Who looks up to the Leviathan? Ideology, political trust, and support for restrictive state interventions in times of crisis

The extent in which voters from different ideological viewpoints support state interventions to curb crises remains an outstanding conundrum, marred by conflicting evidence. In this article, we test two possible ways out from such puzzle. The role of ideology to explain support for state interventions, we argue, could be (i) conditional upon the ideological nature of the crisis itself (e.g., whether the crisis relates to conservation vs. post-materialist values), or (ii) unfolding indirectly, by moderating the role played by political trust. We present evidence from a conjoint experiment fielded in 2022 on a representative sample of 1,000 Italian citizens, in which respondents were asked whether they support specific governmental interventions to curb a crisis, described under different conditions (e.g., type of crisis, severity). Our results show that the type of crisis matters marginally – right-wing respondents were more likely to support state interventions only in the case of terrorism. More fundamentally, political trust affects the probability to support state interventions, but only for right-wing citizens.


Introduction
Who supports restrictive governmental interventions during crises?This question has been at the forefront of the mediatic and scholarly attention since the global outbreak of the COVID-19 pandemic in early 2020.The unfolding of the health and social crisis has generated a wealth of academic and public discussions about government-enforced limitations to personal freedoms, as well as debates about the boundaries between the private and public spheres (e.g., Bjørnskov and Voigt, 2022).Yet, the current debate about individual freedoms versus social responsibility and state interventions for the greater good is neither novel, nor specifically related to these types of health crises.
Crises as diverse as climate change, terrorism, economic catastrophes, and wars oftentimes tend to be associated with governmental policies intended to curb their nefarious consequences.Regardless of the effectiveness of these interventions, and the reasons why some leaders might be more likely than others to promote an interventionist approach (e.g., Medeiros et al., 2022), much attention has been granted in the existing literature to whether such state interventions unfold as trade-offs between social welfare and individual rightsand to ethical considerations related to freedoms and the limits of democracy (Dahl, 1989;Sen, 2005;Posner, 2006).
When it comes to voters, a case likely can be made that different individuals, in terms of, say, personal traits or political preferences, react in different ways to specific limitations (Collis et al., 2022).While research has shown the presence of generalized effectse.g., lockdowns enforced during the pandemic increased trust in democracy and government across Europe in the shortterm (Bol et al., 2021), and a large majority of citizens support the limitation of individual freedoms during crisespatterns of approval can indeed differ greatly across subgroup categories (Hartmann et al., 2022).For instance, Terry et al. (2020) report that younger individuals were more likely to showcase a negative mood toward the pandemic at the beginning of the crisis, Auton and Sturman (2022) demonstrate that greater compliance with COVID-19 restrictions was more likely among more informed respondents, and Modersitzki and colleagues (2021) show that respondents high in extraversion and neuroticism were more likely to perceive measures as more restrictive.Overall, individual differences, including deep psychological constructs like personality traits, seem to matter for individual responses to crises according to multidimensional patterns.
Less clear, mostly due to contradictory extant results, is the role played by a key attitudinal dispositionthe ideological preferences of citizenfor individual support for state intervention.If some evidence exists that partisan considerations had less of a role to play than expected when it comes to the persuasiveness of crisis-related public health messages (Gadarian et al., 2021b), ideological considerations tend to be strong predictors of attitudes toward politics.Yet, the specific role of ideology 1 to shape individual support for restrictive governmental measures remains an outstanding questionand, more specifically, whether compliance to injunctions is more often found amongst progressives or conservatives.On the one hand, evidence exists that conservatives showcase more often compliance with governmental policies (e.g., Sullivan et al. 1982;McClosky and Brill, 1983).For instance, attitudes toward pandemic-tracking technologies were seen more favorably among voters high in right-wing authoritarianism and moral conservatism (Wnuk et al., 2020), and conservatives were more likely than liberals to accept civil liberty restrictions to ensure personal security after 9/11 (Davis and Silver, 2004).On the other hand, conservatives were less concerned with the spread of the COVID-19 pandemic (Conway et al., 2021), and have been substantially less keen to engage in individual behaviors to curb the pandemic (Gadarian et al., 2021a).Clarke et al. (2021) show that, in Australia, right-wing individuals and those high in antiegalitarianism and conventionalism tended to react more negatively to restrictions to stop COVID-19.Similarly, conservatives are typically strongly less likely to endorse restrictive measures to curb climate change, contrary to liberals (McCright and Dunlap, 2011).All in all, whether a conservative or progressive ideology leads toward greater compliance to state interventions during crises remains, surprisingly, an open question.While the differences in responses on the political spectrum may depend on various contextual factorsincluding the nature and framing of the crisisthe lack of consistent behavioral patterns seems somewhat at odds with our current understanding of the fundamental driving role of partisanship and ideology for political behaviors.In this article we tackle this key puzzle by testing two competing narratives: the first is related to the nature of the crisis at hand, and the second to the possible impact of ideology on the role played by political trust.
