Introduction
The global trend of autocratization has raised pressing questions about the robustness and universality of democratic commitments. While many citizens around the world continue to express support for democracy in surveys (Wuttke et al. Reference Wuttke, Gavras and Schoen2022), recent evidence reveals declining democratic endorsement in several advanced democracies, especially among younger cohorts who appear to be ‘democrats in name only’ (Wuttke et al. Reference Wuttke, Gavras and Schoen2022: 416; see also Foa and Mounk Reference Foa and Mounk2016, Reference Foa and Mounk2017), without a deeper understanding of core components of liberal democracy (Kaftan Reference Kaftan2024). However, existing survey instruments may overstate the depth of this support by presenting democracy in isolation, without reference to competing societal priorities.
This study, therefore, asks to what extent citizens prioritize core democratic institutions when faced with competing societal goals such as economic prosperity, public safety, or cultural cohesion. In other words, we explore how responsive citizens’ preferences are to democratic versus non-democratic attributes of governance, and which aspects of democracy – such as free elections, civil liberties, or executive constraints – are most salient.
Our theoretical foundation builds on the classic materialism versus post-materialism framework, which posits that citizens prioritize different political values depending on their levels of economic and physical security (Inglehart Reference Inglehart2005). According to this view, support for democratic norms is strongest among those who have moved beyond concerns for survival and stability. We also draw on the literature on the ‘authoritarian social contract’ (Desai et al. Reference Desai, Olofsgård and Yousef2009; De Mesquita and Smith Reference De Mesquita and Smith2010), which argues that citizens may tolerate or even support authoritarian governance in exchange for economic well-being and public goods provision. These perspectives suggest that preferences for democracy are not absolute, but conditioned by underlying security needs and contextual factors.
Despite these theoretical insights, existing empirical studies often assess democratic preferences through direct survey questions, which assume democratic support in a vacuum. Other work has examined trade-offs in the context of candidate choice, focusing on how democratic violations interact with partisanship, polarization, or competence (Frederiksen Reference Frederiksen2022; Graham and Svolik Reference Graham and Svolik2020). Yet these studies leave open the question of how citizens evaluate societal-level political arrangements that mix democratic and non-democratic features.
To address this gap, we designed a conjoint experiment involving over 35,000 respondents across thirty-two countries, including sixteen authoritarian regimes. We asked participants to evaluate pairs of hypothetical countries and choose where they would prefer to live. These country profiles vary along key attributes: democratic characteristics (free elections, civil liberties, executive constraints), economic prosperity, crime levels, and socio-cultural features (for example, gender equality and ethnic diversity). While the prompt posed to respondents is ‘What kind of society would you prefer to live in?’, our analytical focus is on understanding which dimensions of democracy are prioritized – and under what conditions they are deprioritized – in favor of alternative societal benefits.
In real life, people may have certain preconceptions pertaining to how specific regime types are more likely to offer material and physical security. On the one hand, there is evidence that democracy is linked to different indicators of economic development – including economic growth (see, for example, Broderstad Reference Broderstad2018; Knutsen and Dahlum Reference Knutsen, Dahlum, Coppedge, Edgell, Knutsen and Lindberg2022). Citizens who are aware of this, or just tend to associate democracy with economic growth, may also tend to prefer democracy precisely because they assume that a democratic government will provide better economic conditions. On the other hand, some people may associate authoritarian regimes with the idea of the ‘strong leader’, who can act decisively in response to insecurity. Our design allows us to more systematically isolate the relative salience of democratic institutions from people’s preconception of whether democracy should provide material and cultural outcomes. We test pre-registered hypotheses derived from the ‘security hypothesis’, which posits that perceived insecurity – whether economic, physical, or cultural – shapes openness to non-democratic governance that promises stability (Inglehart Reference Inglehart2005; Matovski Reference Matovski2021). We further examine how these preferences vary by regime type, development level, and individual characteristics such as prior democratic support and economic well-being.
Our results show that free and fair elections consistently emerge as the most salient democratic attribute. However, other democratic features, particularly executive constraints, are often less prioritized. Citizens are frequently responsive to models in which political leaders operate with limited oversight as long as those models are associated with greater economic prosperity. Even core liberties like free speech are sometimes subordinated to material security, though to a lesser degree.
These findings suggest a strong alignment with the appeal of electoral autocracy – a model that combines the formal presence of elections with weakened institutional checks and a focus on economic performance (Von Soest and Grauvogel Reference von Soest and Grauvogel2018). Citizens in both established democracies (for example, Australia, the UK) and autocracies (for example, Venezuela, Singapore) reflect this pattern, indicating that the authoritarian social contract may resonate broadly.
Finally, while electoral autocrats often manipulate democratic procedures to consolidate power, they maintain the façade of elections to legitimize their rule. Our findings underscore the strategic value of this façade: citizens appear to value elections as symbolic and procedural markers of legitimacy, even when other democratic norms are compromised.
Mass Support for Democracy
How robust is citizens’ support for democracy, and under what conditions are they willing to tolerate or endorse more authoritarian alternatives? This question has gained renewed urgency amid evidence of a global wave of autocratization, affecting an increasing number of countries and citizens worldwide (see, for example, Haggard and Kaufman Reference Haggard and Kaufman2021; Maerz et al. Reference Maerz, Lührmann, Hellmeier, Grahn and Lindberg2020). Most contemporary cases of autocratization are driven by democratically elected leaders and parties who, once in power, gradually erode democratic institutions from within, in some cases leading to their complete breakdown. (Bermeo Reference Bermeo2016; Grillo et al. Reference Grillo, Luo, Nalepa and Prato2024; Lührmann and Lindberg Reference Lührmann and Lindberg2019; Levitsky and Ziblatt Reference Levitsky and Ziblatt2019). These leaders often maintain substantial public support even after exhibiting authoritarian tendencies, suggesting that many citizens may be more receptive to non-democratic alternatives than previously assumed (Neundorf et al. Reference Neundorf, Öztürk, Northmore-Ball, Tertytchnaya and Gerschewski2025).
For a long time, it was widely believed that most citizens are strongly committed to democracy. This view has been supported by cross-national surveys such as the World Values Survey, which indicate that large majorities around the world express support for democracy, despite considerable variation across countries (Norris Reference Norris2017; Wuttke et al. Reference Wuttke, Gavras and Schoen2022). However, a growing body of research challenges the assumption that expressed support for democracy reflects a deep or stable commitment.
