Overselling democracy–claiming legitimacy? The link between democratic pretention, notions of democracy and citizens' evaluations of regimes' democraticness

Brunkert, Lennart J.

doi:10.3389/fpos.2022.880709

ORIGINAL RESEARCH article

Front. Polit. Sci., 09 August 2022

Sec. Comparative Governance

Volume 4 - 2022 | https://doi.org/10.3389/fpos.2022.880709

This article is part of the Research Topic Researching Political Legitimacy: Concepts, Theories, Methods and Empirical Studies View all 9 articles

Overselling democracy–claiming legitimacy? The link between democratic pretention, notions of democracy and citizens' evaluations of regimes' democraticness

$\nLennart J. Brunkert$ Lennart J. Brunkert^*

Center for the Study of Democracy, Institute of Political Science, Leuphana University Lüneburg, Luneburg, Germany

Many non-democratic countries anchor the word “democracy” in their national constitutions and everyday rhetoric, while ignoring the conceptual roots of democracy and its scholarly-defined procedural standards. This article argues that governments intentionally “oversell” democracy to their people, in order to exploit the legitimizing effect that the word embodies. This can, however, only succeed if the receiving side is susceptible to such claims to legitimacy. Accordingly, this study investigates how effective “overselling” attempts are in light of individuals' liberal vs. illiberal notions of democracy. Building on congruence theory, it juxtaposes the, at times blatant, “overselling” with individual-level notions of democracy and, thus, investigates whether governments' attempts to claim democratic-procedural legitimacy are contingent on citizens' understanding of the concept. Using multilevel moderation analyses, it shows that illiberal, authoritarian notions of democracy can convert “overselling” into positive evaluations of a regime, whereas prevailing liberal notions unmask “overselling” governments and create additional criticality. The conclusion argues that notions of democracy function as a filter, which matches true and false demand and supply of democracy. The findings help to understand why and how democratization movements can unfold and why some citizens see their country as democratic even though it is not.

Introduction

The ideal of democracy has spread across the globe with more than 90% of the world's constitutions asserting democratic rule (Marquez, 2016, p. 22). Indeed, a truly astonishing variety of political systems around the world use the label democracy in their self-portrayal—from Swiss' direct democracy to Victor Orbán's envisioned “illiberal democracy” (see e.g., Buzogány, 2017; Bogaards, 2018) to the Chinese “socialist consultative democracy.” Aside locally and ideologically sourced labels for different political systems, scholars debate and define very clear criteria, which help to identify when a political system deserves the label “democracy” and when it does not. Procedural democracy—or in Dahl's (1971) words “Polyarchy”—is present when the population is guaranteed effective participation, voting equality, enlightened understanding, control of the agenda and inclusiveness. While some political leaders strive for such ideal-typical democratic systems and respect the “rules of the game,” others bend these rules and maintain “democracies with adjectives” (Collier and Levitsky, 1997) or straight-out autocracies. Though these regimes distort the ideals of democratic rule, many refer to themselves as democrats vis-à-vis their population. In fact: Marquez (2016) finds that of those 184 countries that proclaim democratic standards in their constitutions, only around 50 are certainly governed democratically¹. Constitutional references to the word democracy clearly do not make a political system democratic—just as government's repeated claims to a rational-legal, democratic rule do not guarantee the adherence to the procedural fundamentals that define said rule. But how is this “overselling” of democracy perceived by the population and when and why does it fall on fertile ground?

Political scientists have long argued that non-democratic regimes make strategic use of the positive image that concept of democracy embodies by misappropriating multiparty elections and pseudo-democratic institutions (Diamond, 2002; Levitsky and Way, 2002; Wahman et al., 2013). The ubiquity of the term democracy also reaches individual citizens in varying ways, leading to individual concepts of democracy: some of these are more liberal, some are rather illiberal (Dalton et al., 2007; Schedler and Sarsfield, 2007). Regimes that pretend to be more democratic than they actually are, should, thus, be able to maintain a congruence between their flawed delivery of democracy and their citizens often illiberal notion of the term.

The proposed concept of “overselling” relates the claim—to follow democratic-procedural norms—to the observed level of procedural democracy, as attributed by “objective” scholarly definitions. Empirically, this is operationalized by subtracting a countries level of procedural democracy from its rational-legal claims to legitimacy, which capture governments' claims to adhere to electoral principles and the rule of law. All countries that do not refer to democracy in their constitutions are excluded from the analysis which ensures that these rational-legal claims are backed by some form of constitutional promise of democratic rule. The main aim is to analyze the effectiveness of such governmental attempts to “oversell” procedural democracy to citizens, treating “overselling” as the main country-level explanatory variable. In the multilevel models, the individual's concept of democracy plays a moderating role, translating the governmental “overselling” attempts into individual assessments of current democracy-levels of their respective country. While most citizens are able to identify key components of democracy, others fail to do so and subsume illiberal, anti-democratic practices into their individual concept of democracy (Dalton et al., 2007; Cho, 2015). Concurrent with “democracies with adjectives,” Schedler and Sarsfield (2007) label these “democrats with adjectives.”

But how do “democrats with adjectives” perceive and evaluate “democracies with adjectives” and can “overselling” help to foster legitimacy-perceptions of political regimes?

Building upon previous work about “democracy misunderstood” (Kirsch and Welzel, 2019) and the citizens' “overrating” of their country's democratic quality (Kruse et al., 2019), this article adds an important puzzle-piece to the research on contemporary understanding of democracy and autocracy and its consequences. I argue that the missing piece lies in the interconnection of people's understanding of the concept democracy, the degree of “overselling” used by the ruling elite and the subsequent evaluation of democracy by said individuals. That is individuals' notions of democracy function as a filter used to process and reality-check governments' statements and transfer these into positive or negative evaluations of the incumbent regime. In order to legitimize their rule, political leaders create narratives that maintain a congruence between the concept of democracy that they supply and the concept that citizens demand. When supply and demand are incongruent, dissatisfaction with current levels of democracy and political opposition to the status quo might follow.

In technical terms, this paper investigates how “overselling” democracy affects people's evaluation of their governments' level of democracy and how this evaluation is moderated by different individual notions of democracy—while accounting for alternative explanatory factors. It fills a gap left in the literature, which has so far mostly focused on describing and explaining why people have different notions of democracy (see e.g., Dalton et al., 2007; Schedler and Sarsfield, 2007; Shin and Kim, 2018; Kirsch and Welzel, 2019). Going beyond these previous studies, this article does not inquire about the causes of popular “misunderstandings” of democracy but rather about their implications in the light of governmental legitimation strategies that “oversell” procedural democracy. It bridges the gap between questions of legitimation strategies, notions of democracy and support for democracy or autocracy, respectively.

