Conceptualization of Democracy: A Social Constructionist Approach

Several challenges complicate conceptualizing a multifaceted idea like democracy. First, there is the constructionist-versus-positivist debate surrounding the concept of democracy; social constructionists argue there are many versions of reality and therefore there is not “one” idea of democracy but many interpretations (Adler Reference Adler1997). Then there is the competing view of positivism, which has constraints that allow questions about truth and verifiable answers (Dessler Reference Dessler1999). This implies there is a verifiable understanding of democracy that can be empirically tested. Osterberg-Kaufmann, Stark, and Mohamad-Klotzbach (Reference Osterberg-Kaufmann, Stark and Mohamad-Klotzbach2020) distinguish meanings of democracy from understandings of democracy by arguing that understandings of democracy consist of different representations of an identical object and allow uniform measurement, whereas meanings imply conceptual ambiguity and contestation.

The tension between these two approaches is then further complicated by the debate between a universalist approach toward democracy versus a relativistic approach (Osterberg-Kaufmann, Stark, and Mohamad-Klotzbach Reference Osterberg-Kaufmann, Stark and Mohamad-Klotzbach2020; Reference Osterberg-Kaufmann, Stark and Mohamad-Klotzbach2023). The universalist approach argues that there is one understanding of democracy that can be understood by all people in all nations, since they all have similar reasoning capabilities (Zapf Reference Zapf, Da Rosa, Schubert and Zapf2016). Whereas the relativist approach argues that individuals understand democracy through the lens of culture, so there will be many varying interpretations of what democracy means because not all cultures agree on the same norms being “correct” (Zapf Reference Zapf, Da Rosa, Schubert and Zapf2016).

Newer approaches, such as the Varieties of Democracy (V-DEM) project, use expert ratings to create composite definitions of democracy that span across several interconnecting conceptualizations (Coppedge et al. Reference Coppedge, Gerring, Altman, Bernhard, Fish, Hicken and Kroenig2011), a process that draws from a social constructionist idea of democracy that allows for several definitions but uses positivist methods to measure them. To understand how citizens conceptualize democracy, we adopt V-DEM’s approach of providing several meanings of democracy and focusing on three specific understandings.

The tension between these two debates has led to a myriad of approaches to understanding democracy from the macro to the micro level. By providing richer scenarios for participants to choose from, this study increases the intension of the concept of democracy—that is, the set of defining attributes or properties it encompasses—while also prioritizing its extension, meaning the breadth of its application across participants’ understandings (Osterberg-Kaufmann, Stark, and Mohamad-Klotzbach Reference Osterberg-Kaufmann, Stark and Mohamad-Klotzbach2020; Reference Osterberg-Kaufmann, Stark and Mohamad-Klotzbach2023). The goal of this project is to identify patterns in which conceptualizations of democracy form at the citizen level. To do this, we will be taking a moderate social constructionist perspective.

Dimensions of Democracy: Processes, Protections, and Outcomes

To understand how Americans conceptualize democracy, we must first unpack it as a concept. Democracy is a nuanced idea with many competing scholastic definitions. The most widely used definition is the minimalist approach, which emphasizes democracy as a process centered on elections (Schumpeter [1942] Reference Schumpeter2013). However, this narrow focus overlooks other essential qualities of democratic governance, such as protections against discrimination and freedom to criticize the government. To address these gaps, scholars have broadened their definitions to combine democratic procedures with liberal democratic principles, emphasizing civil liberties and individual freedoms (Coppedge et al. Reference Coppedge, Gerring, Altman, Bernhard, Fish, Hicken and Kroenig2011). Additional frameworks include deliberative democracy, which focuses on mutual communication and reflection on common issues (Bächtiger et al. Reference Bächtiger, Dryzek, Mansbridge and Warren2018; Coppedge et al. Reference Coppedge, Gerring, Altman, Bernhard, Fish, Hicken and Kroenig2011), and egalitarian democracy, which highlights economic equality within a democratic society (Coppedge et al. Reference Coppedge, Gerring, Altman, Bernhard, Fish, Hicken and Kroenig2011). Each of these definitions captures necessary but individually insufficient aspects of democratic governance.