First, it could be the nature of the crisis itselfand, in particular, whether the crisis itself is traditionally framed in progressive or conservative termsthat alters the effects of respondents' ideology, in such a way that stronger support for state interventions exists when the crisis is framed in terms that match the ideological profile of the respondent (e.g., the threat of terrorism for right-wing voters).
Second, the impact of ideology on support for state interventions could be a more indirect one, materializing itself in moderating the effect of one of the most widely discussed predictors of 1 We conceptualize ideology as a specific type of belief system, in line with the classical studies in political and electoral behavior (e.g., Converse, 1964).In this sense, we treat ideology as a configuration of attitudes or ideas that are bound together in a form of interdependence.
support for state interventions: political trust. 2 As far as political trust is important as an explanans of support for state interventionsas it reduces the perception of the risk associated with granting extra-power to the state to deal with a crisisthis effect should be more marked among those citizens who need to be assured the most.As we will discuss, we theoretically expect this to be true in particular among citizens with a right-leaning position.
We investigate these two possible narratives by leveraging novel evidence from a conjoint experiment fielded in Italy in early 2022 on a representative sample of 1,000 citizens.Respondents were asked whether they support specific governmental interventions to curb three different types of crises (a pandemic, an environmental, and a terrorist one), the characteristics of the crisis (type, severity, shared measures) being manipulated in a conjoint setting.
Italy stood out as a potentially insightful case study with its long history of internal terrorism, as well as being one of the most affected European countries by the recent pandemic and by climate change.Moreover, the peculiar government of national unity at the time of the survey made Italy, as we will discuss, a sort of quasi-experimental setting to study the effect of political ideology.Our results show that the type of crisis matters marginallyright-wing respondents were more likely to support state interventions to curb terrorism.More importantly, we show that ideology, as we suspected, mostly matters in an indirect way.In particular, we show that political trust increases the probability to support state interventions, but only for right-wing citizens, hence providing clarity to the (moderating) role played by ideology in explaining support for state interventions during crises.

Ideology, trust, and support for state interventions
The existing literature has produced mixed, often contrasting results about the role of political ideology in shaping patterns of public support for state interventions during crises.Focusing on the standard progressive-conservative divide, 3 scholars argued that citizens with conservative stances are more likely to show compliance with government policies and to construe personal rights as contingent rather than absolute (Sullivan et al., 1982;McClosky and Brill, 1983).
The mechanisms underlining this different perspective on risks and individual freedoms have been long investigated by political psychology.For instance, Jost and co-authors (2003) argue that people tend to adopt political conservative stances to manage uncertainty and threat.Here, conservatives are more responsive than progressives to external and internal threats.In particular, US conservatives appear to be more prone than progressives to accept restrictions on civil liberty to enhance personal security after the terrorist attacks of 9/11 (Davis and Silver, 2004), and significantly more likely to support the use of torture against suspected terrorists (Zugravu et al., 2023).Variations in responses to negative stimuli tend to be correlated with political preferences (Hibbing et al., 2014: 299), and indeed political conservatives have been shown to be more sensitive to threatening stimuli, in particular for physical threats (Crawford, 2017).According, for example, to the motivated social cognition (MSC) perspective (Jost et al., 2003), both existential 2 We refer here to vertical trust (i.e., trust in institutions) and not to horizontal trust (i.e., trust in others), although both interpretations can be correlated (Putnam, 1993).The literature on political trust is vast, pinpointing its crucial implications for both political participation (Hooghe and Marien, 2013) and law-abiding behavior (Marien and Hooghe, 2011). 3 In Western Europe, the 'progressive-conservative' dichotomy (Middendorp, 1978) marked the structure of the political space, where 'progressive' (i.e., the left side) came to indicate support for economic equality and cultural pluralism, and 'conservative' (i.e., the right side) would be associated to the aims of economic freedom and cultural uniformity (Bobbio, 1994).This unidimensional left-right simplification of the political space (Fuchs and Klingemann, 1990;Knutsen, 1995) is seen as an instrument that citizens can use to orient themselves in a complex political word.Similarly, the United States is dominated by the liberal-conservative ideological dimension, especially in the last decades, characterized by a growing partisan polarization (Webster and Abramowitz, 2017).Based on their nature as means for orientation, the left-right (or progressiveconservative) and the liberal-conservative dimensions can be seen as functional equivalents (Fuchs and Klingemann, 1990).We adopt this approach here.motives for threat management (e.g., loss aversion, fearfulness) and epistemic motives for uncertainty management (e.g., avoidance of uncertainty, needs for closure) are associated with political conservatism.Similarly, the negativity bias (NB) perspective posits that conservatives are especially responsive to negative stimuli and events (Hibbing et al., 2014).In this regard, 'the strongest evidence for the MSC and NB perspectives comes from studies in which threat is operationalized as perceived or actual physical harm or danger and when conservatism is operationalized using measures of social rather than economic political positions or identification' (Crawford, 2017: 356).
