1 Introduction

Despite high levels of public support for liberal democracy in polls (e.g., Hernández 2016; Inglehart 2003; Norris 2011; Zilinsky 2019), political scientists have recently documented a crisis of democracy around the world (e.g., Coppedge 2017; Diamond 2020; Foa and Mounk 2016; Lührmann and Lindberg 2019). Nowadays, liberal democratic decline is marked by the gradual deconsolidation of political pluralism, civil liberties, and the rule of law rather than by sweeping coups (Bermeo 2016; Diamond 2020; Maeda 2010). Democratically elected leaders like Viktor Orbán in Hungary and Recep Erdoğan in Turkey use the power vested in them by the people to dismantle and instrumentalize important liberal democratic institutions and the free press (Huq and Ginsburg 2018). This deconsolidation from within is possible because voters support, or at least refrain from punishing, their representatives’ antidemocratic and illiberal behavior (Carey et al. 2022; Luo and Przeworski 2022). This suggests that some people support liberal and democratic norms at the abstract level but do not act accordingly in day-to-day politics.

Ideally, democratic support is conceived as independent from specific political contexts. In the classical understanding (Easton 1965, 1975), regime support—in contrast to specific support for political authorities—is conceived as diffuse and largely unresponsive to changes in day-to-day politics, strongly shaping people’s choices involving regime principles. However, recent research on individuals’ norm support and their willingness to vote for candidates not adhering to liberal and democratic norms challenges this assumption (Carey et al. 2022; Fossati et al. 2022; Graham and Svolik 2020; Lewandowsky and Jankowski 2023; Saikkonen and Christensen 2022), especially in contexts with increased political polarization (Arbatli and Rosenberg 2021; Svolik 2019, 2020; van der Brug et al. 2021) and during crises such as climate change, the COVID-19 pandemic, or economic shocks (Amat et al. 2020; Arceneaux et al. 2020; Ballard-Rosa et al. 2021, 2022; Marbach et al. 2021).

This research points to discrepancies between abstract support and support for the practical application of these norms across different contexts, raising the question of whether citizens are really that principled when it comes to their support for liberal and democratic norms. We investigated whether and how even generic references to broader political issues influence expressions of norm support by evoking trade-off–specific cues, which enable individuals to evaluate the costs and benefits of specific liberal and democratic norms for their political agendas. Depending on their evaluation, people may uphold principles that benefit their agenda but reject norms that impede the implementation of their positions, irrespective of their abstract support for the same principles. Given that many evaluations of democratic stability rely on questions about abstract norm support (cf. European Social Survey European Research Infrastructure 2023), existing empirical results may not be very informative regarding people’s support for the practical application of liberal and democratic norms in real-life settings.

To test the expectation that expressions of support for liberal and democratic norms are affected by trade-off–specific cues, we used data from two online surveys querying German citizens’ attitudes on climate change and the COVID-19 pandemic alongside their support for selected norms in the abstract and against the backdrop of these issues (henceforth called contextualized support). Complementing recent experimental research from Graham and Svolik (2020), Carey et al. (2022), Saikkonen and Christensen (2022), and others, we find considerable differences in individuals’ abstract support for liberal and democratic norms and for the application of these norms when trade-off–specific cues are present. The more concerned individuals are about climate change, and the more precautions against COVID-19 they take, the lower their expressed support for these principles being applied to people on the other side of these issues. Hence, even generic references to political issues prompt people to express different levels of norm support, suggesting that abstract measures substantially overestimate people’s dedication to democratic and liberal principles.

2 Theorizing Support for Liberal and Democratic Norms in Real-World Settings

People who support liberal and democratic principles in the abstract will usually strive to uphold them—at least in theory (cf. Quintelier and Van Deth 2014). In reality, however, goal conflicts may arise between democratic norms and substantive goals. Accordingly, people likely have to make trade-offs in their evaluations of liberal and democratic norms, suggesting that contextualized support for these norms will equal abstract support only if trade-off–specific features do not matter (cf. Kingzette et al. 2021). This could be the case if people always decided in line with their abstract principles or if the goals expected to be attained by transgressing these norms were not important to people; however, this seems unlikely given that many political decisions severely affect people’s interests or deep-seated convictions, such as moral intuitions, personal values, and national or ethnic identity (e.g., Armingeon and Bürgisser 2021; Graham et al. 2009). Hence, individuals confronted with such trade-offs may justify transgressions of liberal and democratic norms to achieve their policy objectives.

