazil	1762	2018
Bulgaria	1558	2017
Chile	3174	2018
Colombia	1520	2018
South Korea	1200	2017
Croatia	1487	2017
Cyprus	1000	2019
Denmark	3362	2017
Ecuador	1200	2018
Slovakia	1432	2017
Slovenia	1075	2017
Spain	1209	2017
United States	2596	2017
Estonia	1304	2018
Ethiopia	1230	2020
Philippines	1200	2019
Finland	1199	2017
France	1870	2018
Georgia	2194	2018
Greece	1200	2017
Guatemala	1203	2020
Netherlands	2404	2017
Hungary	1514	2018
Indonesia	3200	2018
Iceland	1624	2017
Japan	1353	2019
Lithuania	1448	2018
North Macedonia	1117	2019
Malaysia	1300	2018
Mexico	1739	2018
Montenegro	1003	2019
Nigeria	1237	2018
Norway	1122	2018
Peru	1400	2018
Poland	1352	2017
Portugal	1215	2020
United Kingdom	1788	2017
Romania	2870	2018
Russia	3635	2017
Serbia	1210	2018
Sweden	1194	2017
Switzerland	3174	2017
Taiwan	1223	2019
Czech Republic	1811	2017
Tunisia	1208	2019
Turkey	2415	2018

EVS/WVS, 2021

Box B
Variables from the World Values Survey used in the composition of the measure of Democratic Commitment

Variable	Code	Wording
Political system: strong leader	E114	I'm going to describe various types of political systems and ask what you think about each as a way of governing this country. For each one, would you say it is a very good, fairly good, fairly bad, or very bad way of governing this country?
Political system: military regime	E116	I'm going to describe various types of political systems and ask what you think about each as a way of governing this country. For each one, would you say it is a very good, fairly good, fairly bad, or very bad way of governing this country?

EVS/WVS, 2021

Box C
Groups by government ideology

Ideology	Countries
Right-Wing	Argentina, Austria, Australia, Azerbaijan, Bosnia, Bulgaria, Brazil, Belarus, Chile, Colombia, Cyprus, Czech Republic, Germany, Denmark, France, United Kingdom (GBR), Guatemala, Japan, Malaysia, Netherlands, Norway, Peru, Philippines, Serbia, Russia, Turkey, and USA.
Left-Wing	Albania, Bangladesh, Bolivia, Ecuador, Estonia, Spain, Ethiopia, Georgia, Greece, Hungary, Indonesia, Iceland, South Korea, Lithuania, North Macedonia, Mexico, Nigeria, Poland, Portugal, Romania, Sweden, Slovakia, and Taiwan.
Center	Armenia, Switzerland, Finland, Croatia, Montenegro, Slovenia, and Tunisia.

Herre, 2022

Table A
Descriptions of democratic commitment and polarization

Country	C_Dem (%)	Polar
Albania	68.8	1.01
Germany	78.8	1.55
Argentina	38.2	3.32
Armenia	31.7	3.42
Australia	72.6	2.84
Austria	82.9	2.20
Azerbaijan	30.3	2.56
Bangladesh	53.6	3.5
Belarus	33.2	3.14
Bolivia	31.4	3.31
Bosnia	40.4	3.17
Brazil	27.2	3.94
Bulgaria	45.5	1.74
Chile	52.8	1.76
Colombia	34.4	3.18
South Korea	30.8	2.69
Croatia	60.8	2.62
Cyprus	53.6	2.74
Denmark	81.2	0.17
Ecuador	19.7	3.44
Slovakia	62.3	1.49
Slovenia	71.7	2.85
Spain	72.7	1.99
United States	60.7	3.18
Estonia	81.7	1.47
Ethiopia	43.6	2.61
Philippines	13.7	1.61
Finland	83.6	1.23
France	73.1	2.67
Georgia	19	3.36
Greece	88.3	2.58
Guatemala	20.9	1.65
Netherlands	69	1.53
Hungary	75.3	3.81
Indonesia	17.8	3.62
Iceland	87.7	1.13
Japan	69.4	1.23
Lithuania	46.5	0.36
North Macedonia	19.9	3.4
Malaysia	23.9	3.63
Mexico	18.9	2.24
Montenegro	20.9	3.24
Nigeria	26.7	1.62
Norway	85.4	1.1
Peru	23.4	3.05
Poland	76.2	3.93
Portugal	43.2	0.63
United Kingdom	69.8	2.2
Czech Rep.	73.1	0.94
Romania	18.3	2.56
Russia	51.3	2.82
Serbia	32.9	2.21
Sweden	81.4	0.74
Switzerland	79.4	0.7
Taiwan	31.2	1.31
Tunisia	61	2.3
Turkey	45.5	4