Regarding COVID-19, Wnuk and co-authors (2020) show that attitudes toward pandemic-tracking technologies were seen more favorably among people with right-wing authoritarian views and high moral conservatism.Yet, contrasting evidence exists as well: Conway and colleagues ( 2021) have shown that US conservatives have been consistently less preoccupied by the spread of COVID-19 than progressives, and less willing to accept restrictions.In Italy, Ladini and Maggini (2023) demonstrated that right-wing voters are less likely to accept limitations to freedom.Similarly, and regarding climate change, left-wing citizens in Western Europe reported stronger support for action to mitigate climate change (McCright et al., 2016), resembling the same polarization on this issue between Democrats and Republicans in the USA (McCright and Dunlap, 2011).
The main take-home point, as we see it, is that there is no clear link between ideology and acceptance of state interventions during crises.How to make sense of these inconsistent results?In this article, we test two competing expectations: first, that the effect of ideology is contingent on the nature of the crisis; second, that its effect largely materializes indirectly, as a moderating factor on the impact of a fundamental political attitude, political trust.
First, the lack of consistent results with regard to the impact of political ideology on support for restrictive measures to curb crises might simply come from the fact that crisesand the associated threats that potentially go with themare usually not presented in ideologically neutral terms.Consistent evidence exists that crises are 'politicized' and framed along ideological terms through epistemic construction mechanisms (e.g., Krzyżanowska and Krzyżanowski, 2018;Voltolini et al., 2020;Hutter and Kriesi, 2022).Also as a result of these rooted processes of politicization, crises tend to follow narratives that are intrinsically built on ideological elementsto the point that their mere existence, and the threat they pose, is not consensually accepted across the political divide.
For instance, threats to individual safety linked with crime or terrorism are construed as a crisis especially from a conservative standpoint (e.g., Loader, 2020), likely due to the centrality of 'conservation' values (including the need to be safe and secure; Schwartz, 2012) for right-wing conservatives (Jones et al., 2018).Inversely, heightened perceptions of the risk inherent in the climate crisis seem associated with left-wing values.Indeed, contemporary progressivism is increasingly characterized by the defence of post-materialist values, such as multiculturalism, gender equality, and, most relevant here, environmental protection (Ford and Jennings, 2020).Therefore, climate change represents a serious threat to these values.Several studies on climate change confirmed this assumption (Hindman, 2009;Häkkinen and Akrami, 2014;Veenstra et al., 2014): individuals on the ideological left tend to attribute climate change to human activity and be worried about it to a much larger extent than individuals on the ideological rightindependently from education.
Progressives and conservatives are equally likely to support punishment against violations of core values (Wetherell et al., 2013), but they likely differ in which violations ought to be punished.The different support for different public policies by progressives and conservatives is of course a well-documented phenomenon, and we simply extend here a rather trivial mechanismprogressives and conservatives support different public policiesby applying it to differential support for different threats, above and beyond individual preferences for state intervention as a whole.