Supporting this argument, recent research shows that despite high levels of abstract democratic support, voters willingly ignore breaches of democratic norms in exchange for policy congruence (e.g., Carey et al. 2022; Graham and Svolik 2020; Lewandowsky and Jankowski 2023; Saikkonen and Christensen 2022). Strikingly, individuals subconsciously rationalize breaches of liberal and democratic norms within their political in-group as democratic behavior, while condemning similar behavior from opposing politicians (Krishnarajan 2023). This provides further evidence for a discrepancy between abstract and contextualized support for liberal and democratic norms. Existing research also shows that people’s preferences regarding direct democratic procedures are shaped by trade-off–specific cues (Landwehr and Harms 2020; Steiner and Landwehr 2023; Werner 2020), suggesting that large parts of the electorate evaluate procedures based on their expected outcomes (Landwehr et al. 2017; Landwehr and Harms 2020). The mechanism we assume has thus already been documented for other types of democratic attitudes.

Yet not everybody will respond to trade-offs between liberal and democratic principles and other valued goals in the same way. For people whose support of liberal and democratic principles is based on strong convictions, these principles should always prevail over competing goals. The reactions observed in existing literature may be more likely when abstract norm support is weak or not anchored by reasoning about substantive principles, or when competing goals are perceived as sufficiently important to override these principles. In this case, asking about abstract support in isolation will lead to very different evaluations than questions including trade-off–specific cues (cf. Zaller 1992; Zaller and Feldman 1992). As people want to see issues tackled in a specific way, support or opposition to certain principles may seem beneficial depending on their issue positions. Considering the different evaluations of (un)democratic behavior of in- and outgroup members (Krishnarajan 2023) and the relevance of trade-off–specific cues for the support of direct democratic procedures (Landwehr and Harms 2020; Werner 2020), people can be expected to uphold norms for one issue while rejecting the very same norms for another issue. Hence, we hypothesize that expressions of support for liberal and democratic principles in the context of political issues are influenced by individuals’ predispositions toward these issues, as the included cue allows most people to ascertain their potential gains and losses from supporting a norm. By implication, we expect that contextualized support is not fully determined by abstract support for the same principles.

This mechanism should be enhanced during crises such as climate change and the COVID-19 pandemic, which can only be resolved collectively and need near-universal adherence to one solution, impeding their resolution with democratic means (cf. Howe 2017). In light of the urgency of these two crises, the inertia inherent in democratic processes may persuade people who feel strongly about these issues that breaches of liberal and democratic norms are justified to avoid life-threatening ramifications (cf. Mittiga 2021). For climate change, this urgency manifests in the rising number of extreme weather events around the world (cf. Eckstein et al. 2021) and the popularity of protest movements such as Fridays for Future (e.g., Sommer and Haunss 2020; Thackeray et al. 2020), which both contribute to an increasing awareness of the impacts of climate change in Western democracies (Flynn et al. 2021).

The COVID-19 pandemic involves an even sharper trade-off between public health and liberal and democratic principles (Amat et al. 2020; Engler et al. 2021). To contain the pandemic, many countries resorted to restricting civil rights such as the freedom of movement and assembly. To enable their governments to react quickly to the pandemic situation, parliaments handed over vast executive power, although countries with strong democratic institutions were more hesitant to breach democratic norms to contain the pandemic (Engler et al. 2021). On the individual level, lockdown policies generally gave rise to authoritarian attitudes (Marbach et al. 2021). The pandemic also led to a sharp increase in support for technocratic government and prompted people to relinquish civil liberties more readily than in the context of climate change or terrorism (Amat et al. 2020).

3 Research Design

To explore differences between abstract and contextualized support for liberal and democratic norms and to test our hypothesis, we used two separate online surveys querying German citizens’ attitudes in the context of climate change and the COVID-19 pandemic. Germany is a highly stable democracy (see V‑Dem 2023, cited as Coppedge et al. 2023; Pemstein et al. 2023) in which citizens routinely express high levels of support for liberal democracy (e.g., Inglehart 2003; Wuttke et al. 2020), suggesting resilience to antidemocratic tendencies. Therefore, diverging levels of abstract and contextualized support in Germany would indicate even more severe misperceptions of democratic support in other cases.