Monolithic definitions of democracy fail to capture how understanding of democratic governance varies both across and within nations. Research reveals stark differences between democratic and nondemocratic societies: citizens in nondemocratic nations tend to associate democracy with economic prosperity, obedience, and populist leadership, while those in democratic nations emphasize elections and rights protections (Kirsch and Welzel Reference Kirsch and Welzel2019; Zagrebina Reference Zagrebina2020). Even within established democracies, citizens differ in how they prioritize democratic components: while free and fair elections are broadly recognized as essential, the perceived importance of procedural elements varies significantly (Davis, Goidel, and Zhao Reference Davis, Goidel and Zhao2021). These multiple, often conflicting conceptualizations of democracy across and within nations challenge the validity of existing research on democratic support.

Thriving democracies must ensure citizens have equal access to political decision making and receive fair societal outcomes. Failure to do so results in a crisis of regime legitimacy, and citizens are less likely to trust the government and comply with its decisions (Tyler Reference Tyler2006). However, citizens are often willing to prioritize some goods over others. Lu and Chu (Reference Lu and Chu2021) identify three categories of citizens: those who believe democratic principles are essential regardless of circumstances (principle holders), those who are willing to trade those principles for more favorable outcomes (benefit seekers), and those who are unsure (fence sitters). Depending on whether one’s favored party is in power and how democracy is understood, Americans may shift between these roles (Krishnarajan Reference Krishnarajan2023). To observe this process of trade-offs, we need to split our overall conceptual definition of democracy into more distinct categories.

To simplify the concept of democracy and better observe trade-offs between principles and outcomes, we assert that democratic governance from a citizen perspective should be divided into three components: procedural democracy, distributive democracy, and liberal democracy. These three concepts represent separate dimensions of democracy that can often be conflated by the public: processes, outcomes, and protections. By separating these attributes of democracy, we can test whether participants value some aspects of democratic government over others, and whether voters believe “the ends justify the means.”

We draw inspiration for this structure from Pippa Norris, Sarah Cameron, and Thomas Wynter (Reference Norris, Cameron and Wynter2018) who employ a similar categorization when examining electoral integrity. Norris, Cameron, and Wynter distinguish between confidence in political institutions, regime performance, and regime principles. They define regime principles as reflecting the core procedures and institutions that are at the heart of functioning democracies, whereas they define regime performance as system “outputs” such as economic growth. These factors represent distinct priorities that citizens may emphasize when conceptualizing democratic government. More recent work has used a similar strategy to disentangle the effects of collinear concepts, identifying the defining features of a political system as access to power, the policy-making process, and performance (Papp et al. Reference Papp, Navarro, Russo and Nagy2024). We adopt a similar approach for parsimony but adapt these three categories to fit specific understandings of democracy.

Procedural and distributive democracy extend the concepts of procedural and distributive fairness to the broader framework of governance, highlighting the trade-off between democratic principles and outcomes. Procedural justice suggests that involving citizens in the decision-making process makes the outcome more acceptable (Lind and Tyler Reference Lind and Tyler1988), emphasizing individual participation and control over results (Thibaut et al. Reference Thibaut and Walker1975). Similarly, procedural democracy focuses on adherence to democratic processes and legal norms. What sets democratic leaders apart from nondemocratic ones are the norms guiding their acquisition of power and the laws holding them accountable (Schmitter and Karl Reference Schmitter and Karl1991). Thus, procedural democracy draws from this literature, defining democracy as a free and fair system in which decisions are made and power is equally distributed with citizens’ input.

Distributive fairness concerns the allocation of resources and opportunities within a country (Schnaudt, Hahn, and Heppner Reference Schnaudt, Hahn and Heppner2021). Applying this concept, a distributive view of democracy prioritizes positive social and economic outcomes over adherence to a free and fair process. For example, researchers suggest that socioeconomic status affects regime support: those in higher socioeconomic positions tend to support the status quo, while those in lower positions often oppose it, even if the status quo is democratic (Ceka and Magalhães Reference Ceka and Magalhães2020). This pattern also appears in how perceived levels of economic inequality correlate with support for democracy in East Asia and Latin America (Wu and Chang Reference Wu and Chang2019). Whether individuals have their economic needs met influences their evaluation of democracy and affects whether they view a regime as democratically legitimate (Ceka and Magalhães Reference Ceka and Magalhães2020).