The analysis uses data from the World Values Survey (WVS), the Varieties of Democracy project and additional sources, specified below. Using multilevel modeling, it clearly shows the importance of civic, political education when it comes to unmasking “overselling” governments' democratic rhetoric for what it really is—propaganda. The findings show that people with authoritarian notions of democracy fall for false claims and evaluate their country's level of democracy more positively than those who possess liberal notions of democracy. But even more so, liberal notions of democracy enable criticalness which allows citizens to see through the democratic façade of “overselling” regimes.

Previous research shows that education, emancipative values and individual-level socioeconomic conditions are the main drivers of individuals' liberal notions of democracy (see e.g., Cho, 2015; Kirsch and Welzel, 2019; Zagrebina, 2020). Accordingly, this study does not assume that regimes' “overselling” strategies shape people's notions of democracy. In that case, different popular notions of democracy could not moderate whether governmental “overselling” leads people to inflationary or deflationary ratings of their regimes' democratic legitimacy due to a lack of variance within countries². As there are many alternative explanations for satisfaction with democracy (see e.g., Claassen, 2020b; Kriesi, 2020; Wegscheider and Stark, 2020), I do not deem these findings as a penultimate causal explanation of how “overselling” works, but rather as highly relevant and informative description of how “overselling” and notions of democracy interact in forming democracy evaluations of citizens. The findings help to better understand, how and why seemingly authoritarian propaganda falls on fertile ground and why some populations reach a tipping point which makes them doubt their “overselling” governments and demand true democratic procedures. Potential causes for changing notions of democracy and their implications are picked up again in the discussion. Additional robustness checks presented in the Appendix further increase the credibility of the presented study and support the drawn conclusions.

The remainder of the paper is structured in the following way. The next section anchors the proposed argument on a theoretical foundation and presents testable hypotheses. Thereafter I discuss the conceptualization of “overselling democracy” and individuals' “notions of democracy” and operationalize the remaining variables. The section on data and methods summarizes the used modeling approach. Descriptive graphs and regression tables present the results, before turning to the evaluation of the hypotheses and a discussion of the results. The conclusion completes the previous discussion and embeds the findings in the larger political science literature.

Theoretical perspectives and hypotheses

This study argues that a government's “overselling” and individuals' notions of democracy are the two main components that define whether or not citizens' legitimacy demands and the respective government's supply are congruent and thus helpful for the maintenance of a regime's legitimacy. Congruence between what governments say and what citizens want to hear leads to a “perceived obligation to obey,” making the demand and supply logic of legitimation “a relational concept that connects a dominant A to a subordinate B” (Gerschewski, 2018, p. 653–55). Also Suchman (1995) relies on congruence to elaborate on his idea of legitimacy. He argues, that “legitimacy is socially constructed in that it reflects a congruence between the behaviors of the legitimated entity and the shared (or assumedly shared) beliefs of some social group” (1995, 574). These considerations make congruence theory suitable to this research, as it proposes that a well-functioning political system be contingent on a sufficient level of congruence between governmental authority structures and the citizens' authority beliefs (Eckstein and Gurr, 1975; Eckstein, 1997). Following Haldenwang's (2017) supply and demand approach to legitimate rule, this analysis posits that governments offer different interpretations of democracy—exaggerated or understated—to their subordinates, who wish for their individual demands to be met. Hence, congruence—and in this sense also legitimacy—can be achieved on two paths. An uninformed public matched with an exaggerating government or a democratic-enlightened public matched with true adherence to democratic standards that do not require additional legitimation attempts. So why is this relevant for our understanding of political regimes?

Legitimation forms a major pillar of political rule by enabling a governmental system to be seen as “proper and just” (Tyler, 2006, p. 375) and thereby allowing ruling elites to govern without the constant need for justification (Beetham, 1991; Gilley, 2009; Levi et al., 2009; Gerschewski, 2013). Legitimate rule is based on the adherence to established laws and the absence of “arbitrariness and despotism” (Tannenberg et al., 2020, 2). Two strands of research coexist here. On one hand, the search for the ideal political system which is legitimate in and by itself, based on a normative, philosophical understanding of legitimacy. On the other hand, the empirical-analytical understanding of legitimacy, coined by Weber (1978 [1922]) and more recently Beetham (1991, 1993), allows governments and regimes to be seen as more or less legitimate (Haldenwang, 2017). Suchman (1995) stresses the subjectivity and manipulability of legitimacy, while arguing that it is something that cannot be attributed to a system by any individual, but is created by the general acceptance of those that are subject to its decisions. Governments' legitimation strategies—exemplified here by the “overselling” of democracy—are an example of the empirical-analytical approach, as the need for justification exemplified by these strategies precludes an inherent legitimacy of the political system. Mimicking democracies and selling this façade to the population, thus, becomes a way to attempt to procure legitimacy (Schedler, 2002). Kruse et al. (2019, p. 319) argue that (non-democratic) regimes “[…] indoctrinate people, so as to make them believe to live in a democracy, even though they do not.” Their analysis points to the effectiveness of governmental legitimation attempts by explaining how and why people “overrate” their country's level of democracy and hence ascribe democratic-procedural legitimacy to their regime or do not. Complementing their analysis, in my article the question is no longer how right or wrong ordinary people are in their assessment of their country's level of democracy. Rather, the question is how effective are governmental attempts to claim democratic-procedural legitimacy and how is this effectiveness moderated by different popular notions of democracy?

Summing up the argument: When demand for and supply of democracy are in congruence, evaluations of the regime's democraticness should be more positive than in times of incongruence, reflecting the support for the regime and perceived legitimacy in the eyes of the ruled. If democracy, however understood, is in high demand everywhere³, then evaluations of a government's current level of democracy resemble either desire for change or satisfaction with the status quo. Building on the premises of congruence theory, this article asks how different portrayals of democracy interact with different notions of democracy in establishing regime evaluations. The main proposition is that popular notions of democracy moderate governmental “overselling” in such fashion that:

H1: Overselling increases ratings of the regime's democratic legitimacy among citizens with prevailing illiberal notions of democracy.