EVS/WVS, 2021

Table B
Model without interaction

	Demo
Predictors	Odds Ratios	CI	p
(Intercept)	11.40	1.11-117.08	0.041
Idade [21-40]	1.07	0.94-1.23	0.307
Idade [41-60]	1.21	1.06-1.40	0.007
Idade [61-110]	1.32	1.13-1.54	<0.001
Educ	1.25	1.16-1.35	<0.001
RendFx	1.03	1.01-1.04	0.002
Sex	0.99	0.93-1.05	0.694
Relig	1.07	1.02-1.12	0.011
IdeolExtr [Ext.Left]	1.09	0.95-1.24	0.228
IdeolExtr [Left]	1.22	1.12-1.34	<0.001
IdeolExtr [Right]	0.89	0.82-0.97	0.011
IdeolExtr [Ext.Right]	0.75	0.67-0.84	<0.001
PolarPol	1.07	0.78-1.48	0.668
Ideol nov [direita]	0.65	0.15-2.83	0.563
Ideol nov [esquerda]	0.41	0.10-1.66	0.210
Years	1.01	1.00-1.01	0.087
Gini	0.92	0.87-0.98	0.004
PIB1000	1.00	0.97-1.02	0.892
Random effects
σ2	3.29
τ00 cntry_AN	0.44
ICC	0.12
N cntry_AN	28
Observations	34274
Marginal R2 / Conditional R2	0.120 / 0.225

Note: bold indicates p-value less than or equal to 0.05.EVS/WVS, 2021Boix et al., 2013Herre, 2022Coppedge et al., 2024

Table C
Model with interaction between individual ideology and political polarization

	Demo
Predictors	Odds Ratios	CI	p
(Intercept)	10.59	0.99-113.70	0.051
Idade [21-40]	1.07	0.94-1.23	0.315
Idade [41-60]	1.21	1.06-1.40	0.007
Idade [61-110]	1.31	1.13-1.53	<0.001
Educ	1.25	1.16-1.35	<0.001
RendFx	1.03	1.01-1.04	0.003
Sex	0.99	0.93-1.05	0.710
Relig	1.07	1.02-1.13	0.007
IdeolExtr [Ext.Left]	0.92	0.60-1.40	0.685
IdeolExtr [Left]	1.54	1.17-2.03	0.002
IdeolExtr [Right]	1.41	1.06-1.87	0.018
IdeolExtr [Ext.Right]	1.25	0.82-1.90	0.308
PolarPol	1.15	0.83-1.60	0.408
Ideol nov [direita]	0.65	0.15-2.89	0.572
Ideol nov [esquerda]	0.41	0.10-1.68	0.213
Years	1.01	1.00-1.02	0.076
Gini	0.92	0.87-0.97	0.004
PIB1000	1.00	0.97-1.02	0.863
IdeolExtr [Ext.Left] ×
PolarPol	1.07	0.91-1.27	0.401
IdeolExtr [Left] ×
PolarPol	0.90	0.81-1.02	0.090
IdeolExtr [Right] ×
PolarPol	0.83	0.75-0.93	0.001
IdeolExtr [Ext.Right] ×
PolarPol	0.82	0.70-0.96	0.013
Random effects
σ2	3.29
τ00 cntry_AN	0.45
ICC	0.12
N cntry_AN	28
Observations	34274
Marginal R2 / Conditional R2	0.126 / 0.231

Note: bold indicates p-value less than or equal to 0.05.EVS/WVS, 2021Boix et al., 2013Herre, 2022Coppedge et al., 2024