The first survey was fielded in spring 2020 and focused on climate change attitudes. In 2019, Germany experienced a sharp rise in climate change protests, and the issue dominated the political discourse (Statista 2019). Although the first wave of the COVID-19 pandemic hit before the collection period and would soon become the defining political topic, climate change was still highly salient at the time of the survey (Umweltbundesamt 2020).

The second survey (Navarrete et al. 2022) was collected in fall 2020 and queried German citizens’ attitudes about the COVID-19 pandemic (see Table 1 for detailed information on both surveys). At the time, Germany had just entered a new stage of containment policies, mandating the closing of bars and restaurants and limiting personal contacts and movement (Bundesministerium für Gesundheit 2023). Although protests against the containment policies had started to gain traction prior to the collection time (Hippert and Saul 2021), public support for the government remained high (Wagschal et al. 2020). In contrast to the close cooperation of most parliamentary parties to contain the pandemic, the far right Alternative for Germany party (AfD) increasingly opposed the containment policies and even allowed protesters access to the Bundestag, where they harassed members of parliament and cabinet members (Hippert and Saul 2021).

Table 1 Overview of survey data used
Full size table

Both surveys used samples drawn from online access panels based on representative quotas for sex, age, and education. Although quotas cannot prevent the realized sample from being biased toward internet users, the proposed mechanism is expected to work independently of respondents’ online affinity. Even if this assumption failed, the validity of the results should not be affected, as we focus on intraindividual rather than interindividual differences.

To assess how often support for liberal and democratic norms conflicts with issue-specific positions, we first examined how many respondents have strong preferences about these norms as well as climate change or the COVID-19 pandemic. We measured abstract norm support with a battery of items querying respondents’ agreement with the statements in Table 2 on a five-point scale from “strongly disagree” to “strongly agree.” For the analyses, we rescaled the responses to range between 0 and 1 and reversed items where higher values indicated support for norm transgressions for consistency (see Online Appendix 1 for question wording and coding).

Table 2 Items measuring support for liberal and democratic norms
Full size table

Rather than aiming to cover the full range of liberal and democratic norms, we presented a diverse set of measures capturing support for the current liberal democratic system in Germany. The items listed in Table 2 speak to core liberal democratic principles such as the division of powers, the protection and political inclusion of minorities, the freedom of speech, and equality before the courts, which are routinely used to measure the quality of democracy (see, e.g., Freedom House Index [Freedom House 2023] and V‑Dem [Coppedge et al. 2023; Pemstein et al. 2023]). Less common in research on democratic stability, the consideration of group interests captures fair political competition, which is linked to minority protection as interest groups mediate between citizens’ interests and the political systems (Binderkrantz 2020; Schlozman et al. 2012). Lastly, support for violence in political conflict provides a warning sign for democratic deconsolidation (Levitsky and Ziblatt 2018; Saikkonen and Christensen 2022).

Individual predispositions toward climate change were measured with an item asking respondents how worried they were about climate change, with answers ranging from 1 (“not worried at all”) to 5 (“extremely worried”). We relied on respondents’ concern about climate change instead of their belief in climate change or their positions on more specific climate policies because concern offers an indication of respondents’ general stance toward climate action independent of their support for specific climate policies, which tend to come with considerable trade-offs (Armingeon and Bürgisser 2021). However, when we measured respondents’ stances on climate change as the mean of their positions on four mitigation and three adaptation items, the effects were substantively unchanged and even more pronounced (Online Appendix 3).

Unfortunately, an analogous measure for respondents’ concern about the COVID-19 pandemic was unavailable. We therefore used a battery of questions about COVID-19 prevention measures to construct an additive index counting the number of measures respondents had taken in the last week. Measures included actions such as washing one’s hands longer and more often, wearing a mask even where not required, or avoiding personal contacts, indicating that respondents had gone beyond the current directives to prevent infections. Because more measures taken suggest greater worry for themselves or their loved ones, the constructed index should reasonably capture concern about the pandemic (see Online Appendix 1 for question wording and coding).

Figure 1 displays the percentages of respondents who value both the selected norms and are concerned about the respective crisis, those who either value the norm or are concerned, and those who value neither. Conflicts may arise in the first group if the trade-off–specific cue indicates that norm support is detrimental to the implementation of people’s issue preferences. Depending on the norm, between 3% (minority protection) and 55% (freedom of speech) of respondents experience conflicts between policy goals and norm support. On average, every fourth respondent thus has reason to compromise their norm support in order to strengthen their preferred policy agenda.