Survey items measuring abstract support for democracy or authoritarianism may overstate the extent of genuine democratic attachment (Claassen et al. Reference Claassen, Ackermann, Bertsou, Borba, Carlin, Cavari, Dahlum, Gherghina, Hawkins and Yphtach2024; Snagovsky and Werner Reference Snagovsky and Werner2024). In particular, numerous studies show that citizens’ willingness to stand up for democracy and democratic principles is sensitive to a wide range of other concerns, such as partisanship (Frederiksen Reference Frederiksen2024; Graham and Svolik Reference Graham and Svolik2020; Mazepus and Toshkov Reference Mazepus and Toshkov2022), the political competence of candidates running for office (Frederiksen Reference Frederiksen2022; Breitenstein and Hernández Reference Breitenstein and Hernández2025), trust in institutions (Gidengil et al. Reference Gidengil, Stolle and Bergeron-Boutin2022), or policy concerns (Lewandowsky and Jankowski Reference Lewandowsky and Jankowski2023; Saikkonen and Christensen Reference Saikkonen and Christensen2021). Providing further evidence on the fluidity of democratic preferences, Wunsch et al. (Reference Wunsch, Jacob and Derksen2022) finds that willingness to punish anti-democratic candidates varies based on people’s understanding of democracy (see also Grossman et al. Reference Grossman, Kronick, Levendusky and Meredith2022), while Krishnarajan (Reference Krishnarajan2023) demonstrates that partisans reshape their understanding of democracy to match their biases and political preferences – overlooking undemocratic acts by favored candidates while overstating similar violations by opponents. In sum, these findings suggest that while democracy remains a valued ideal, its salience can diminish when other political or policy concerns become more pressing.
While existing research provides compelling evidence that many citizens place other priorities above democracy, it leaves several key questions unanswered regarding the trade-offs people make between democracy and other societal concerns. First, previous studies, which largely focus on choices between hypothetical political candidates, may overestimate citizens’ willingness to accept authoritarian systems at the macro level. Citizens might tolerate seemingly undemocratic statements or actions by candidates because they do not believe such behavior will actually undermine democracy. This could be due to a perception that candidates will not follow through on undemocratic promises, confidence in the strength of democratic institutions to resist such threats, or a lack of recognition that these statements violate democratic principles (Grossman et al. Reference Grossman, Kronick, Levendusky and Meredith2022; Krishnarajan Reference Krishnarajan2023). In other words, we might be willing to support a candidate in an election who violates democratic norms and practices, but still value living in a democracy. To better understand these dynamics, it is necessary to examine not only the priorities voters have when choosing between candidates but also their preferences for democracy (or its absence) compared to other societal outcomes.
Prior research has underscored political factors such as partisanship and perceptions of competence in explaining tolerance for undemocratic actions. However, less attention has been paid to the role of economic, cultural, and social concerns. Although candidate choice designs can incorporate issue positions, they are limited in capturing how citizens weigh democracy itself against other dimensions of societal well-being – such as economic prosperity, public order, or cultural cohesion.
Addressing these limitations, Adserà et al. (Reference Adserà, Arenas and Boix2023) examine the extent to which citizens in the United States, France, and Brazil prioritize democratic institutions, particularly elections, relative to other societal outcomes. They find that ‘having democratic institutions strongly trumps all other considerations in determining the optimal society for an overwhelming majority of citizens’ (Adserà et al. Reference Adserà, Arenas and Boix2023, 1). However, their focus on elections – and on two long-established democracies plus one intermediate case – may lead to an overly hopeful picture of democratic commitment. Their findings are consistent with broader evidence suggesting that elections are seen as democracy’s most essential component. Chu et al. (Reference Chu and Yeung2025) broaden the scope to six countries, finding that people rarely sacrifice elections except in exchange for physical safety, while Ferrer et al. (Reference Ferrer, Hernandex, Prada and Damjan2025) demonstrate that liberal-democratic principles such as media freedom or judicial independence are more easily compromised than elections themselves.
Building on this foundation, our study offers new insights into the salience and stability of democratic support across an unprecedented thirty-two-country sample, including sixteen autocracies. We move beyond the limited focus on predominantly democratic cases in earlier research by enabling systematic cross-regime comparisons and providing a global assessment of democratic responsiveness. There are good reasons to expect that these patterns may differ in authoritarian contexts, as recent research has increasingly highlighted the role of political propaganda and indoctrination in shaping citizens’ attitudes (Akaliyski and Welzel Reference Akaliyski and Welzel2020; Neundorf et al. Reference Neundorf, Nazrullaeva, Northmore-Ball, Tertytchnaya and Kim2024; Rosenfeld and Wallace Reference Rosenfeld and Wallace2024). Under such conditions, political indoctrination and propaganda may contribute to the devaluation of democratic institutions in the eyes of citizens, while increasing the salience of cultural or economic concerns. Alternatively, the concept of thermostatic support for democracy (Claassen Reference Claassen2020) implies that citizens under authoritarianism may become even more supportive of democratic principles.
Our conjoint experiment disaggregates liberal democracy into its core components – elections, civil liberties, and executive constraints – within a unified experimental framework, allowing us to assess not only whether but which democratic institutions citizens prioritize. We also embed these democratic trade-offs within the broader logic of the ‘authoritarian social contract’, examining how material and physical security shape responsiveness to democracy. Finally, we expand the scope of this literature by incorporating cultural sources of (in)security, specifically, attitudes towards gender equality and ethnic diversity, which previous studies have largely overlooked. This addition allows us to show that cultural concerns, alongside economic ones, can also condition democratic preferences. Taken together, these innovations provide a more comprehensive understanding of how citizens evaluate democracy relative to competing societal goals and help explain the persistent appeal of electoral autocracy across diverse regime contexts.
Democracy Versus Economic and Non-Economic Security
Prominent theories on political attitudes and preference formation suggest that citizens’ perceptions of insecurity are a powerful influence on their political preferences, including their support for democratic institutions. Early contributions such as Fromm (Reference Fromm1957) argue that, when faced with an uncertain world and a lack of direction, individuals may seek psychological stability by turning away from democratic openness and embracing more hierarchical or controlled systems. At the micro level, this perspective often draws on theories of hierarchical needs, which posit that individuals prioritize more fundamental needs before they attend to higher-order goals. In particular, Maslow’s (Reference Maslow1958) influential ‘hierarchy of needs’ suggests that existential security – the assurance of basic survival – must be attained before individuals can pursue goals related to autonomy, freedom, and self-actualization. Basic needs such as food, shelter, and protection from harm – typically guaranteed or regulated by the state – form the foundation upon which more aspirational values, including civic and political engagement, can be built.