H2: Overselling decreases ratings of the regime's democratic legitimacy among citizens with prevailing liberal notions of democracy.

Conceptual background and variables

What exactly is “overselling”?

“Overselling” democracy—as defined here—refers to the claim to follow democratic procedures, without translating these into political practice. For example: Following the ousting of Mursi's Muslim-Brotherhood led government, Egypt's 2014 election-winner al-Sisi asserted to have the goal of following a previously established “roadmap to democracy,” while simultaneously preventing opposition candidates from competing and culminating in the return to military rule (Debre and Morgenbesser, 2017). In a similar vein, the Communist Party of China refer to their system of government as a “socialist consultative democracy.” Their permanent mission to the United Nations states that “China's democratic system has been continuously improved, and the forms of democracy are becoming more varied. The people are exercising fully their right to be masters of the state.” (Perm. Miss. of PRC to the UN, n.d.). Both countries are ranked as unfree by Freedom House (2020). Still, they make references to democratic principles and make claims they cannot or do not uphold. They are “overselling” democracy in order to profit from its universal positive connotations. Aside these two examples, the third wave of democratization (Huntington, 1991) brought with it a wave of democratic pretention, as most of the regimes that began transition did not fully consolidate and created a democratic gray zone—mostly populated by “hybrid” or “competitive authoritarian” regimes (O'Donnell, 1996; Carothers, 2002; Diamond, 2002; Levitsky and Way, 2002, 2010). These regimes rely on the “dictator's toolkit” (Frantz and Kendall-Taylor, 2014) to prevent meaningful regime change. I argue, that one of these tools is the “overselling” of democracy. International aid and status are often subject to improvements toward democracy (see e.g., Dietrich and Wright, 2015) and, more generally, democracy has a universal positive connotation and is deemed desirable by most citizens around the world (Bratton and Mattes, 2001; Inglehart, 2003; Dalton et al., 2007; Chu et al., 2008; Rose et al., 2011; Klingemann, 2014)⁴. Hence, it seems only logical for rulers to wave a democratic flag, anchor the term in their constitutions and thus exploit the lip-service to the word as a way to procure legitimacy. This sounds rather uninventive and easy to see through, but by anchoring a rational-legal rhetoric in everyday politics, citizens without the proper education and conceptual understanding of democracy potentially take these claims at face value and rate their respective regime as more democratic than it actually is: “overselling” succeeds. On the other hand, we know that individual-level socioeconomic conditions, education (Almond and Verba, 1963), emancipative (Kirsch and Welzel, 2019) and liberal democratic values (Nathan, 2020) improve individuals' conceptual knowledge of democracy in the long run and thus increase pressure on the supply side of the political system.

Measuring these governmental claims and relating them to measures of procedural democracy is a challenge, in so far that governments' claims differ in their intention and scope and are not readily quantifiable and comparable. Fortunately, the V-Dem Project introduced new expert-coded questions on claims to legitimacy (Coppedge et al., 2020, p. 208–10). Of these, the claim that most closely relates to concepts of procedural democracy measures regimes' use of rational-legal justification (Tannenberg et al., 2020)⁵. Rational-legal claims “pertain to legal norms and regulations as laid out for instance in the constitution regarding access to power (e.g., elections) as well as exercise of power (e.g., rule of law).” Thus, the rational-legal claims evolve around two main concepts, (1) electoral access to power and (2) rule of law-based exercise of power. For each country-year coders then evaluate the extent that “the current government refer[s] to the legal norms and regulations in order to justify the regime in place” (Coppedge et al., 2020, p. 209–10; Tannenberg et al., 2020). Though electoral access to power and rule of law are not sufficient to call a country a democracy, they form a necessary procedural basis of democracy⁶. Being well aware of these empirical limitations, rational-legal claims are the closest and most credible approximation for which a large set of empirical data is available. To ensure that regimes are not referring to undemocratic legal norms and regulations when resorting to rational-legal claims to legitimacy all countries that do not refer to “democracy” or “democratic” in their constitutions are excluded based on the coding of the Comparative Constitutions Project (Elkins et al., 2005)⁷.

The final variable “overselling democracy” is created by subtracting a measure of procedural democracy from the claims to rational-legal rule. This subtrahend of the equation, needs to reflect the logic of the rational-legal claim, while being inclusive enough to measure levels of democracy. Accordingly, it mirrors the claims' two main concepts: (1) electoral access to power and (2) rule of law-based exercise of power. The final “procedural democracy” variable is created by combining the V-Dem's variables clean elections, elected officials and suffrage (access to power) with the rule of law index⁸ (exercise of power). The aggregation combines complementarity—and substitutability-based understandings of democracy (see e.g., Brunkert et al., 2019) by using an equally weighted average of the sum and the product of the index's subcomponents (all variables taken from V-Dem V10 by Coppedge et al., 2020, p. 281–82, 42)⁹.

Figure 1 visualizes the idea of “overselling” democracy by plotting procedural democracy against rational-legal claims to legitimacy. Subsequently, the space can be separated into “oversellers” and “undersellers” along a 45° line. In wave six, New Zealand uses barely any claims but features high levels of procedural democracy. Zimbabwe occupies the opposite end of the scale and Georgia sits right in the middle. Georgia, thus, backs up its modest claims with modest procedural democracy, while Zimbabwe's claims are far from their actual democratic performance. After subtracting procedural democracy from rational-legal claims to legitimacy, governments that rely heavily on these while simultaneously not backing them up via institutional and procedural performance score high on “overselling.” The other end of the scale is occupied by countries that show very high levels of procedural democracy and completely or predominantly refrain from using related claims to legitimacy¹⁰.

FIGURE 1

Figure 1. “Overselling” and “underselling” democracy.

To put these observations into perspective, the two extreme cases—Zimbabwe and New Zealand—can serve as further insightful examples. Zimbabwe is in the WVS sample for wave six and seven and maintained a V-Dem coding of high levels of rational-legal claims to legitimacy even after long-term president Robert Mugabe was removed from office in 2017. His successor and former vice president Emmerson Mnangagwa has since taken over and reaffirmed his claim to power in the contested 2018 election, which he won with a majority of 50.8% of the votes (BBC News, 2018). In a New York Times opinion piece, he commits:

“[…] that in the new Zimbabwe, all citizens will have the right of free speech, free expression and free association. At the heart of this will be free and fair elections, to be held as scheduled in 2018, with all impartial observers who wish to witness the Zimbabwean democracy at work welcome to attend.” And “[…] that we are bringing about a new era of transparency, openness and commitment to the rule of law” (Mnangagwa, 2018).