Chu, Williamson, and Yeung (Reference Chu, Williamson and Eddy2024) identify that some citizens have a “substantive” view of democracy, where democracy is conceptualized in terms of tangible outcomes like economic benefits. This idea of distributive democracy has been supported by other recent survey work. When Pew asked 30,000 people across 24 countries how democracy could be improved, addressing basic needs like jobs, rising prices, and livable wages emerged as a key theme (Silver et al. Reference Silver, Fagan, Huang, Clancy, Chavda, Baronavski and Mandapat2024). Moreover, this tension between procedural democratic principles and socioeconomic benefits underpins populist movements throughout Latin America, where maintaining democratic processes often necessitates addressing inequality (Saffon and González‐Bertomeu Reference Saffon and González‐Bertomeu2017). Citizens may value ensuring economic prosperity and quality public goods such as education and infrastructure over competitive elections.

Finally, the concept of liberal democracy addresses civil liberties, such as the protection of human rights. It is distinct from both democratic processes and the distribution of social and economic benefits. For this concept, we draw on work by Dahl (Reference Dahl1983), who incorporates liberal democracy into his concept of polyarchy when he details the right to freedom of expression, the right to seek out alternative sources of information, and the right to organize as necessary freedoms in a democracy. Liberal democracy emphasizes the critical need to safeguard individuals and minority groups from oppression, whether by government power or majority rule (Coppedge et al. Reference Coppedge, Gerring, Altman, Bernhard, Fish, Hicken and Kroenig2011). This can be understood as a “negative” view of power—that is, liberal democracy represents protections, a “freedom from” discrimination and targeting.

These three dimensions of democracy are not mutually exclusive; it is possible to have democratic governance integrating all three dimensions. However, even if procedural democracy emphasizes including citizens in the democratic process, citizens may be willing to trade democratic principles for better distributive democracy if a regime fails to deliver favorable outcomes (Saffon and González‐Bertomeu Reference Saffon and González‐Bertomeu2017). To understand what prompts citizens to prioritize certain aspects of democracy above others, we employ a conjoint experiment that randomizes characteristics of these democratic categories and identify what individual-level characteristics correspond with these trade-offs.

The relationship between democratic performance and democratic principles also deserves careful examination, particularly the question of how economic outcomes might moderate citizens’ commitment to democratic values when faced with trade-offs (Saffon and González‐Bertomeu Reference Saffon and González‐Bertomeu2017). For instance, individuals are more likely to justify undemocratic actions from politicians when policies like healthcare and social spending align with their political interests (Krishnarajan Reference Krishnarajan2023), suggesting that some Americans consider democratic principles worth sacrificing for the sake of certain outcomes.

Building on this understanding, research shows that socioeconomic status further influences regime support: Ceka and Magalhães (Reference Ceka and Magalhães2020) argue that in democratic societies with widening income gaps, those in poverty are likely to be less committed to democracy than their well-off counterparts, indicating that economic well-being moderates attitudes toward democracy. Supporting evidence for this relationship extends globally: a 2021 Pew survey of 17 advanced economies—of which 16 were democracies—found that economic pessimists were more likely to desire significant political change (Wike and Fetterolf Reference Wike and Fetterolf2021). Taken together, these findings suggest that the distributive aspects of democracy, particularly economic well-being, may significantly condition individuals’ commitment to other procedural or liberal attributes of democracy against potential alternatives. Given this, we expect that preferences for normative levels of procedural and liberal democracy are highest when there is broad economic prosperity in which everyone’s needs are met, and housing and healthcare are affordable. In turn, preferences for normative levels of procedural and liberal democracy are lowest when people feel economically insecure and basic needs like housing and healthcare are economically challenging.

Therefore, we hypothesize the following:

Defining Democracy: Existing Work and Methods

Accurate measurement of citizens’ democratic preferences is important, because this influences whether they view democratic principles as fundamental or negotiable in exchange for preferred outcomes (Lu and Chu Reference Lu and Chu2021). However, though current projects like V-DEM can provide rich expert-level opinions, they cannot provide insights into “bottom-up” citizen-level support. Likewise, while large-scale surveys such as the World Values Survey, Eurobarometer, and other national barometers give us valuable breadth across nations, these datasets can only provide self-report information at the individual level.