This framework suggests that citizens are more likely to prioritize democratic values once security is established (Inglehart Reference Inglehart2005; Welzel Reference Welzel2013). As different foundational needs are met, individuals gain the resources and autonomy needed to pursue a broader range of goals and experiences. This increased capacity, in turn, raises the salience of values associated with freedom, political voice, and institutional accountability. Accordingly, citizens with greater existential security may place a higher priority on democratic rights, such as civil liberties and free elections, than those who remain preoccupied with day-to-day survival and protection from perceived threats.
By contrast, under conditions of hardship and insecurity, democratic principles may become less immediately salient. In such contexts, citizens may place greater importance on political systems – of any kind – that they perceive as capable of responding effectively to their pressing needs. This is not necessarily an endorsement of authoritarianism, but rather an indication that democratic procedures and civil liberties may lose priority when survival and stability are at stake. Indeed, while authoritarian regimes are sometimes associated with swift and decisive action, there is a limited empirical basis for assuming that they are especially effective in addressing problems like crime or poverty. In many cases, authoritarian governance has been linked to corruption, inefficiency, and systemic instability. What seems more plausible is that, during periods of acute insecurity, citizens gravitate toward whatever form of governance appears responsive and capable of delivering order, security, and basic welfare – regardless of its democratic credentials.
This insight is echoed in the literature on regime legitimation in non-democratic contexts. The ‘authoritarian contract’ hypothesis, for example, holds that citizens may accept limitations on political rights if they believe the state can deliver economic or physical security in return (Desai et al. Reference Desai, Olofsgård and Yousef2009, 93). Similar dynamics are observed in rentier states, where governments use resource wealth to maintain political stability by ensuring material welfare, thereby reducing public pressure for political liberalization (Beblawi and Luciani Reference Beblawi and Luciani1987; Ross Reference Ross1999). This insight is consistent with the notion that preferences for democracy are partly or fully endogenous to other preferences, such as economic security (see, for example, Gratton and Lee Reference Gratton and Lee2024; Luo and Przeworski Reference Luo and Przeworski2021) or other policy outcomes (Svolik Reference Svolik2019).
The implication of this theoretical perspective is not that citizens categorically reject democracy under duress, but rather that the salience of democratic principles may be diminished when basic security is under threat. We therefore propose the following hypothesis.
HYPOTHESIS 1 People place greater priority on material and physical security than on democratic institutions and political rights.
At the same time, the relative priority citizens assign to democracy versus security is likely shaped by contextual and individual-level factors. Building on extensive literature on public opinion and political psychology, we expect that preferences are not formed in a vacuum but reflect citizens’ prior experiences, structural conditions, and deeply held beliefs (Zaller et al. Reference Zaller1992). We identify different reference points that may condition how people assess the importance of democracy in relation to security: the national economic context, the national political context, individuals’ own material conditions, their pre-existing democratic orientations, and their level of social conservatism.
First, we expect that citizens in economically underdeveloped or unstable national contexts will place greater emphasis on material and physical security, whereas citizens in more prosperous societies – where basic needs are more consistently met – are likely to view democratic institutions as more essential. Similarly, individuals experiencing personal economic hardship are expected to assign greater salience to issues of material security. This leads to our second hypothesis.
HYPOTHESIS 2 Compared to those living comfortably, people experiencing economic hardship assign greater priority to material and physical security over democratic institutions and political rights.
Second, we argue that the broader political context plays a crucial role in shaping how citizens evaluate the relative importance of democratic responsiveness versus other system goals such as economic prosperity or physical security. One might expect that citizens living under autocratic rule – or in recently transitioned democracies – would have a heightened appreciation for democratic institutions, having experienced first-hand the limitations of non-democratic regimes. However, research on authoritarian legacies suggests a more complex reality. In young democracies, public support for democracy tends to remain fragile and ambivalent, often shaped by lingering distrust, disillusionment, or weak institutional socialization inherited from the authoritarian past (Neundorf Reference Neundorf2010; Pop-Eleches and Tucker Reference Pop-Eleches and Tucker2017).
By contrast, in long-established democracies, democratic norms and institutional trust are more deeply embedded in society. Citizens are more likely to have internalized democratic values through consistent exposure to civic education, public discourse, and intergenerational transmission via family and community (Fuchs-Schündeln and Schündeln Reference Fuchs-Schündeln and Schündeln2015). These mechanisms reinforce the normative status of democracy, making citizens more likely to prioritize democratic institutions and political rights, even in the face of competing concerns such as material hardship or insecurity. As such, we propose the third hypothesis.
HYPOTHESIS 3 Compared to those living in autocracies, people in democracies assign greater priority to democratic institutions and political rights over material and physical security ( H3a ). This relationship is stronger in older democracies ( H3b ).
Hypothesis H3b reflects the idea that the prioritization of democratic responsiveness is not only shaped by regime type but also by the maturity of a democratic society. In newer democracies, where democratic institutions may still be consolidating and where citizens may have mixed experiences with their performance, democratic principles may coexist in tension with concerns over economic stability, safety, or other immediate needs. As such, citizens in these contexts may exhibit a more conditional or ambivalent commitment to democratic priorities. We therefore expect that only in older, more deeply institutionalized democracies will democracy be consistently prioritized over competing system goals.
Next, we turn to individual-level variation. While national context shapes the broader environment for democratic socialization, individuals differ in the strength of their support for democracy. Prior research shows that those with stronger normative or affective democratic commitments are more likely to endorse democratic procedures even when outcomes are unfavorable and are less susceptible to authoritarian or populist alternatives (Kaftan and Gessler Reference Kaftan and Gessler2025). Such attachments are often rooted in personal experiences, education, and political socialization, and can serve as a critical lens through which individuals evaluate political priorities.
We argue that these individual commitments influence not only abstract attitudes but also concrete preferences. When asked to weigh competing system goals, individuals who strongly support democracy should consistently prioritize democratic institutions – such as elections, civil liberties, and institutional checks – even when these come at the expense of economic stability or physical security. By contrast, individuals with weaker democratic commitments may more readily favor material outcomes over political rights. We thus expect democratic support to act as a psychological anchor that shapes how individuals evaluate alternative political systems. Thus, the fourth hypothesis.
HYPOTHESIS 4 Those with stronger stated support for democracy assign greater priority to democratic institutions and political rights over material and physical security.
So far, we have focused on economic and physical threats, but perceived insecurity can also stem from perceived cultural threats (Norris and Inglehart Reference Norris and Inglehart2019). Cultural backlash theory suggests that progressive social change – on issues like multiculturalism, gender equality, and reproductive rights – can provoke resistance among socially conservative groups who feel culturally marginalized (Norris and Inglehart Reference Norris and Inglehart2019). Socially traditional individuals tend to favor conformity over diversity, prioritize in-group loyalty, and resist changes that challenge established norms and values (Feldman Reference Feldman2003; Stenner Reference Stenner2005). When confronted with rapid cultural change, feelings of insecurity may lead them to value stability and order over freedom and democratic institutions. In this context, citizens may even accept authoritarian governance if it promises to defend their cultural values, for instance, by restricting immigration or limiting liberal social policies. This logic underlies the expectation that cultural threat perceptions increase support for authoritarian leaders and movements (see, for example, Wright and Esses Reference Wright and Esses2019; Donovan Reference Donovan2021). This yields an additional, pre-registered hypothesis.