As we discuss in the methodological section, a third type of crisisnot directly and explicitly framed in left/right ideological termswill serve as a control of sort: a pandemic crisis.Of course, as the COVID-19 crisis has certified, a pandemic crisis cannot be understood necessarily as politically neutral in all contexts.The inconsistency of the findings about the relationship between ideology and acceptance of restrictive measures in such instance might in fact reveal different frames in different countries, depending on the ideology of the party in government.However, this has a twofold implication: (1) there has not been a direct, explicit, and consistent politicization of the recent pandemic crisis in left/right ideological terms that is context-independent; (2) the government-opposition explanation, rather than the genuine ideological explanation, appears to largely account for the different political framing of the COVID-19 crisis across contexts (with partisanship influencing support for restrictive measuressee Arceneaux et al., 2020).As previously mentioned, the peculiar government of national unity at the time of the survey, conducted in Italy, should downplay government-opposition differences.Above and beyond these matters, the pandemic crisis itself was likely less intrinsically ideologically framed than terrorism (intrinsically related to matters of security and conservation) and climate change (intrinsically associated with post-materialist values such as environmentalism).We thus have Second, the lack of consistent results with regard to the impact of political ideology on support for restrictive measures to curb crises could indicate that the effect of ideology is not a direct one.More specifically, we argue here that the role of political ideology could manifest itself as a moderating factor on the impact played by one of the most important determinants of political attitudes, namely political trust.
Contrary to the role played by ideology, the empirical evidence connecting trust to support for state interventions during crisis is strong.A wealth of literature has in fact shown that trust in political institutions plays a crucial role in shaping under what conditions citizens are willing to comply with measures enforced by their governments (see 'institutional theory', e.g., Baumol and Blinder 2008).Importantly, when large-scale crises threaten public and private security, people who hold a solid trust in the institutions that respond to such threats are consistently more likely to comply with public curbing policies.A textbook case, in this sense, are the measures enforced by the USA after the terrorist attacks on 9/11: the lower the trust in the US Government and President, the higher the resistance from Americans opposed to limitations of their freedom to enforce anti-terrorism policies (Davis and Silver, 2004).
Recently, scholars assessed as well the legitimacy of medical and non-medical measures to curb the spread of infectious diseases, like the viruses Ebola and H1N1, confirming this relation between institutional trust and the likelihood of accepting restrictions (Tang and Wong, 2003;Prati et al., 2011;Vinck et al., 2019).The same applies to COVID-19, as this empirical regularity has been confirmed at both the aggregate (Bargain and Aminjonov, 2020; Barrios and Hochberg, 2021) and individual levels (Ladini and Maggini, 2023), though individual-level studies have sometimes shown mixed results (Dohle et al., 2020;Jørgensen et al., 2021).One key mechanism to explain this relation are the 'cascades of confidence' during the first phases of new crises, where people rally around their government and fellow citizens to respond to exceptional times (Guglielmi et al., 2020).On the other hand, lack of trust is associated with scepticism toward government interventions even during crises (e.g., Eberl et al., 2021;Casiraghi and Bordignon, 2023).
Scholars unveiled similar dynamics on climate change.Cologna and Siegrist (2020) demonstrated that trust in institutions is correlated with support for climate change mitigation public measures, but only weakly with private climate-friendly behaviors.Smith and Mayer (2018) found that trust and risk perceptions are generally positively associated with public willingness to support policies addressing climate change.Similarly, Fairbrother and colleagues (2019) argued that people who are more supportive of higher taxes on fossil fuels are not more aware or concerned about climate change.Rather, political trust, hence more than climate-specific attitudes, is a better predictor of citizens' likelihood of finding restrictions legitimate.Indeed, the pivotal role of trust in shaping patterns of approval for pro-environment policies was already noted more than a decade ago (Konisky et al., 2008).
Still, both trust and ideology are fundamental heuristics that citizens rely on to build their positions on political issues, especially in exceptional times.Indeed, and crucially for our explanation, right-wing citizens have been generally associated with being more risk-averse (Jost et al., 2003;Crawford, 2017).
Studies have also shown that risk aversion typically increases during large-scale crises that threaten personal and collective security, such as the financial crisis after 2007 (Guiso, 2012).Given that conservatives usually react more negatively than progressives to threats, they become arguably more risk-averse during major crises.Here it should be stressed that risks might be connected not only with the consequences of the crisis (and shaped by its ideological framing, as postulated in H1), but also with its management by public authorities.Trust arguably plays a key role in influencing how citizens relate and react to such momentous decisions.According to Hetherington's theory (2005), the activation of the trust heuristic helps citizens to sacrifice their own self-interest for the sake of others and is connected with the perceived risk tied to a particular policy.
Relying on this theory, other scholars (Rudolph and Evans, 2005;Rudolph, 2009) broadened the concept of sacrifice to include considerations of ideological interest, showing that the effect of political trust is significantly more pronounced among citizens who pay a higher ideological cost for supporting specific policies, for instance conservatives as regards government spending (Rudolph and Evans, 2005).Similarly, we argue that during major crises, when public authorities enforce limitations to individual freedoms for the greater good, the cost of this sacrifice could be therefore cognitively higher for right-wing people given that conservatism is traditionally associated with individualism at least since the '80s (Gray, 1990).