While these claims stand prominent in the media, Zimbabwe has been found to crack-down on opposition actors, limit the space for free speech and held a highly contested election (Feldstein and Steven, 2022). In contrast to Mnangagwa's claims, V-Dem's measure for rule of law is on a downward trend since 2013. Freedom of expression and association are in decline since the 2018 election, though still above the levels experienced under Mugabe. At the same time, the references to the word “democracy” or “democratic” reached an all-time high with Zimbabwe's new constitution of 2013 (from 9 to 19 references). Taken together, the public statements of Mnangagwa and the countries continuing struggle to introduce real political reform and respect individual liberties paint a very clear picture of what this article defines as “overselling” democracy.

For New Zealand (NZ) the story is much simpler. According to the Comparative Constitutions Project (Elkins and Ginsburg, 2021), there is no mention of the word “democracy” or “democratic” in the long history of New Zealand's constitution. Simultaneously, NZ is a prime member of several scientific democracy scoreboards. Freedom House ranks them as “free” with a score of 97–99 out of 100, Polity IV awards them their highest score from 1945 to today and also in V-Dem's coding the country scores just a few decimals below the hypothetical maximum of one. These academic measures of democracy go hand in hand with an unobtrusive use of rational-legal claims to legitimacy by leading politicians in the country. Even though these claims are used, these are truthful depictions of institutional reality and not used to “oversell” something that does not exist, as shown in the case of Zimbabwe. The modesty of the used claims and the high level of democratic performance to back it up, thus, make NZ a quintessential case of underselling democracy. In fact, the NZ government could stress their adherence to established rule of law, electoral integrity and turnover much more extensively, without coming close to falling into the “overseller”-category.

Individual notions of democracy

The literature on support for democracy points out with increased frequency that democracy is not a universally understood concept (Dalton et al., 2007; Cho, 2014, 2015; Shin, 2015; Shin and Kim, 2018; Kirsch and Welzel, 2019), while its philosophical propositions—freedom, equality and self-determination—can be regarded as universal human values (Sen, 1999; Beetham, 2009). Dalton et al. (2007) combine existing survey research and use recoded open-ended questions about the content of democratic government to assess what the world's citizens associate with democracy. Their evidence suggest a relatively universal, positive notion of democracy, relating it foremost to individual freedoms and civil liberties. Besides overwhelming overt support for democracy (Inglehart, 2003), most individuals are able to identify the key components of democracy, defined as free elections, civil liberties and the rule of law (Cho, 2015). While being able to identify these key components, a surprisingly large share of citizens from new democracies, autocracies and even from established democracies fail to reject false statements about democracy and attribute illiberal governmental practices to the term democracy (Schedler and Sarsfield, 2007; Cho, 2015; Kirsch and Welzel, 2019). These divergent “notions of democracy” help explain the seemingly ubiquitous support for democracy, while, in extreme cases, reverting the meaning of support for democracy to support for autocracy (Kirsch and Welzel, 2019; Kruse et al., 2019). However, democratic experience can help overcome misunderstandings and long exposure to democratic institutions can engender the acculturation of democratic values (Mishler and Rose, 2002). Better knowledge of democracy also helps citizens to evaluate their countries levels of democracy more precisely (Wegscheider and Stark, 2020), while general support for democracy can speed up democratic consolidation and de-stabilize autocratic rule (Claassen, 2020a). From a political culture perspective, the elimination of existential threats and societal and economic modernization have been found to enable individuals and societies to move from survival to self-expression and emancipative values and subsequently drive demand for political change (Inglehart and Appel, 1989; Welzel et al., 2003; Welzel, 2013; Brunkert et al., 2019).

The moderator, notions of democracy, is conceptualized following Cho (2015) and Kirsch and Welzel (2019). For their analysis, Cho argues that “only those who were able to evaluate both sets of democratic and non-democratic regime characteristics accurately are rated […] as fully informed about democracy” (p. 241). Following Kirsch and Welzel (2019), this is measured by the variable “liberal notions vs. authoritarian notions of democracy” (henceforth LND-vs.-AND), which captures individuals' appraisal of liberal notions while simultaneously rejecting authoritarian notions of democracy. The variable is created using the item-battery on “essential characteristics of democracy” included in waves five to seven of the World Values Survey (Inglehart et al., 2020). Of these 12 items, four are best suited to represent the procedural understanding of democracy, featured here. The selection of political leaders via free elections and the attribution of the protection of civil liberties onto democracy form the liberal notions of democracy (aggregation via arithmetic mean). In opposition to these, authoritarian notions attribute to democracy the role of “religious authorities in interpreting laws” and the “takeover of the military in case of an incompetent government” (aggregation via arithmetic mean). LND-vs.-AND are then created by subtracting the values of the latter from the former¹¹. Hence, this measures a liberal notion of democracy in its higher scores that is free from illiberal, authoritarian notions.

Perceived democraticness of own country

The dependent variable of the multilevel model builds on the WVS' variable, in which respondents are asked “[…] how democratically is this country being governed today” on a 10 point scale. It is rescaled to a range of zero to one. Contrary to the questions on support for democracy or support for democratic government in general, this variable captures individuals' assessment of their current government's democratic performance. If democracy is seen as generally good and desirable, then evaluation of each government's democratic performance involuntarily become a measure of support for the current government, based on the notions of democracy which were inquired upon previously. Hence, it is only applicable to equate democracy ratings with support for the regime under the condition that individuals also express general support for democratic governance. To account for this conditionality, the dependent variable is multiplied (weighted) with the WVS' variable which asks respondents about their personal importance of democracy on a ten-point scale. High scores now indicate support for democratic governance and high evaluations of the current government, which I equate with a perceived legitimacy of the regime in place. Low values indicate that respondents either (1) do not support democratic governance in general and do not see their country as democratic, (2) do not support democratic governance, but think that their current regime is democratic or (3) support democratic governance but do not see the current regime as democratic.

Country-level control variables

To take into consideration the democratic learning perspective emphasized by Mishler and Rose (2002), the duration of the current regime, as well as the countries “democracy-stock” (Gerring et al., 2005) are included. The latter is created using each country's cumulative sum of their liberal democracy score (Coppedge et al., 2020) up to the year in which the country was surveyed. It captures the cumulative experience with liberal democracy since 1945. To account for skewness due to some longstanding highly democratic countries in the sample, the variable is log-transformed. The regime duration is captured by counting the years since regime change identified by V-Dem's regime information variable (Coppedge et al., 2020, p. 130).