Hypothesis 5 Compared to those with progressive values, people with more socially and culturally traditional attitudes place a higher value on socially conservative societies over democratic institutions and political rights.
Research Design
We conducted an online, well-powered, paired-profile conjoint experiment across thirty-two countries, involving a total of 35,281 respondents and generating 282,560 country-profile observations. This study was designed to test our hypotheses and received ethical approval on 17 May 2022 (approval number: 400210195). The study was pre-registered on 14 December 2022, before data collection commenced. Additional details on ethical considerations can be found in Appendix A.Footnote 1
In the experiment, we randomly assigned characteristics to hypothetical country profiles across seven dimensions. The first three dimensions – elections, constraints on the executive, and free speech – correspond to our expectations about democracy. The next two, the economy and crime, speak to our expectations about physical and material security. In addition, we also include the two following cultural characteristics: gender equality and cultural diversity.
Respondents were shown randomized combinations of country attributes and asked to choose which country they would prefer to live in. This forced-choice design avoids ‘cheap talk’ by requiring prioritizing between competing features. Unlike traditional survey items – such as rating the importance of democracy – our approach places all attributes on a common scale, compelling respondents to prioritize. A key strength of the design lies in the independent randomization of attributes, which allows us to isolate responsiveness to each feature. While this imposes orthogonality between attributes that may be correlated in reality, we see this as a manageable limitation, outweighed by the design’s advantages over standard importance or salience measures.
We acknowledge, however, the potential for violations of the information equivalence assumption (Dafoe et al. Reference Dafoe, Zhang and Caughey2018). Respondents may infer information about one attribute based on another – for example, assuming that a democratic country must also have a strong economy. While we mitigate this by providing explicit information on all key dimensions, some respondents may still interpret profiles through pre-existing associations. Crucially, our design does not require perfect perception of each attribute. It is sufficient that, on average, respondents who are shown a ‘high democracy–poor economy’ profile rate the economy more negatively than those shown a ‘high democracy–strong economy’ profile. We revisit this issue in the interpretation of our results.
Data Collection
Data collection for this project took place between 16 December 2022 and 21 May 2023. The surveys were first conducted in Turkey (December 2022), followed by English-speaking countries (January–February 2023), and Spanish-speaking countries (April–May 2023). Data collection in each country typically required about one week to complete. Table A.1 in the Appendix provides field dates for all countries.
Participants were recruited via paid advertisements on Facebook and Instagram – a method increasingly used in comparative political science for its global reach and flexibility (Boas et al. Reference Boas, Christenson and Glick2020; Zhang et al. Reference Zhang, Mildenberger, Howe, Marlon, Rosenthal and Leiserowitz2020). Social media recruitment enabled broader country coverage and more diverse samples than commercial panels like YouGov or Qualtrics, particularly in non-Western contexts.Footnote 2 This approach allowed us to expand the scope beyond previous studies using similar designs, which were typically limited to three (Adserà et al. Reference Adserà, Arenas and Boix2023), six (Chu et al. Reference Chu and Yeung2025), or seven mostly democratic countries (Ferrer et al. Reference Ferrer, Hernandex, Prada and Damjan2025). Moreover, using a consistent recruitment method across all countries enabled systematic cross-national comparisons, including those of underrepresented contexts such as autocracies and countries in the Global South.
Opt-in online surveys, regardless of recruitment method, are susceptible to sampling bias due to self-selection – especially in countries with lower internet access. To mitigate this, we offered a prize draw (one $500 online shopping voucher) to attract less-educated and less politically engaged respondents (Neundorf and Öztürk Reference Neundorf and Öztürk2025). We also used Facebook’s targeting tools to balance ad delivery by age, gender, and education across subgroups (Neundorf and Öztürk Reference Neundorf and Öztürk2023; Zhang et al. Reference Zhang, Mildenberger, Howe, Marlon, Rosenthal and Leiserowitz2020).
Overall, our advertisements reached 2.5 million people, yielding 35,281 valid responses to at least one country comparison. Appendix H compares our sample demographics with population estimates from representative surveys in each country. While our samples are generally representative in terms of age and gender, they still overrepresent highly educated and politically interested individuals in some countries (as is usually the case in online surveys). Respondents with liberal attitudes on gender issues are also overrepresented in certain countries, but our samples come close to population benchmarks in most countries in terms of nativist attitudes. These sample comparisons are presented in detail in Online Appendices H and J. However, subgroup analyses presented in Appendices F and G show that these biases do not invalidate our results, as they remain consistent across core demographic variables.
Furthermore, as Krupnikov et al. (Reference Krupnikov, Nam and Style2021) points out, there is growing evidence that experimental treatment effects drawn from online convenience samples – like ours – reliably replicate across more representative probability samples. Nevertheless, we are cautious about the implications of our findings for the external validity of sample average treatment effects (Boas et al. Reference Boas, Christenson and Glick2020). Another potential threat to data quality in online samples is bot activity. In Appendix C, we present a detailed analysis of data quality in this sample. Our analysis finds no indication of bot activity during participant recruitment.
Case selection
Some hypotheses focus on the contextual factors influencing respondents’ evaluations of hypothetical countries, particularly economic (H2) and political (H3) conditions. Case selection was guided by the need for variation in democracy (democratic v. autocratic) and economic development (low v. high). Practical constraints also influenced case selection. First, our reliance on paid social media advertisements limited the sample to countries with at least two million regular Facebook users.Footnote 3 Second, we prioritized countries where English or Spanish, the primary survey languages, are widely understood.Footnote 4 Turkey was included as a representative electoral autocracy with relatively higher development than most autocracies in our sample.
Figure 1. Case selection by democratic and human development.
Source: V - Dem and UNDP 2021 values.
Based on these criteria, we included thirty-two countries in our study.Footnote 5 Appendix Table A.1 lists the number of respondents and profile observations per country. Figure 1 visualizes case selection by liberal democracy scores (V-Dem) and the Human Development Index (UN), showing substantial variation in these contextual variables.