Given however that political trust, as we have highlighted above, reduces risk aversion, we expect that the ideology of citizens can affect the role played by the former aspect.In particular, and bringing to its logical conclusion what was just noted above, we expect that ideology should moderate the relationship between trust and the acceptance of restrictive measures.As a result of that, the role played by political trust in these instances should be magnified where it is needed the most, that is, among right-wing individuals.
We thus have H2.Individuals high in political trust are more likely to accept restrictive measures during crises, especially if they are right-wing.

Setting and data
We test our hypotheses via a conjoint online survey experiment administered by the polling company Demetra to a representative sample of Italian citizens in January 2022 (N = 1,000)see online Appendix C for a description of the main socio-demographic features of the sample.A conjoint design was selected as it allows us to assess the independent effects of different attributesin our case, the multidimensional characteristics of an exogenous crisison respondents' preferences through a fully randomized vignette (Hainmueller et al., 2014).Such a design is furthermore particularly indicated to investigate potential variations in the effects of such attributes across different subgroupssuch as trust or ideology.
In light of our expectations, our experiment focuses on three different crises: (i) a crisis traditionally framed in conservative terms (the chance of a terrorist attack), a crisis traditionally framed in progressive and post-materialist terms (climate crisis), and a new pandemic (ideologically not intrinsically relatable to conservative or progressive values).All three crises have a global scope, that is, likely have implications affecting countries on the whole and thus reasonably requiring a governmental response or intervention.
Italy is a particularly well-suited case to investigate such dynamics.Among Western countries, Italy was hit early and hard by COVID-19, and experienced heated discussions about the effectiveness and legitimacy of the exceptional measures implemented by the government.Similarly, both climate change and terrorism are consistently debated issues in the country.The latter is linked to the history of far-right and far-left domestic terrorism in Italy during the 'Years of Lead' in the seventies and eighties, coupled with the recent wave of Islamic and far-right attacks in Europe, whereas the former is frequently associated with recent waves of natural disasters across the countryaccording to the head of the Civil Protection Department, 'all of Italy is at risk', with upwards of 94% of all municipalities under the threat of landslides, flooding, and coastal erosion (Giordano, 2022).
Additionally, the timing of the survey is particularly important because since February 2021, Italy had a national government cabinet led by Mario Draghi.Among major parties, only the right-wing Brothers of Italy (Fratelli d'Italia, FdI) did not participate in the government. 4As such, Italy represented a quasi-experimental setting, whereby we can (almost) neutralize the impact of supporting or not the government and how this affected citizens' attitudes toward freedom limitations during a global crisis.This is not irrelevant to assess the 'net' impact of ideology on the support for restrictive measures or otherwise.Arceneaux et al. (2020) conducted, in fact, a series of experiments in the USA and UK during the first wave of the COVID-19 pandemic, finding that citizens are more likely to accept restrictive measures if they are championed by politicians of their preferred political party.However, in our case, (almost) all major parties were in government.As a result, any detected impact of the variable ideology happens net of the possibility that respondents' own party is actually in the cabinet.In addition, the experiment was conducted a considerable amount of time after the first phases of the pandemic crisis.In this way we are testing our hypothesis in a conservative framework, far from the first 'cascades of confidence' that typically characterize the early outburst of a crisis.

Design and measures
In our experiment, respondents were presented with two vignettes, side by side, each presenting a different crisis scenario.Each respondent was administered two pairs of the vignettes.Vignettes manipulated the type of the crisis (pandemic, terrorism, climate change), its severity (low, high), and whether restrictions were imposed in other EU countries (see Table 1).In addition, we also varied the nature of the measures implemented: (i) control measures, namely enhancing state control on citizens through CCTVs in public or controlling private messages on social media; (ii) restrictive measures, namely limitations to freedom of movement through, for instance, lockdowns or reducing car or internet usage; and (iii) punitive measures, namely the sanctioning and fining of specific behaviors, typically not respecting the exceptional restrictions enforced.