A free media would potentially counter governments' attempts to convince the public of their democraticness and unmask political corruption and propaganda. Hence, the regime's censorship efforts are included using a combined variable (aggregated using the arithmetic mean) for broadcast, print and internet censorship efforts (Coppedge et al., 2020).

Also political violence might bias individual assessments of the current state of democracy of their country and is accordingly controlled for using the political terror scale (Gibney et al., 2020). Lastly, the log-transformed gross domestic product is added, to account for performance-based regime support (World Bank, 2019). Period affects are controlled via the inclusion of wave dummies.

Individual-level control variables

More educated and politically interested citizens should be better equipped to see through governments' false claims (Almond and Verba, 1963; Geddes and Zaller, 1989; Inglehart and Welzel, 2010; Campbell, 2019). On the other hand, education also reinforces the visions of the current political system in younger generations (e.g., Alwin and Krosnick, 1991; Sears and Valentino, 1997). Thus, the effect might be ambiguous. Education is recoded to harmonize previous WVS waves with the latest wave seven. Political interest is measured as the mean of the two WVS variables “interest in politics” and “important in life: politics,” after rescaling both to a zero to one range. Also financial satisfaction and income-scales might lead to ambiguous regime evaluations, if individuals attribute their economic situation to governments' performance (see e.g., Chu et al., 2008). Emancipative values (Welzel, 2013) are included at the individual and the country level, as these have been found to enhance criticalness and stand in opposition to authoritarian values (Kirsch and Welzel, 2019).

Data and methods

To probe the hypotheses, two analytical steps follow. First, a descriptive analysis outlines the basic relationship between the variables. Second, multilevel moderation analysis with cross-level interactions, random intercepts and slopes tests the proposed relationships while controlling for alternative explanatory factors. The nature of the data and research goal necessitate the use of a multilevel framework (Rabe-Hesketh and Skrondal, 2012), as individual survey participants are nested within countries and waves. The total scope of the analyzed sample comprises 70 countries, of which 31 have been surveyed only once, 27 have been surveyed twice and 12 have been surveyed three times¹². All level one variables are country-wave mean-centered, which is recommended especially in an interaction analysis, since it “[…] minimizes the possibility of finding spurious cross-level interaction effects” (Hofmann and Gavin, 1998; Aguinis et al., 2013, p. 1512;). Additionally, Heisig and Schaeffer (2019) argue for the inclusion of random slopes for lower level variables in cross-level interactions, as estimates can otherwise be biased anti-conservatively. Following their recommendation, the effect of notions of democracy is allowed to vary for each country-wave. The main model, with evaluations of current democracy levels as the dependent variable, can be spelled out as follows:

\begin{array}{l} Y_{i j} = β_{0} + β_{1} z_{i j} + β_{2} x_{j} + β_{3} x_{j} z_{i j} + β_{4 - 10} I_{i j} + β_{11 - 19} C_{j} \\ + u_{0 j} + u_{1 j} z_{i j} + e_{i j} & (1) \end{array}

Where β₀ denotes the constant, β₁ the coefficient for the individual level moderator liberal vs. authoritarian notions of democracy (z_ij) and β₂ the coefficient for the country-level variable “overselling” (x_j). The coefficient of the interaction term (x_jz_ij) is captured by β₃. Additionally, the overall intercept and the slope for the level one moderator are allowed to vary across countries. Thus, the random part of the model includes the random intercept u_0j, the random-slope parameter u_1jz_ij, as well as the level one error e_ij. I_ij denotes a vector of level one control variables with β₄₋₁₀ as its set of coefficients. C_j is a vector of the level two control variables, with its associated coefficients β₁₁₋₁₉.

All individual level data is taken from wave five to seven of the World Values Survey (WVS). By now, the WVS covers a majority of the world's population and provides the most extensive data to test the derived hypotheses. Each country-wave is considered as a separate level-two entity. Regime specific effects which might have changed in-between survey-waves are absorbed by the level-two random effect attributed to each country-wave cluster.

There exists some valid criticism toward the reliability of survey-responses in authoritarian countries (see e.g., Kuran, 1997). Citizens might fear negative repercussions when criticizing their government's performance—for instance appraising democracy while living under autocratic rule. To account for this potential bias, the notes (Appendix Notes 1) and robustness-check section (Table 3 in Appendix, model 2) of the appendix discuss this in detail and provide additional analyses, which validate the results presented here.

Descriptive analysis

Figure 2 visualizes the main argument and shows the level two variable “overselling,” the country-level average of the level one moderator LND-vs.-AND and the dependent variable in relation to each other. The first four graphs depict a selection of democratically run countries, in which “underselling” meets a democratic, critical and liberal citizenry which is then transferred into positive evaluations of each country's level of democracy. The second line shows countries with unclear or arbitrary relationships and the last line depicts the effectiveness of “overselling” democracy when combined with prevailing authoritarian notions in the citizenry. The “arbitrary” group does not seem to support any clearly observable patterns, however, when put into relation with the more extreme examples, a clearer picture emerges. Without the LND-vs.-AND in-between the other two variables, there is no observable pattern, as high levels of democracy evaluations seem to exist for different levels of “overselling.” Especially the first and last line emphasize this observation—for example Germans and Chinese see their country as similarly democratic. The variance of “overselling” alone cannot explain the democracy-ratings of the individual citizens. As hypothesized, the notions of democracy now function as a filter through which the claims of the regime are assessed and transferred into regime-evaluations. The bottom-line graphs show that if exaggerated claims to democratic legitimacy meet a flawed notion of democracy, regimes are evaluated more democratically than one would expect. On the other hand, if high levels of democracy and low levels of claims to legitimacy meet an “enlightened,” liberal public, the evaluations also turn out positive, but move closer to their associated notions of democracy.

FIGURE 2

Figure 2. “Overselling,” LND-vs.-AND and Democracy Evaluations: Country-wave mean values.