Conjoint Experimental Design
The experiment compared two hypothetical countries. Each respondent evaluated five pairs of countries,Footnote 6 indicating their preferred country, and assessed the quality of life in each country (results for the latter are presented in Appendix D). Each country profile was randomly assigned one attribute from seven dimensions: elections, constraints on the executive, free speech, the national economy, crime, equal pay between genders, and ethnic diversity. Table 1 outlines all dimensions and attributes. To minimize ordering effects, dimensions were grouped into thematic blocks (democracy, security, society), which were randomized and then fixed for each respondent across all five comparisons.Footnote 7
Table 1. Dimensions and attributes for conjoint experiment
The core dimensions in our experiment are the three political attributes: elections, executive constraints, and free speech. These were selected as they represent the key components of liberal democracy most relevant in contemporary contexts.Footnote 8 Recent trends in autocratization often involve executives undermining institutional checks, especially the judiciary and legislature (Lührmann and Lindberg Reference Lührmann and Lindberg2019; Grillo et al. Reference Grillo, Luo, Nalepa and Prato2024). While some respondents may view unconstrained leadership as promoting efficiency, such preferences would not align with a strong democratic commitment.
The economic and crime attributes capture societal security – differentiating between material (economic) and physical (non-economic) security.Footnote 9 To explore the role of cultural features, we also include two additional items: one on gender equality (measured by equal pay) and one on societal diversity. Although gender equality may be influenced by regime type, it can vary within both democracies and autocracies, making it relevant to test whether respondents are willing to prioritize cultural values over democratic institutions.Footnote 10
The final list of dimensions might vary in their level of abstraction, which could influence the degree to which respondents prioritize them. For instance, aspects such as the constraints on the executive and free speech may be harder for some respondents to grasp than the concept of free and fair elections. However, this is not necessarily a methodological issue but rather a reflection of reality: certain characteristics of society – and democracy in particular – are more abstract and therefore likely harder to understand. This could explain why some aspects of democracy are valued more than others. To better explore respondents’ perceptions of these different democratic elements, the empirical section tests the extent to which these items influence how respondents assess and value a country’s political characteristics.Footnote 11
The selected dimensions balance generality and comprehensibility to ensure applicability across all thirty-two countries. Our design allowed us to include a diverse set of cases but necessitated abstract attributes to ensure that respondents from diverse political and developmental backgrounds could relate to and understand each concept.Footnote 12 We also sought to avoid presenting too many items, as doing so could make the task of choosing between two countries overly complex. Consequently, we excluded potentially interesting dimensions, such as nationalism, foreign policy, and religion.
Finally, since we aim to measure and compare responsiveness to different country features, the external validity of our design depends on how well the attribute contrasts reflect real-world variation. As outlined above, we constructed the experiment to mirror meaningful differences across countries – especially regarding liberal democratic features – while maintaining enough abstraction to ensure cross-national applicability. This necessarily results in some dimensions (for example, prosperity v. hardship) appearing more starkly contrasted than others (for example, executive constraints v. no constraints), which may influence observed effect sizes (Blumenau and Lauderdale Reference Blumenau and Lauderdale2024). However, if respondents perceive these contrasts differently, and this is reflected in their choices, we interpret it as a substantive finding – not a design flaw – suggesting that certain distinctions (for example, economic hardship) are simply more salient than others (for example, executive oversight).
Moderators
We examined contextual and individual-level heterogeneity in respondents’ responsiveness to democracy and other societal factors. We therefore split our sample into distinct groups and present the results separately. We consider the following country-level moderators. First, to test the impact of ‘macro-level development’ (H2), we use the Human Development Index (HDI) provided by the United Nations Development Programme (2022). This index, based on life expectancy at birth, expected years of schooling, and gross national income per capita, captures a country’s level of human development. Our results are presented using a median split of HDI for our sample, which is 0.7.
Second, to examine the impact of ‘regime type’ (H3a), we use the ’Regimes of the World’ (ROW) indicator from V-Dem to classify cases as either (electoral or liberal) democracies or (closed or electoral) autocracies. Additionally, we define the ‘age of democracy’ (H3b) using the ROW indicator by counting the years since a country’s last democratic transition. Autocracies are assigned a value of 0, while democracies are categorized as young (1–5 years), middle-aged (up to 99 years), or old (over 100 years).Footnote 13
We also include three individual-level moderators, measured before respondents were exposed to the country profiles. First, ‘personal economic standing’ (H2) is assessed with the question: ‘Which of the following comes closest to how you feel about your household’s income nowadays? (1) Living comfortably on present income to (4) Very difficult on present income’. For the analysis, we collapse categories 1 and 2 (no struggle) and categories 3 and 4 (struggle). Descriptively, 40 per cent of the sample falls into the no-struggle group, while 60 per cent falls into the struggle group. This measure is significantly correlated with macro-level development in the expected direction.
Second, individuals’ ‘support for democracy’ (H4) is measured with the question: ‘Do you agree or disagree with the following statement: Democracy may have its problems, but it is still the best form of government. (1) Strongly agree; (3) Neutral; (5) Strongly disagree’.Footnote 14 We collapse categories 1 and 2 (high support) and categories 3 to 5 (low support). High support accounts for 71 per cent of the sample, while low support constitutes 29 per cent. These proportions are roughly consistent across democracies and autocracies.
Third, individual-level social and cultural traditional values (H5) is measured by an index variable that is built on three questions measuring attitudes towards gender equality, homosexuality, and immigration, measuring agreement with the following three statements: ‘When jobs are scarce, employers should give priority to people born in this country over immigrants’, ‘When jobs are scarce, men should have more right to a job than women’, ‘Homosexuality is justified’. Responses ranged from strongly disagree (1) to strongly agree (5). In the analysis, we take the median split on this index to contrast liberal and conservative respondents. In the Appendix, we conduct a robustness check contrasting the 10 per cent most liberal and conservative respondents (see Figure A.21).
Modeling
We use simple linear regression with respondent-clustered standard errors to test our hypotheses (Hainmueller et al. Reference Hainmueller, Hopkins and Yamamoto2014). Specifically, we regress our preference-to-live outcome on the country attributes. We calculate average marginal component effects (AMCEs) – denoting the average effect of each conjoint factor averaged over the other (randomly assigned) factors – and, to further bolster the intuition of our findings, show responsiveness to democracy and economic prosperity simultaneously using a factorial interaction between the two dimensions. To the latter end, we sum the three democracy features to a scale ranging from 0 (not democratic at all) to 3 (fully democratic).Footnote 15
To test the heterogeneous effects hypotheses (H2–H5), we calculate AMCEs by subgroup without control variables, as our goal is to describe – not causally explain – differences in responsiveness to country features. To strengthen these comparisons, we also assess each subgroup’s general responsiveness to all country attributes. This accounts for the possibility that some subgroups are more responsive overall, which could affect interpretations of responsiveness to individual attributes. We construct a summary scale reflecting the average AMCE across all attributes for each subgroup, estimated in separate regressions to avoid multicollinearity.Footnote 16
We use AMCEs rather than marginal means in all subgroup analyses. Given our binary attributes and forced-choice outcome, marginal means would duplicate information without added insight (for example, an AMCE of 0.24 implies marginal means of 0.38 and 0.62). Importantly, AMCEs reflect not majority preferences but the strength and direction of responsiveness to attribute contrasts – a measure that combines both preference intensity and consistency (Abramson et al. Reference Abramson, Kocak and Magazinnik2022).