Fig. 1
figure 1

Potential for conflict between liberal and democratic norms and issue positions. Depicted are the percentages of respondents who value the respective norm and are concerned about climate change or the COVID-19 pandemic (conflicted), those who either value the norm or are concerned (decided), and those who value neither (indifferent). Respondents are assumed to value the respective norm or be concerned if their answer is above the midpoint of the scale. Source: Survey 1 for climate change, survey 2 for the COVID-19 pandemic

Full size image

To test whether trade-off–specific cues affect support for liberal and democratic norms, we considered the differences between respondents’ abstract and their contextualized norm support. Contextualized support was measured with a battery of questions that were worded very similar to the items for abstract support but that relate to a specific political issue.Footnote 1 For instance, the cued item on parliamentary control reads as follows: “To effectively combat [climate change | the spread of pandemics], the government’s capacity to act must not be limited by the parliament.” (See Online Appendix 1 for the wording of all items.) The framing of the cued items clearly favored one position, making it easy for respondents to judge whether protecting the norm would hurt or advance their agenda. Abstract and contextualized support were queried in the same survey, precluding the possibility that shifts in contextualized support were based on unobserved shifts in abstract support. In both surveys, respondents were asked to report their abstract norm support before answering any questions on the substantive issue, and they saw at least 15 other screens between the batteries on abstract and contextualized support.

We used regression analysis to examine to what degree citizens’ contextualized norm support was determined by their abstract support and their positions on cued issues. To this end, we included respondents’ abstract support for the respective norm as well as their positions on climate change and the COVID-19 pandemic as independent variables in an ordinary least squares (OLS) regression. Analogous to a lagged dependent variable (e.g., Keele and Kelly 2006), the inclusion of respondents’ support for the abstract norm as an independent variable controls for the influence of factors that affect both abstract and contextualized support. In consequence, the effects of the remaining independent variables relate to the variance in contextualized support that cannot be explained by the determinants of abstract support. Thus, the model only needs to include independent variables that explain contextualized but not abstract support for liberal and democratic norms. By implication, standard controls such as sociodemographic characteristics or the extremity of ideological positions are not relevant here, as they influence abstract as well as contextualized support (e.g., Kingzette et al. 2021; Kokkonen and Linde 2023).Footnote 2,Footnote 3 Our main explanatory variables are respondents’ predispositions on climate change and the COVID-19 pandemic, which we expected to influence people’s contextualized but not their abstract norm support. To test the robustness of the results, we additionally estimated the influence of people’s issue positions on the difference between their abstract and contextualized support (Online Appendix 5), again with substantively unchanged results.

4 Results

To confirm that people’s abstract support for liberal and democratic norms differs from their contextualized support for these norms, we first considered the mean support for norms with and without trade-off–specific cues. Figure 2 depicts respondents’ average norm support in the abstract and in the context of climate change (squares) or the COVID-19 pandemic (circles). On average, respondents are more willing to limit legal recourse and curtail the freedom of speech when trade-off–specific cues imply that upholding these norms would adversely affect the agenda pursued by the majority of the respondents in the sample. Although the difference is considerably smaller, respondents are even significantly more accepting of political violence in the fight against climate change and the COVID-19 pandemic, indicating that the most fundamental cornerstones of liberal democratic rule are questioned when they impede people’s political agendas.

Fig. 2
figure 2

Average support for liberal and democratic norms by issue. Depicted are mean levels of abstract (black) and contextualized (gray) support with 95% confidence intervals. Issues are climate change (squares) and the COVID-19 pandemic (circles). Source: Survey 1 for climate change, survey 2 for the COVID-19 pandemic

Full size image

The picture is less clear for parliamentary oversight, as the difference just reaches statistical significance in the expected direction for the COVID-19 pandemic but is reversed for climate change. Similarly, support for the influence of interest groups remains unaffected in both scenarios. Yet respondents are more supportive of protecting the interests of individuals over general public interests in the context of COVID-19. In line with our larger argument, norms benefitting minorities or the opposition, which comprise their own trade-off–specific cues, received comparatively low support even in the abstract. In contrast, norms generally benefitting all citizens—namely freedom of speech, the legal process, and the monopoly of violence—received higher levels of abstract support. Adding trade-off–specific cues then produced the greatest effects for the latter, i.e., for norms without inherent positional information. In short, citizens’ support for liberal and democratic norms appears to vary depending on how norms impact their political agendas.