For instance, Cho (Reference Cho2014) finds that individuals who understand liberal principles as core democratic elements are more likely to support liberal democratic governance. However, Cho’s study was limited to World Values Survey data, and the metric used to measure understanding of democracy evaluated whether participants defined democracy in terms of popular elections and civil liberties in contrast to a military takeover or a religious interpretation of the law. Understanding democracy in terms of popular elections and civil liberties is not incorrect but is also not comprehensive, as it does not leave room for distributive interpretations of democracy that can drive democratic satisfaction (Wike and Fetterolf Reference Wike and Fetterolf2021).

Additionally, the dependent variable for Cho’s study involved the explicit acceptance of democratic government compared with the rejection of authoritarian government. Acceptance of democracy as opposed to other systems of government does not isolate what individuals value about democracy; instead, it mainly quantifies relative dislike of authoritarian governance. Relying on self-reported understandings of democracy, therefore, does not necessarily reveal true preferences and predict what principles matter most to individuals.

This project diverges from previous studies by not attempting to assess the extent to which people hold a “correct” understanding of democracy. Democracy is a multifaceted concept that may vary both within and between nations according to sociocultural context (Davis, Goidel, and Zhao Reference Davis, Goidel and Zhao2021). Instead, we seek to introduce multiple conceptions to identify which are most preferred among the public. This helps us to draw out Americans’ tendencies to be principle holders or benefit seekers. Drawing on prior research and the three dimensions of democracy outlined above (Lu and Chu Reference Lu and Chu2021; Norris, Cameron, and Wynter Reference Norris, Cameron and Wynter2018), we break down democracy into four attributes: political equality, rule of law, freedom of expression, and economic well-being. These four attributes are manifestations of our three overarching understandings of democracy: political equality and rule of law represent the concept of procedural democracy, economic well-being represents the idea of distributive democracy, and freedom of expression represents liberal democracy.

These attributes of democracy improve upon past measures because they differentiate between democratic outcomes and processes. This is an important distinction because these conceptualizations of democracy are not inherently dependent on one another; each concept on its own is a necessary but individually insufficient condition for a democratic society. For instance, it is possible to have a nation that fairly votes in an illiberal candidate who does not value civil liberties or violates democratic norms, as seen in the 2018 presidential election in Brazil and the 2016 and 2024 presidential elections in the US.

To leverage individual-level data and evaluate citizens’ preferences for democracy, it is necessary to provide multiple conceptualizations of democracy and acquire individuals’ revealed preferences rather than their self-reported understanding of democracy. We address this open conceptual and methodological question of alignment between self-reported procedural, liberal, and distributive understandings of democracy and individuals’ revealed preferences for normative democratic governance by testing the following hypothesis in our study:

Research Design

Our central motivating questions are as follows: What are the attributes of the “ideal” democracy in which Americans wish to live? Do distributive outcomes of democratic governance condition preferences for normative procedural and liberal democracy? And do American’s self-reported beliefs align with their revealed preferences when they are forced to make choices and trade-offs? To answer the first question, we use a question format commonly employed on cross-national surveys such as the World Values Survey (Haerpfer et al. Reference Haerpfer, Inglehart, Moreno, Welzel, Kizilova, Diez-Medrano, Pippa Norris and Puranen2022) and the European Social Survey (Ferrín Pereira, Hernández, and Landwehr Reference Ferrín Pereira, Hernández and Landwehr2023; Quaranta Reference Quaranta2018) that asks respondents about the importance of different attributes of democracy. We answer the second and third questions by leveraging a conjoint experiment conducted with the same national sample of participants. Employing a framework developed by Hainmueller, Hopkins, and Yamamoto (Reference Hainmueller, Hopkins and Yamamoto2014), we generated randomized country profiles with four attributes and three levels. These four attributes—political equality, rule of law, freedom of speech, and economic well-being—represent our three dimensions of democracy.