Results
In this section, we present the results in three sub-sections. First, we explore which country features citizens perceive as democratic by showing the effects of all factors on the perceived democraticness of the country profiles, another outcome we included in the conjoint. This helps us understand the mechanisms behind our findings, sheds light on core assumptions of our setup, and provides general knowledge about how citizens understand democracy. Second, we investigate citizens’ responsiveness to the various country features. Third, we condition the findings on theoretically interesting contextual and individual-level factors.
Manipulation Check: Do People Understand Democracy?
As discussed above, citizens’ responsiveness to aspects of democracy and other societal characteristics may be influenced not only by their relative value placed on democracy but also by which practices they perceive as democratic. Previous literature has shown that citizens’ understanding of democracy is often quite shallow (Kirsch and Welzel Reference Kirsch and Welzel2019). For instance, many citizens tend to associate the label democracy with other desirable features, such as economic stability or peace (Dalton et al. Reference Dalton, Shin and Jou2007; Ridge Reference Ridge2021). Others tend to mainly understand democracy as elections, while not considering liberal aspects such as institutional constraints and political rights (see, for example, Chu et al. Reference Chu, Williamson and Yeung2024; Kaftan Reference Kaftan2024; Wunsch et al. Reference Wunsch, Jacob and Derksen2022). Hence, to gauge whether citizens are conscious about being responsive – or not – to objective aspects of democracy as aspects of democracy, we also want to find out what they perceive as democratic.
To explore this, we present results from models investigating whether our dimensions tapping into democracy affect respondents’ assessments of how democratic a country is, drawing on the following outcome question – measured from 0 (very low) to 10 (very high) and rescaled to 0 to 1 – from our survey: ‘How would you rate the level of democracy in each country?’ As assessments of democraticness are influenced by other societal features than democratic institutions and practices, we also test whether our dimensions tapping into security and cultural characteristics shape assessments of democraticness.
Figure 2. Average marginal component effects of all attributes on perceived democraticness across the entire sample (manipulation check).
Note: 282,421 observations. Linear regression with country characteristics as independent variables and perceived democraticness as dependent variable. Standard errors are clustered on the respondent level.
Our results, summarized in Figure 2, indicate that all three democratic dimensions increase respondents’ assessments of how democratic a country is. Yet, elections and free speech matter considerably more than checks and balances. Factors such as economic security and wage equality are positively related to assessments of democracy, but the true, objective democratic features generally affect perceived democraticness the most. Hence, our respondents seem to have a relatively good grasp of what democracy is, which implies that responsiveness in preferences does to some extent reflect accurate considerations about democratic features. However, it is noticeable that constraints on the executive, for example, are not seen as a very important aspect of democracy, which is important to bear in mind when interpreting the main results.
Main Effects: Citizens Prioritize Elections and Economic Well-Being
As we have now established that the included dimensions of our experiment tap into realistic understandings of what constitutes a democracy, we turn to the question of which societal characteristics the preferences of citizens are most responsive to Figure 3 presents our main results, pooling all respondents and showing the AMCEs for each of our seven country attributes on our binary preference-to-live outcome.Footnote 17
Figure 3. Average marginal component effects of all attributes on living preferences across the entire sample.
Note: 282,560 observations. Linear regression with country characteristics as independent variables and living preferences as the dependent variable. Standard errors are clustered on the respondent level.
Overall, elections and economic prosperity have the strongest effects. Specifically, economic prosperity – compared with economic hardship – affects the preference-to-live outcome by 15 percentage points, whereas free and fair elections – compared with not having free or fair elections – affect the outcome by 16 percentage points. Second, the political features (elections, checks and balances, and free speech) more consistently affect the outcome than the security and diversity features. The difference between the effect of the economy and crime – which does not yield any significant effect – is particularly noticeable.Footnote 18
However, citizens are slightly less responsive to civil liberties (free speech versus absence of free speech) and much less responsive to horizontal accountability (constraints on the executive versus no constraints) than to economic well-being. Thus, responsiveness to checks and balances in particular is overshadowed by responsiveness to the economy, which we interpret as signaling that people prioritize the citizen-centered aspects of democracy (free speech and elections) to a larger extent than institutional aspects (executive constraints). We thus confirm authoritarian contracts, as people are more responsive to economic well-being than non-electoral aspects of democracy, which provides partial support for our first hypothesis.
Regarding executive constraints, our findings – together with the manipulation check results – suggest that citizens may not be unresponsive to this feature because they reject it as democratic, but rather because they do not perceive it as central to democracy. Nonetheless, regardless of subjective interpretation, our data show that citizens are objectively unresponsive to this attribute in their preferences.
To further test Hypothesis 1, Figure 4 presents marginal means on the preference-to-live outcome across all combinations of economic prosperity and the democracy scale, based on a factorial interaction between the two dimensions (as described in the Modeling section). The figure confirms that citizens are highly responsive to economic prosperity – more so than to the addition of a single democratic feature (one point on the 0–3 democracy scale). This is visible in the figure, where vertical shifts from hardship to prosperity produce larger changes in preferences than horizontal shifts across the democracy scale. However, respondents are, on average, more responsive to the addition of two democratic features, which typically yield a 25 percentage point increase in the outcome – exceeding the effect of economic prosperity.
Figure 4. Living preferences for high (low) democracy and low (high) security country bundles.
Note: 282,560 observations. Linear regression with our democracy scale and economic prosperity as independent variables, including an interaction term between the two, and living preferences as the dependent variable. Standard errors are clustered on the respondent level.
As shown in Figure 3, responsiveness to democratic features is also highly uneven: elections elicit stronger responses than civil liberties, and especially more than executive constraints. Taken together, Figure 4 provides partial support for Hypothesis 1 – citizens are clearly responsive to economic prosperity, but a combination of democratic features can outweigh that effect.Footnote 19
As discussed above, a key caveat is that we may not have fully disentangled democratic features from economic performance. Respondents might, for instance, associate country profiles that have free and fair elections with better economic conditions, which could help explain some of the observed effects of elections. At the same time, the effect of elections may also stem from a genuine ideological commitment to elections, or from expectations about other desirable policy outcomes that are seen as following from democratic processes. We leave it to future research to examine more precisely why and how free and fair elections resonate with citizens.