The varying angles that can be seen at the LND-vs.-AND-link between “overselling” and the dependent variable can be interpreted as varying levels of criticalness. The pointier the V-shape becomes, the less critical the population. This occurs most frequently in South-East Asian countries and Central-Asian countries, where citizens often attribute economic performance to democratic governance as the findings of Chu et al. (2008) and Shin and Kim (2018) suggest. The most critical publics approximate the shape of an inverted letter L—with LND-vs.-AND, on the one hand, and democraticness ratings, on the other, tending to be at a par. Understood quite visually, the notions of democracy become the turning point, which defines when unqualified support for an “overselling” regime turns into qualified criticalness. Once the public embraces liberal notions of democracy, the angle shifts from a positive evaluation toward a more critical evaluation, which does not take the government's claims at face value and evaluates their level of democracy accordingly.

Indonesia can serve as an example for such a within-country shift. In-between wave five and seven, liberal notions have given way to authoritarian notions (Δ_{LND−vs.−AND} = −0.13^***), while “overselling” remained stable. The result is an increased effectiveness of “overselling” and higher attributions of democracy to the incumbent government (Δ_DemocEval = 0.09^***) due to lower normative evaluation-standards. Unfortunately there is no new data available for Belarus, as the recent protests and the exposure of governmental lies would suggest that the public has moved toward liberal notions of democracy and has reverted the weak convex relationship visible in Figure 2 into a weak concave relationship with the notions of democracy resembling the turning point from unqualified support to qualified criticality.

Clustering the different types of regimes into “overseller,” “intermediate” and “underseller” and the individuals into “illiberal,” “mixed,” and “liberal” can help to better understand the distribution of these two main variables and their interaction in the following multilevel models. Table 1 shows the cross-tabulation of these two variables. As with most transformations from continuous to categorical measurement, the right selection of cut-off values is crucial¹³. Keeping these limitations in mind, the cross-tabulation shows that in each type of regime, individuals with all three categorized notions of democracy exist. However, there also is a premium of illiberal individuals in “overselling” regimes and of liberal individuals in underselling regimes. The category of mixed individuals makes up the largest fraction. This is in line with the findings of Schedler and Sarsfield (2007), Cho (2015), and Kirsch and Welzel (2019), who find that most individuals are able to identify key elements of democratic rule but—at the same time—fail to reject authoritarian practices as clearly undemocratic. Re-using the example of Indonesia, this categorization reveals that from wave five to seven, the share of individuals with illiberal notions of democracy increased by 5% (from 3.5 to 8.8) and that of individuals with liberal notions decreased by 10% (23– 13%), which explains how and why “overselling” falls on fertile ground. Additional notable examples of “overselling” regimes with large shares of individuals who subscribe to liberal notion of democracy include: Kazakhstan (wave 6: 41% LND), Moldova (wave 5: 49% LND), Belarus (wave 6: 42% LND), Armenia (40% LND), Burkina Faso (wave 5: 42% LND), Ukraine (wave 6: 48% LND), Russia (wave 6: 46% LND), and Morocco (wave 5: 44% LND).

TABLE 1

Table 1. Cross-tabulation of categorical versions of “overselling” and “notions of democracy.”

If I were asked to make any predictions based on the claims of this article, it would be that, these countries are most likely to experience protest—leading them to a new equilibrium between their regimes' democratic performance, claims to legitimacy and public demands for civil liberties, free and fair elections and the rule of law. In fact, we already saw an outburst of protest in early 2022 in Kazakhstan. Though fueled by rising prices, the protest included demands for more civil liberties and the fight against corruption (Khashimov and Couch, 2022). Morocco was part of the Arab Uprisings in 2011–2012, but opposition was appeased and co-opted by the existing regime. In line with this article's story, Thyen and Gerschewski (2017, p. 49) find that “[…] in Morocco, the perceived incongruence between claims to democratize the political regime and the perceived violation of these claims influenced the decision to participate in the 2011 protests.” Also Moldovans repeatedly protested against corruption and false promises and Belarussian pro-democracy protests dominated the media in 2020–2021. These examples exemplify how notions of democracy, that are present within a society, matter. Though, without more in-depth qualitative data, it is difficult to draw a causal arrow from widespread liberal notions of democracy to specific protest movements.

Multilevel regression results and discussion

The results of the multilevel moderation analysis are summarized in Table 2, whereof model one contains only the main variables and their interaction. Model two adds all individual level control variables and model three additionally tests for country-level confounders.

TABLE 2

Table 2. Evaluation of democratic quality.

The initial results are robust to the inclusion of a wide array of alternative explanatory factors. Since the level one moderator LND-vs.-AND is country-wave mean-centered, it captures the within variance partition. That is, comparing citizens with mean levels of LND-vs.-AND with those that score one unit higher. As extreme changes regarding “overselling” are not expected, the variable has been rescaled to a mean of zero and a standard deviation of one. In substantial terms and in the unaltered version of this variable the standard deviation amounts to 0.26. Thus, a one standard deviation increase in the model (toward more “overselling”) implies either a decrease of procedural democracy by 0.26, an increase of rational-legal claims to legitimacy by 0.26 or a combination of both (on a hypothetical range from −1 to 1 in the unaltered variable). These come close to changes that we can observe in i.e., cases of regime change or a leadership change, which may bring new claims to legitimacy.

In the main models including the interaction “Overselling,” LND-vs.-AND, as well as their interaction show high levels of significance. In interaction analyses, the “main” effect of the two interacted variables is contingent on another. This implies that a significant finding for a single coefficient shows that this variable has an effect on the outcome, when the second variable of the interaction is zero (here its mean value). Thus, “overselling” is perceived negatively and decreases the perceived level of democracy by citizens who lie on the mean value of LND-vs.-AND, which falls into the “mixed individuals” category of Table 1. Similarly, for LND-vs.-AND, the results show that for a case which shares the global mean value of “overselling” a more liberal notion of democracy enhances positive evaluations of a country's level of democracy. However, the interaction reverses this effect. When “overselling” meets a democratically well-informed citizen, their evaluation of their country's level of democracy becomes significantly more negative, speaking in favor of the hypothesized moderation. If authoritarian notions of democracy prevail, “overselling” can successfully enhance assessments of democratic performance.