We tested the hypothesis that individuals’ contextualized support for liberal and democratic norms is influenced by their issue positions using linear regression models explaining respondents’ contextualized norm support as a function of their abstract support and their positions on climate change and the COVID-19 pandemic. Figure 3 shows that, while abstract support is positively related to contextualized support, this link is surprisingly weak considering that the measures queried the same concept, only adding trade-off–specific cues. In line with the descriptive findings, respondents who are very concerned about climate change are willing to compromise all considered liberal and democratic principles to advance climate protection. Accordingly, concerned respondents’ contextualized support is between five (legal process) and 25 (parliamentary control) percentage points lower than their abstract support, strengthening the notion that abstract and contextualized support differ systematically and substantively.

Fig. 3
figure 3

Effects on support for liberal and democratic norms in the context of climate change. Depicted are linear coefficients with 95% confidence intervals (see climate change models in Online Appendix 2 for full regression results). Source: Survey 1

Full size image

Abstract norm support is linked slightly more strongly to contextualized support queried against the backdrop of the COVID-19 pandemic (Fig. 4), although the effects remain small given the inherent link between the two measures. Respondents who take the maximum number of COVID-19 prevention measures are less supportive of six out of seven principles, expressing between 11 (minority protection) and 38 (interest groups) percentage points less contextualized norm support compared to respondents who take no prevention measures.

Fig. 4
figure 4

Effects on support for liberal and democratic norms in the context of the COVID-19 pandemic. Depicted are linear coefficients with 95% confidence intervals (see COVID-19 models in Online Appendix 2 for full regression results). Source: Survey 2

Full size image

5 Conclusion

Using climate change and the COVID-19 pandemic as examples, this research was designed to study whether individuals’ support for the application of liberal and democratic norms to specific issues differs from their abstract support for the same norms, as well as how these differences are informed by individuals’ issue positions. We found that abstract norm support differs substantively from contextualized support and is stronger for principles that benefit all citizens, as opposed to clearly defined groups such as minorities. Specifically, we show that individuals’ political incentives matter for their expressions of support when applied to climate change and the COVID-19 pandemic. Individuals who are more concerned about climate change are less willing to uphold liberal and democratic norms when they are in the way of climate protection; the same relationship can be observed for people taking more precautions against COVID-19 and upholding norms that impede action against the COVID-19 pandemic. These findings challenge the informative value of abstract norm support, as people are either unwilling or unable to translate their abstract support into principled decisions. Considering that the application of norms rather than abstract support is decisive for democratic (de)consolidation, these results are concerning.

Expanding the emerging literature on democratic support (e.g., Carey et al. 2022; Fossati et al. 2022; Graham and Svolik 2020; Lewandowsky and Jankowski 2023; Saikkonen and Christensen 2022), we demonstrate that political incentives not only prevent citizens from punishing norm violations by candidates and parties but also significantly lower their expressions of support for these norms. These effects can be read in different ways: People may have real preferences for regime principles, which are updated when learning about additional information and are thus subject to adaptive political evaluations. Alternatively, abstract norm support could be stable but not well elaborated, making it hard for people to consistently apply their abstract convictions in specific settings (Wuttke et al. 2023). The available evidence cannot tell us which reading more accurately describes reality, and both accounts may apply at least to some people.

Several starting points for future research arise from the limitations of the available data. Due to the framing of the survey questions, we could not test whether our assumptions also hold if we flip the target of the norm violation, i.e., if the norms were indicated to benefit the opposite group. Moreover, because norm support was queried only for one issue per survey, we cannot compare how trade-off–specific cues related to different political issues to influence the support for individual respondents. It would be especially relevant to compare the effect for individuals whose incentives are reversed for different issues. In addition, the wording of the abstract and contextualized items varied in some cases. Most important, the abstract interest group item suggested that these groups harm the public welfare, whereas the cued version did not reference harm, resulting in a slightly different connotation. Although the direction of these differences suggests that we underestimated the real effects, we cannot preclude that deviations in question wording influenced some results. Lastly, newly collected surveys should measure respondents’ own perceptions regarding their membership in the majority or minority to allow for more fine-grained analyses of opportunity structures.