We also provide different levels of our four democratic attributes, which represent different situations for our respondents. Two of our levels are illiberal majority and illiberal minority in reference to Zakaria’s (Reference Zakaria1997) definition of illiberal democracy: systems with competitive elections that fail to protect civil liberties and minority rights while disregarding constitutional constraints on executive power. We chose these varying levels of each attribute because individuals’ perspectives on democracy can be skewed by their subjective experience of it (Burger and Eiselt Reference Burger and Eiselt2023; Lindholm and Rapeli Reference Lindholm and Rapeli2023; Lindholm, Lutz, and Green Reference Lindholm, Lutz and Eva2024). Drawing on the emerging literature on illiberalism and majoritarian understandings of democracy, we define a core dimension of illiberal democracy as electoral majoritarianism—the rejection of limitations and rights-based constraints on political power (e.g., Enyedi Reference Enyedi2024; Laruelle Reference Laruelle and Laruelle2024; Scheiring Reference Scheiring, Sajó, Uitz and Holmes2021; Wunsch, Jacob, and Derksen Reference Wunsch, Jacob and Derksen2025). Both illiberal levels for each attribute in our experiment reflect this conceptualization, as they violate the principles of equal treatment and minority protection but differ in the participants’ perspectives. Thus, by varying the relationship of participants to liberal and illiberal democracy in each country profile, our design induces their subjective experience, much like real-world elections, prompting them to weigh certain democratic attributes over others and to reveal whether they prioritize self-interest and political positionality over liberal democratic norms. This will allow us to determine whether some Americans are benefit seekers or principle holders (Lu and Chu Reference Lu and Chu2021).

Participants viewed two country profiles at a time and were asked to choose the country that represented their ideal version of democracy for five rounds. This structure allows complete randomization of our profiles. Table 1 provides all attributes and levels used. We measure participants’ revealed preferences on these dimensions through their choice of country profiles and measure their self-reported understanding of democracy via survey items asked after the conjoint experiment section. An example of possible country profiles is included in figure 1.

Table 1 Attributes and Levels Included in Conjoint Experiment

Figure 1 Example Profiles for Conjoint Experiment

Case Selection, Samples, and Implementation

The sample for this project consists of 623 participants, drawn from an opt-in online panel in August 2024. Quota sampling was employed to ensure the sample was comparable to the US population on race, age, gender, and educational attainment. We chose a US sample because it presented a unique case study of a society with an increasing income gap (Muczyk, Nance, and Coccari Reference Muczyk, Nance and Coccari2009) and a national election in 2024. The existing political tension and economic climate during this period in the US gave us a salient test of whether individuals in an established democracy remain principle holders in unfavorable conditions or switch to being benefit seekers.

Revealed Preferences for Attributes of Democracy

Limitations and Generalizability

For this study we combined Bayesian and frequentist analysis strategies. Our main effects and attribute interactions derive from experimental data, but importance measures were generated using cjbart’s trained models to predict distributions across all possible conjoint profiles. The cjbart package employs robust Bayesian statistics and machine learning, allowing us to more accurately capture heterogeneity. However, despite the robustness of the analyses, the study had some constraints. For instance, we rely on cjbart’s model assumptions and accept its limitations. Though the cjbart model can handle any number of covariates with its use of Bayesian methods and machine learning, the VIMP process requires more traditional frequentist power analyses and therefore reduced our power for the importance heatmap. Additionally, IMCEs in this project shown in the importance heatmap and subsequent graphs demonstrate correlational relationships and descriptive patterns but are unable to prove causality.

Furthermore, though our self-report measures for liberal (RMSE = 0.0061), procedural (RMSE = 0.0014), and distributive democracy (RMSE = 0.0047) had good fit, our measures had weak internal consistency. Kendall’s W for all three measures was below 0.3873 (see table S3 in the supplementary materials), indicating a weak to moderate level of reliability. This presents a constraint for our analyses but also reinforces the fact that self-report measures are often insufficient for measuring abstract concepts like understandings of, or preferences for, democracy. We believe this is consistent with prior research highlighting the conceptual ambiguity and multidimensional nature of public understandings of democracy (e.g., Ariely and Davidov Reference Ariely and Davidov2011; Davis, Goidel, and Zhao Reference Davis, Goidel and Zhao2021; Quaranta Reference Quaranta2018). We interpret the low agreement as evidence that citizens do not hold uniformly structured or tightly integrated conceptions of democracy, and thus treat it as a key finding rather than as only a limitation.