Table D
Model with interaction between individual ideology, national polarization, and government ideology

	Demo
Predictors	Odds Ratios	CI	p
(Intercept)	6.19	0.66-58.30	0.111
Idade [21-40]	1.04	0.93-1.15	0.490
Idade [41-60]	1.23	1.11-1.37	<0.001
Idade [61-110]	1.24	1.11-1.38	<0.001
Educ	1.57	1.49-1.65	<0.001
IdeolExtr [Ext.Left]	2.47	0.59-10.27	0.214
IdeolExtr [Left]	2.30	1.14-4.65	0.020
IdeolExtr [Right]	0.84	0.45-1.56	0.585
IdeolExtr [Ext.Right]	0.23	0.08-0.72	0.011
PolarPol	0.85	0.43-1.68	0.641
Ideol nov [direita]	0.71	0.12-4.35	0.714
Ideol nov [esquerda]	0.54	0.09-3.43	0.515
Years	1.01	1.00-1.01	0.153
Gini	0.94	0.90-0.97	<0.001
PIB1000	1.02	1.00-1.04	0.092
IdeolExtr [Ext.Left] ×
PolarPol	0.78	0.48-1.26	0.309
IdeolExtr [Left] ×
PolarPol	0.81	0.63-1.05	0.111
IdeolExtr [Right] ×
PolarPol	1.07	0.84-1.36	0.579
IdeolExtr [Ext.Right] ×
PolarPol	1.40	0.90-2.16	0.133
IdeolExtr [Ext.Left] ×
Ideol nov [direita]	0.27	0.06-1.17	0.080
IdeolExtr [Left] × Ideol
nov [direita]	0.62	0.29-1.31	0.208
IdeolExtr [Right] × Ideol
nov [direita]	1.71	0.87-3.33	0.118
IdeolExtr [Ext.Right] ×
Ideol nov [direita]	4.84	1.45-16.20	0.010
IdeolExtr [Ext.Left] ×
Ideol nov [esquerda]	0.43	0.10-1.87	0.260
IdeolExtr [Left] × Ideol
nov [esquerda]	0.62	0.30-1.28	0.195
IdeolExtr [Right] × Ideol
nov [esquerda]	1.67	0.87-3.20	0.124
IdeolExtr [Ext.Right] ×
Ideol nov [esquerda]	5.57	1.73-17.94	0.004
PolarPol × Ideol nov
[direita]	1.09	0.53-2.26	0.808
PolarPol × Ideol nov
[esquerda]	1.26	0.61-2.60	0.527
(IdeolExtr [Ext.Left] ×
PolarPol) × Ideol nov
[direita]	1.63	0.98-2.72	0.060
(IdeolExtr [Left] ×
PolarPol) × Ideol nov
[direita]	1.22	0.93-1.61	0.158
(IdeolExtr [Right] ×
PolarPol) × Ideol nov
[direita]	0.78	0.60-1.02	0.066
(IdeolExtr [Ext.Right] ×
PolarPol) × Ideol nov
[direita]	0.61	0.38-0.97	0.036
(IdeolExtr [Ext.Left] ×
PolarPol) × Ideol nov
[esquerda]	1.26	0.76-2.09	0.368
(IdeolExtr [Left] ×
PolarPol) × Ideol nov
[esquerda]	1.12	0.85-1.47	0.411
(IdeolExtr [Right] ×
PolarPol) × Ideol nov
[esquerda]	0.78	0.61-1.01	0.057
(IdeolExtr [Ext.Right] ×
PolarPol) × Ideol nov
[esquerda]	0.57	0.36-0.89	0.015
Random Effects
σ2	3.29
τ00 cntry_AN	0.47
ICC	0.13
N cntry_AN	57
Observations	77859
Marginal R2 / Conditional R2	0.201 / 0.301

Note: bold indicates p-value less than or equal to 0.05.EVS/WVS, 2021Boix et al., 2013Herre, 2022Coppedge et al., 2024

Publication Dates

• Publication in this collection
  26 Aug 2024
• Date of issue
  2024

History

• Received
  02 May 2023
• Reviewed
  13 Nov 2023
• Accepted
  31 Jan 2024