Thus, notions of democracy indeed function as a filter, which helps individuals to separate true democratic, procedural performance from empty claims. Figure 3 disentangles this complex relationship and visualizes the interaction effect. It shows that liberal vs. authoritarian notions of democracy completely reverse the possible effect that “overselling” can have: even in case of strong moderations, complete reversals in the direction of an effect are rarely seen. The marginal effects plot supports the findings from the previous descriptive analysis, while simultaneously controlling for all potentially confounding variables (kept at mean values). Hence, hypothesis H1 cannot be rejected. When “overselling” falls on fertile ground—that is individuals with authoritarian notions—individual evaluations of democracy turn out positive and supportive of the regime in place. Though there is statistically significant support for this claim, these results should be taken with a grain of salt as the majority of the observations lays outside of the range that supports H1 and the previous cross-tabulation shows that only a minority of people hold truly authoritarian notions of democracy. However, hypothesis H2 cannot be rejected. Figure 3 makes clear that citizens with LND-vs.-ANDs around their respective country mean and above are well-equipped to critically assess their government's “overselling” attempts, leading to lower ratings of the regime's democratic legitimacy compared to citizens who score below the country-mean. That is, if “overselling” is met with predominantly liberal notions of democracy, citizens see through this façade and rate their country's level of democracy accordingly. In more substantial terms: Citizens with above mean LND-vs.-AND-scores that are subject to “overselling,” rate their countries level of democracy 0.05–0.1 points lower, which amounts to 5–10% of the dependent variable's scale. Strictly authoritarian citizens, on the other hand, see their country around 5% more democratic when they are a subject to “overselling.” Taking into consideration, that many other variables—many of which have been included here—influence citizens' evaluation of their countries democraticness, the findings of these models are substantial to warrant further in-depth analyses in the future.

FIGURE 3

Figure 3. Marginal effects plot.

In summary, congruence between citizens' demands and governments' supply of democracy is an important factor which helps explain evaluations of governments and subsequently their legitimacy in the eyes of the ruled.

Robustness checks

The results are robust to the exclusion of all countries that have been surveyed several times, keeping only the latest wave for each country (see Table 3 in Appendix, model 1) and to the exclusion of extremely repressive countries (see Table 3 in Appendix, model 2). Also the use of V-Dems original, unaltered democracy measures does not lead to substantially different results (see Table 3 in Appendix, model 3–4). Additionally, it might also be that LND-vs.-AND mediates the effect of “overselling.” Mediation in this case, would imply that “overselling” affects notions of democracy which then in turn influence survey-based assessments of a country's current level of democracy. To lend further credibility to the used multilevel moderation model, Figure 9, Table 4 in Appendix summarize the supplementary test for mediation. The results show no significant indirect effect and re-affirm that LND-vs.-ANDs are mostly driven by education, emancipative values and individual-level socioeconomic conditions, rather than by “overselling” of democracy. I see this as additional support for the moderation analyses employed here.

Control variables

The control variables mostly point in the expected direction. All individual level controls except for education have a significant effect on the evaluation of democratic quality. More interest in politics and individual level emancipative values create additional criticality and are associated with weaker perceived democracy levels. Higher income and financial satisfaction go along with better evaluations of the dependent variable, supporting a resource-based argument about the understanding of democracy (Dalton et al., 2007; Chu et al., 2008). Also confidence in the media results in supportive evaluations, as governmental communications are potentially evaluated less critical. Though the individual trust in the media has a positive effect, governmental censorship efforts do not go unnoticed and lead to a devaluation of democratic quality.

Discussion and implications

The possibility to benefit from democratic-procedural legitimacy, opens the door for non-democrats to justify their rule using democratic narratives while maintaining their grip on power. Many of the “democracies with adjectives” (Collier and Levitsky, 1997) that exist since the third wave of democratization (Huntington, 1991) have halted their development toward further liberal interpretations of democracy (Diamond, 2002; Levitsky and Way, 2010; Lührmann and Lindberg, 2019). Depending on the “democrats with adjectives” (Schedler and Sarsfield, 2007) that populate their electorate, they can maintain a congruence between popular demands and governmental supply of democracy by establishing their own versions of democracy that satisfy the contemporary demand. Previous studies have shown that this demand can shift due to increasing education, growing critical liberal desire for democracy (Nathan, 2020) or the development of emancipative values (Kirsch and Welzel, 2019). As lip service to democracy seems to be universal, it cannot be understated how important questions of democratic understanding in the form of “notions of democracy” are for scholarly analyses in this field of research. Without these, we fly blind and ignore substantial differences that can make the difference between the acceptance and condemnation of “overselling” governments.

Cultural demand for democracy moves at a glacial pace, slowly accumulating and increasing the pressure on the supply side (Welzel et al., 2016). If this pressure is not met by accommodating some of these demands, it can force a new equilibrium by challenging governments via popular dissent, coups, protests and the withdrawal of diffuse support (Easton, 1965). Growing liberal notions make the demand for true democratic procedures more widespread and increase incongruence with unprogressive governments. As also Nathan (2020) shows, liberal democratic values decrease support for illiberal leaders, whose “overselling” becomes increasingly ineffective—as this study proves. Hence, bottom up democratization outcomes are closely linked to the prevailing notions of democracy and the citizens' ability to see through “overselling” attempts. If movements for democracy are not anchored in an untainted demand for democracy, then the following supply will also not be truly and sustainably democratic. This becomes evident in half of the observed transitions being transitions from autocracy to autocracy (Geddes et al., 2014).

Conclusion

Shin and Kim (2018) summarize this field's puzzle quite well by arguing that “[i]ronically, those who consider themselves to be avid supporters of democracy show support for authoritarian rule, while those who live in authoritarian regimes view their countries to be more democratic than those who live in democratic ones” (p. 225). The literature on support for democracy and understanding of democracy has already uncovered many missing pieces in its research agenda and provides extensive explanations for misinterpretations of the term “democracy.” Still, it is obviously puzzling how high democracy ratings found around the world clearly do not map on the objective truth in countries governed by non-democrats. So why do people see their country as democratic even though it is not and why are some true democrats more critical of their democratic governments than others? To create a clearer picture, this article adds another missing piece to this field of research. Namely, the “overselling” of democracy by governments around the world.

The term democracy carries a multitude of positive connotations within and can be found in a majority of contemporary constitutions, public speeches and governmental narratives. Its ubiquitous appraisal creates a “legitimizing ideology” which governments have learned to utilize. By claiming to be democrats they harvest the fruits of democratic pretention, while maintaining their grip on power. However, this strategy is only effective if it falls on fertile ground. Individuals' notions of democracy shape the outcome of governments' “overselling” and can change unqualified support for autocrats into qualified criticality. The descriptive evidence and multilevel moderation analysis support this claim and clearly exemplify the importance of a well-educated and democratically enlightened public for truthful evaluations of a government's democratic vigor. Prevailing authoritarian notions of democracy can convert “overselling” into support for the regime, whereas liberal notions of democracy make people reject these false claims and evaluate governments more negatively. We already know that value change, education and modernization are drivers of liberal notions of democracy. Hence, these are enablers of qualified criticality. Many citizenries have not yet reached this tipping point, where “overselling” is no longer an effective tool of regime legitimation. Thus, “[a] major conclusion from these findings is that authoritarian regimes may seek to justify and strengthen their rule by attempting to induce constituents' perception of living in a democracy” (Thyen, 2017).