Subgroup Differences
Thus far, our results play perfectly into the hands of contemporary electoral autocrats, as citizens are responsive to what the typical electoral autocrat claims to provide: elections and (economic) prosperity. But is this the case for all citizens? We now turn to the hypotheses about heterogeneous effects by individual- and macro-level well-being, regime type, and support for democracy. We report subgroup AMCEs for people who do and do not struggle economically (on the macro-level as well as individually) in Figure 5, for people living in autocracies, new democracies (1–2 years since democratization), intermediate democracies (22–43 years since democratization), and old democracies (more than 100 years since democratization) in Figure 6, and for people with low and high support for democracy in Figure 7.Footnote 20
For H2–H5 to gain support, we should see that democracy is prioritized less relative to security when economic struggle is present (H2), in autocracies (H3a), in new or intermediate democracies compared to old democracies (H3b), and when support for democracy is low (H4). In addition, we should expect that democracy is prioritized less relative to cultural security for individuals who have socially and culturally traditional values (H5). In the presentation of the results, all attributes are included for completeness. Additionally, we benchmark subgroup responsiveness to individual country features against the general responsiveness of the subgroup as explained in the Modeling section. This helps us assess whether heterogeneous effects reflect general differences in responsiveness or genuine differences in responsiveness regarding the specific feature in question.
Figure 5. Average marginal component effects of all attributes on living preferences by individual- and macro-level economic conditions.
Note: 111,608–170,952 observations depending on the size of economic conditions categories. Linear regression with country characteristics as independent variables and living preferences as the dependent variable, conditioning on our measures of individual- and macro-level economic conditions. Standard errors are clustered on the respondent level.
Figure 5 shows that people who do not struggle economically are more responsive to shifts in country features in general, but the relative priority of the economy versus democracy does not depend on personal economic hardship. The subgroup differences are similarly general – and if anything, weaker – across macro-level security. We therefore reject the hypothesis that relative prioritization of security versus democracy depends on economic hardship – whether on the micro or macro level (H2).Footnote 21, Footnote 22
Figure 6. Average marginal component effects of all attributes on living preferences by regime.
Note: 8,818–150,242 observations depending on the size of regime categories. Linear regression with country characteristics as independent variables and living preferences as the dependent variable, conditioning on our measures of regimes. Standard errors are clustered on the respondent level.
Turning to regime differences, Figure 6 shows that the old democracies – the US, the UK, and Australia – stand out by producing larger effects on the democracy factors.Footnote 23 The differences on the democracy factors between the old democracies and the rest (approximately 10 percentage points) substantively exceed general differences in responsiveness (approximately 5 percentage points). In contrast, differences in responsiveness to economic prosperity are similar to differences in general responsiveness.
Figure 7. Average marginal component effects of all attributes on living preferences by individuals’ pre-existing democracy support.
Note: 76,248–206,312 observations depending on the size of democracy support categories. Linear regression with country characteristics as independent variables and living preferences as the dependent variable, conditioning our measures of support for democracy. Standard errors are clustered on the respondent level.
Citizens living in autocracies are not more responsive to security relative to democracy than citizens in new or intermediate democracies. Rather, they tend to be less responsive to shifts in country features in general. There are interesting differences within democratic features, as citizens in autocracies tend to prioritize free speech less, whereas citizens of intermediate democracies prioritize executive constraints less. In sum, we do not find support for the hypothesis that citizens living in democratic contexts convincingly prioritize democracy more relative to security (H3a), whereas we do find support for the hypothesis that such a trend materializes with the age of democracy, especially among very old democracies (H3b).
Next, we find support for the hypothesis that citizens with higher support for democracy prioritize living in a democracy relative to security more than citizens with low support for democracy (H4). Specifically, Figure 7 shows that the effects of the democracy attributes are stronger among citizens with high support for democracy than among citizens with low support, whereas we find no differences concerning economic prosperity. Subgroup differences in prioritization of executive constraints should be interpreted with some caution, as they do not exceed general differences in responsiveness, but they are nevertheless larger than the differences related to economic prosperity.
These findings by no means strike us as obvious, as recent studies suggest that support for democracy is a poor predictor – as bad as flipping a coin – of punishment of undemocratic behavior by politicians (see, for example, Svolik Reference Svolik2019). Despite recent criticism, our findings thus lend some credibility to measures of support for democracy and show that such measures are good predictors of citizens’ preferences of where to live.
Additionally, it is noticeable that support for democracy yields particularly large heterogeneity concerning the effects of free and fair elections. This suggests that citizens mostly have elections in mind when they say they support democracy. Importantly, citizens with high support for democracy do not value economic prosperity, low crime, gender equality, or ethnic diversity more than people with low support for democracy. This suggests that citizens do not have these aspects – which are less central to procedural democracy – in mind when saying they support democracy. We take this – which is largely consistent with our findings regarding understandings of democracy in Figure 2 – as evidence that citizens have a solid understanding of what democracy is and that the expression of general support for democracy (our moderator) might not only describe ‘democrats by name only’ (Wuttke et al. Reference Wuttke, Gavras and Schoen2022, 416).
Looking at Figure 8, we reject the hypothesis that individuals with socially and culturally traditional values (referred to as ‘conservative values’ in the figure) should prioritize conservative cultural country traits over democracy. We find that liberal citizens do value gender equality and ethnic diversity more than conservative citizens, but even conservative citizens view these traits somewhat positively. Additionally, even though liberal citizens value democracy somewhat more than conservative citizens, the effects of the democratic attributes also outweigh the effects of the cultural attributes among conservative citizens. The results imply that despite the widely documented cultural tensions in many democratic countries, these tensions seem not to affect people’s general commitment to democratic institutions, especially elections and individual liberties, as the main priority for a country.
Figure 8. Average marginal component effects of all attributes on living preferences by individuals’ pre-existing social attitudes.
Note: 154,960–175,074 observations depending on the size of social attitude categories. Linear regression with country characteristics as independent variables and living preferences as the dependent variable, conditioning our measures of social attitudes. Standard errors are clustered on the respondent level.
Discussion and Conclusion
Across a diverse set of countries and individuals, our results show that citizens – despite some subgroup differences – value both economic prosperity and free and fair elections when envisioning an ideal society. The strong emphasis on economic prosperity aligns with authoritarian contract theory, which posits that material security often takes precedence over liberal political rights. As the theory suggests, core liberal-democratic elements such as civil rights and executive constraints tend to be deprioritized by citizens.