Additionally, for our conjoint experiment we provided four conceptualizations of democracy for our participants to weigh. However, there are other attributes of democracy beyond political equality, rule of law, freedom of expression, and economic well-being that our participants may consider when deciding to be principle holders or benefit seekers. For instance, they may consider separation of powers to be distinct from political equality or rule of law, or may value concepts associated with deliberative democracy. Likewise, future research should consider aspects of distributive democracy other than the economy, such as education and job opportunities.

Moreover, we cannot generalize our results to past or future events given that our data are cross-sectional. Since our data were collected three months before a polarizing national election in the US, it is possible that the idea of democracy was more salient than in other years. We also encourage future research extending to both democratic and nondemocratic contexts, since individuals from other established democracies and different regime types likely experience democracy differently than Americans (Lu and Chu Reference Lu and Chu2021; Zagrebina Reference Zagrebina2020). However, our study’s approach and outcomes still offer a framework for similar investigations in other political settings. Our results suggest that there is a disconnect between self-reported understanding of democracy and revealed preferences for democratic government, which should be considered for nations beyond the US.

Our statistical analysis cannot conclusively determine the directionality of the moderation observed in our results. However, our findings align with our theoretical framework, suggesting that economic well-being is the most likely moderating variable. Prior research has shown that socioeconomic status influences the strength of individuals’ commitment to democratic principles (Ceka and Magalhães Reference Ceka and Magalhães2020). The variance pattern for economic well-being remains consistent when interacted separately with each democratic attribute. Additionally, economic circumstances—and by extension, economic well-being—tend to exhibit greater variance (Board of Governors of the Federal Reserve System 2025) than political equality, rule of law, and freedom of expression, which generally shift with election cycles, at least within the US. This aligns with historical patterns showing that incumbent candidates’ success fluctuates with the state of the US economy during election years (Guntermann, Lenz, and Myers Reference Guntermann, Lenz and Myers2021), reinforcing the idea that economic well-being is a consistent and influential factor in Americans’ political preferences and priorities.

Furthermore, the wording of the levels in our conjoint experiment differs from most designs in that we purposefully invoke the participant’s identity as part of the group involved in the profiles. Diverging from prior work, we sought to mitigate the potential influence of social desirability in our conjoint experiment by framing attribute levels to evoke respondents’ subjective experiences—much like real-world electoral events. This design more realistically simulates how individuals weigh normative democratic principles against potential economic or political benefits for their own group.

At the same time, we acknowledge that this approach may introduce some degree of social desirability bias, as respondents may hesitate to select profiles that appear self-serving or illiberal. We believe this bias would work against our findings. Social desirability would likely inflate support for normative democratic principles and suppress the selection of illiberal or self-interested profiles, making it less likely that respondents would appear willing to trade procedural or liberal democracy for distributive gains. The fact that we still observe meaningful variation in revealed preferences, especially the decline in support for political equality, rule of law, and freedom of expression under economic insecurity, suggests that our results are conservative estimates. In short, the true extent of benefit-seeking behavior may be greater than what we find here. Finally, though our analysis primarily focuses on the US, supporting evidence for the relationship between economic well-being and democratic preferences extends globally. For instance, a 2024 conjoint experiment across France, Hungary, and Italy found that respondents identify good economic performance as an important feature of a well-performing democracy. Critically, the authors note that the state of the economy had the largest impact on satisfaction with democracy across all three nations (Papp et al. Reference Papp, Navarro, Russo and Nagy2024).

This connection between economic well-being and democratic preference also manifests in recent studies on democratic backsliding. Rau and Stokes (Reference Rau and Stokes2025) find that cross-nationally, economic inequality is one of the strongest predictors of where and when democracy erodes. The prioritization of outcomes over democratic principles enables illiberal leaders to frame themselves as democratic by emphasizing their advancement of social and economic conditions (Chu, Williamson, and Yeung Reference Chu, Williamson and Eddy2024). In conjunction, individuals often rationalize undemocratic behavior from politicians when it serves policies they support (Krishnarajan Reference Krishnarajan2023; Saikkonen and Christensen Reference Saikkonen and Christensen2023). Taken together, these findings suggest that the distributive aspects of democracy, particularly economic well-being, may significantly condition individuals’ commitment to other procedural or liberal attributes of democracy against potential alternatives. Without considering how a nation distributes economic opportunities and social welfare—key decision factors for citizens—we are unable to capture the full scope of democratic support in a nation.