It remains to be seen in which direction individuals' notions of democracy develop in the future. Recent examples, such as the protests in Belarus or Thailand show the role that increased criticality plays in response to electoral irregularities like disbanding opposition parties or the claiming of victory by long-term incumbents. It also remains a question of the relation between demand and supply. Increasing liberal notions of democracy might be appeased by the accommodation of some demands, thereby re-establishing a new equilibrium close to the status quo. Recent debate about autocratization processes in established democracies point toward an alternative development (see e.g., Inglehart and Norris, 2017; Levitsky and Ziblatt, 2019; Lührmann and Lindberg, 2019). People with illiberal notions of democracy might perceive globalized democracy as an elite project detached from the individual citizen and the nation. The complexity of this project, combined with an impure notion of democracy, decreases the congruence between demand and supply and benefits those that simplify their language and question key democratic principles. These thought-experiments, mights and woulds deserve further attention and cannot be answered by this article. Having shown that there exist a generalizable degree of interaction between governments' over—or underselling and notions of democracy, this article creates a starting point for further in-depth inquiries. Ideally, this complex relationship would be explored with qualitative interview data or focus group discussions.

In summary—claiming adherence to electoral and rule-based standards, while neglecting democratic procedures is a widespread phenomenon, exemplified by competitive or hybrid regimes around the globe that did not successfully consolidate true democratic principles but rather consolidated the use of the word “democracy” as a legitimizing narrative vis-à-vis their citizenry.

Data availability statement

The original contributions presented in the study are included in the article/Supplementary material, further inquiries can be directed to the corresponding author.

Author contributions

The author confirms being the sole contributor of this work and has approved it for publication.

Funding

This publication was funded by the Open Access Publication Fund of the Leuphana University Lüneburg.

Acknowledgments

I wish to thank Christian Welzel, Christian von Soest, and Stefan Kruse for their invaluable comments on early versions of this article.

Conflict of interest

The author declares that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Publisher's note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Supplementary material

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fpos.2022.880709/full#supplementary-material

Footnotes

1. ^As Marquez (2016) points out, apart from a few exceptions, almost all regimes around the world describe themselves as a democracy. Leaving the definition of democracy to the countries' regimes would, thus, make the distinction between democracy and non-democracy meaningless. Therefore, the term “non-democracy' refers here and throughout the manuscript to “non-democracies' due to the scholarly standards of academic experts, in deliberate juxtaposition to how regime elites brand their political order.

2. ^An additional mediation analysis Appendix Figure 9 and Table 4) supports this assumption.

3. ^As Appendix Figure 1 clearly shows.

4. ^Appendix Figure 1 visualizes this ubiquitous, overt support for democracy based on two WVS questions.

5. ^For more detailed information on alternative regime legitimation-strategies, see Appendix Notes 2.

6. ^for conceptual debate on the rule of law see e.g., Møller and Skaaning (2012).

7. ^Interestingly, the countries that make the least references to the word democracy are some of the most established democracies (e.g., USA, Belgium, Denmark). Obviously, those are not excluded. Jordan, Malaysia, Singapore, Yemen and Cyprus are dropped from the sample, as they make no references to democracy in their constitution (Elkins et al., 2005).

8. ^Which in itself is a composite variable formed by Bayesian factor analysis. It combines the following indicators: Compliance with high court; compliance with judiciary; high court independence; lower court independence; executive respects constitution; rigorous and impartial public administration; transparent laws with predictable enforcement; access to justice for men; access to justice for women; judicial accountability; judicial corruption decision; public sector corrupt exchanges; public sector theft; executive bribery and corrupt exchanges; executive embezzlement and theft. (Coppedge et al., 2020, 282)

For more details consult the V-Dem Codebook v10 pp. 281-282.

9. ^Using V-Dem's unaltered, original “Liberal Democracy” or “Polyarchy” instead of the procedural democracy measure does not change the results of the analyses (Appendix Table 3, model 3 and 4).

10. ^Both components are measured in the year preceding each national survey of the WVS' waves five to seven to account for temporal ordering and the possibility that the survey interviews were conducted early in the specified year. Descriptive statistics for all variables can be found in the Appendix Table 2.

11. ^Herein I slightly deviate from Kirsch and Welzel, as they add gender equality to LNDs and obedience to authorities to ANDs. While I generally support their approach, the aggregation used here is capturing the procedural understanding of democracy better. As the logical opposite of “liberal' would be “illiberal', I repeatedly refer to authoritarian notions of democracy as illiberal notions of democracy throughout the text.

12. ^For a detailed list of all countries, waves and number of interviewees see Appendix Table 1.

13. ^The original WVS items (scaled 1-10) were each recoded into a binary variable, which shows support of a notion of democracy, if the individual gave it a score of 8 or higher, meaning that this aspect is a very important part of democracy for this individual. The binary scores are then summarized by subtracting the sum of important authoritarian notions from the sum of important liberal notions (resulting in five categories from−2 to 2). The categories are then labeled in such a fashion, that only those that fully subscribe to “free elections' and “civil liberties' and reject “religious laws' and “military takeover' are seen as fully liberal. Individuals who fully subscribe to one or two of the authoritarian notions and reject the liberal notions are seen as fully authoritarian. All others are classified as mixed. Changing the cutoff-value to 7 shifts the distribution slightly in favor of the liberal individuals. For the classification of regimes into “underselling', “intermediate' and “overselling', I use the variable metrics to identify the categories. Those regimes that lie ½ standard deviation (~0.13) above or below zero—the point where claims and institutions are in balance—are identified as “intermediate. Regimes outside this threshold are “oversellers' for values above this cutoff and “undersellers' for values below.

References

Aguinis, H., Gottfredson, R. K., and Culpepper, S. A. (2013). Best-practice recommendations for estimating cross-level interaction effects using multilevel modeling. J. Manag. 39, 1490–1528. doi: 10.1177/0149206313478188