Yet, a somewhat unexpected finding – also observed in Adserà et al. (Reference Adserà, Arenas and Boix2023),Chu, et al. (Reference Chu and Yeung2025), and Ferrer et al. (Reference Ferrer, Hernandex, Prada and Damjan2025) – is the consistent salience of elections. Taken together with these recent experimental studies, our findings reinforce a remarkably consistent pattern: across diverse societies, citizens view free and fair elections as the defining and most valued element of democracy. Like Adserà et al. (Reference Adserà, Arenas and Boix2023),Footnote 24 we find that electoral procedures carry strong symbolic and substantive weight, while liberal-democratic safeguards – such as executive constraints and judicial independence, as shown by Ferrer et al. (Reference Ferrer, Hernandex, Prada and Damjan2025) – are more easily traded away.
This shared emphasis on elections underscores both the resilience and the limits of democratic commitment: people want democracy to the extent that it means voting, but not necessarily when it implies institutional restraint or political pluralism. The fact that citizens across such a wide range of contexts assign high importance to free and fair elections offers an optimistic lens on democratic ideals, but it also rests on two assumptions: first, that people can recognize when elections are not genuinely free or fair, and second, that they view elections as an inherently democratic, rather than potentially autocratic, feature.
On the first assumption, there is limited evidence that citizens possess a clear and accurate understanding of free and fair elections. The existing literature explores how individuals form opinions about the quality of national elections and the extent to which these perceptions are accurate (Coffé Reference Coffé2017). Overall, citizens’ evaluations are, at best, moderately accurate and subject to various cognitive and political biases.Footnote 25 For example, assessments are influenced by partisanship and whether one’s preferred party won the election (Cantú and García-Ponce Reference Cantú and García-Ponce2015; Mochtak et al. Reference Mochtak, Lesschaeve and Glaurdić2021; Schaffner Reference Schaffner2021; Ugues Jr Reference Ugues2018). These biases are further compounded by exposure to disinformation (Mauk and Grömping Reference Mauk and Grömping2023) or by living under authoritarian regimes (Jöst et al. Reference Windecker, Vergioglou and Jacob2022; Wilking Reference Wilking2011). Access to reliable information remains crucial in shaping perceptions of electoral legitimacy (Clayton and Willer Reference Clayton and Willer2023; Karp et al. Reference Karp, Nai and Norris2017; Robertson Reference Robertson2017). In sum, while many individuals may lack a detailed understanding of electoral fairness, they nonetheless associate the presence of elections with democratic governance.
The second assumption – that free and fair elections are universally recognized as a cornerstone of democracy – is further complicated by how autocratic regimes manipulate electoral norms. The symbolic power of elections is often co-opted by autocrats to legitimize authoritarian rule (Brunkert Reference Brunkert2022; Schedler Reference Schedler2013). For instance, in May 2023, Erdogan declared victory in Turkey’s presidential election amid widespread allegations of authoritarian control over the media, the judiciary, and state institutions (Esen et al.Reference Esen, Gumuscu and Yavuzyılmaz2024). Yet, he portrayed the election as a ‘festival of democracy’.Footnote 26 This tactic is not uncommon. Many autocracies claim democratic legitimacy – enshrining it in their constitutions (Márquez Reference Márquez2016) or even adopting names like the Democratic Republic of the Congo – and hold elections as frequently as democracies, with only a few exceptions (for example, Saudi Arabia, North Korea).
Modern electoral autocracies combine ‘de jure’ competitive elections with visible success in providing material welfare and public order to enhance their legitimacy (Morgenbesser Reference Morgenbesser2017; Von Soest and Grauvogel Reference von Soest and Grauvogel2018). Our findings suggest that this model resonates with public preferences. Given the lack of nuanced public understanding of electoral fairness – shaped by biases, disinformation, or regime propaganda – and the ability of authoritarian governments to influence economic perceptions through media control (Öztürk Reference Öztürk2023), autocracies can generate genuine popular support.
These results offer insight into the global pattern of democratic backsliding and authoritarian resilience observed over the past decade. Autocratic leaders often erode democracy by targeting institutional checks and balances, particularly mechanisms of executive oversight (Lührmann and Lindberg Reference Lührmann and Lindberg2019, 1098). As our findings show, such constraints are the least prioritized aspect of democracy among citizens, which may explain why many are willing to tolerate institutional weakening so long as elections appear responsive and economic conditions are favorable.
In conclusion, the revealed priorities across different democratic components may help explain how democratic decline has occurred with public acquiescence in many countries. Citizens are not only tolerant of democratic violations by elected leaders, as previous research has shown, but may actively favor the hybrid political systems emerging in autocratizing states. Finally, the widespread willingness to reduce the salience of executive constraints in favor of economic gains – even in long-established democracies – highlights the enduring appeal of electoral autocracy as a globally resonant model of governance.
An important direction for future research concerns the link between security and democratic support. Our findings diverge from those of Chu et al. (Reference Chu and Yeung2025), who identify physical safety as the main condition under which citizens are willing to sacrifice democracy. By contrast, our crime dimension – measuring everyday insecurity – elicits far weaker responsiveness. This discrepancy may reflect contextual and design differences: our broader global sample includes high-crime societies where insecurity is normalized, and our framing of crime as a persistent social problem may not evoke the acute sense of personal threat emphasized by Chu and colleagues. Future work should examine how distinct forms of insecurity – from chronic crime to existential threats – affect democratic preferences across political and cultural contexts, and when demands for order and stability begin to erode democratic commitments.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/S0007123425101129.
Data availability statement
Replication data for this article can be found in Harvard Dataverse at: https://doi.org/10.7910/DVN/MFAWCE.
Acknowledgments
A previous draft of the paper was presented at the EPSA (June 2023), APSA (Sep 2023) Annual Conference, and VI Jornada de Comportament i Opinió Pública de Catalunya (Nov 2023). We thank the participants of these panels for their valuable feedback, and also thank Julian Schuessler for his methodological guidance and Souleymane Yameogo for his research assistance.
Financial support
This research was generously funded by the ERC Consolidator Grant ‘Democracy under Threat: How Education Can Save it’ (DEMED) (grant number: 865305, PI - Anja Neundorf) and the Norwegian Research Council’s Young Research Talent project ‘When will citizens defend democracy? (DefDem)’ (grant number: 315257, PI - Sirianne Dahlum).
Competing interests
The authors do not declare any competing interests.
Ethical standards
This research relies on the contribution of human participants and adheres to the APSA Principles and Guidance for Human Subjects Research. The study was peer-reviewed and received ethical approval (number: 400210195) from the University of Glasgow’s Ethics Board before any data collection commenced.
Open access