The specific level that attributes can assume was randomized, and all levels were independent and had equal probabilities (Bansak et al., 2021).To increase external validity we decided to frame the two conjoint profiles as a report by a group of independent experts that describe the crisis and recommend certain measures.In contrast to typical conjoint profiles, which employ tables with precise data or numbers, we provided respondents with a relatively short textual vignette.The final vignette is therefore a news feature similar to those that people read on social media, which respondents should be rather familiar with.Citizens typically read news and inform their preferences through this type of short articles, rather than comparing different precise data in tables (Dafoe et al., 2018).Indeed, as argued by established studies, tables constitute a poor analogue of real-life decision-making (e.g., Lau and Redlawsk, 2001).
After seeing the vignette, respondents were asked to evaluate each profile on how 'legitimate' they thought the suggested measures werefrom 0, not legitimate at all, to 10, perfectly legitimate (a measure then rescaled as a 0 to 1 variable).As a result, we are implementing what is called in the literature a 'rating-based conjoint analysis' (Hainmueller et al., 2014: 57), given that the evaluation performed by the respondents of the vignette is a rating rather than a simple dummy variable (0 and 1).Before the vignette, we also asked various questions to measure subgroup preferences, in particular how much respondents trusted the Italian Parliamenta 0-10 proxy for political trustas well as respondents' political self-placement on a standard left-right scale (from 0, left, to 10, right).With respect to the former variable, we rescaled it to 1 for any value higher than 5 and 0 otherwise.Overall, 37% of respondents appear to have a value of 1 for the political trust variable.This percentage is in line with other analyses and surveys5 and corresponds to the typical operationalization adopted in the literature when trust is estimated along a continuous scaleas it is in our case.We have also explored different operationalizations of the political trust variable (i.e., rescaling the trust in the Italian Parliament variable to 1 for any value higher than its mean value in our database -4.3or higher than 6that corresponds to the third quartile; and employing threelow, intermediate, and high trustrather than two categories).All our results remain unaffected (see online Appendix B).
With respect to the ideological variable, we recoded the self-placement scale into 3 categories, namely 5 for the center (25.4% of the total sample), 0-4 for the left (38.2%), and 6-10 for the right (36.4%).Also in this case, our results are robust to different operationalizations of the ideological variable (see the discussion below and Appendix B).

Attribute Levels
Crisis type Terrorism: ' : : : on the risk of terrorist attacks performed by national or international actors' Climate change: ' : : : on climate change and its effects on the weather and natural disasters such as hurricanes, floods, or droughts' Pandemic: ' : : : on the potential appearance of new global viral or bacterial pandemics' Crisis severity Medium: ' : : : slightly alarming, in line with what we expected' High: ' : : : extremely alarming, much more dramatic than what we could expect' Measures: Type Punishments: ' : : : a substantial increase of punitive measure on misbehavioursthrough, for instance, the creation of a public list that determines a limited access to public services, mobility or access to financial services for individuals who were responsible of dangerous behaviours (online or offline)' Control: ' : : : a substantial increase of measures to control the populationthrough, for instance, the placement of CCTV cameras in public spaces or the control of text messages on social media' Restrictions: ' : : : a substantial increase of restrictions to the freedom of movement of the populationthrough, for instance, measures to reduce mobility among cities or during busy hours, limitations to the access to public or crowded places, to acceptable behaviour in private life, or restrictions to the use of internet or social media' Measures: Diffusion Not shared: ' : : : unique for now in Europe' Shared: ' : : : already implemented in many other European countries'

Results and discussion
Concerning result estimations, we computed the marginal means (MM), a method particularly suited to compare subgroup effects in a conjoint framework (Leeper et al., 2020). 6All the results reported below employ survey weights constructed by employing a ranking method, whereby cell counts are adjusted so that the marginal totals match the control totals.Note that our results remain substantially the same if we replicate the analysis without any weights.
Figure 1 shows the results of the conjoint analysis across the entire sample.The most important result here is that the MM score of all attributes is never significantly distinguished from the overall mean in the ratings (see the vertical line at 0.472 in the Figure ).This implies that respondents were generally quite sceptical of the proposed measures, regardless of the specific characteristics of the crisis.Across the attributes, the only significant difference is that, unsurprisingly, crises that are more severe convince, more consistently, citizens to accept restrictions.
Our two hypotheses, however, relate to the possible existence of systematic differences in the evaluations reported in Fig. 1 according to some specific subgroup preferences/attributes.First, following H1, we expect different levels of support for state interventions according to the type of crises across the ideological spectrum of the respondents.In particular, we expect that left-wing respondents are more likely to support state interventions in the case of climate change, whereas right-wing respondents are more likely to support state interventions in the case of terrorism.
Figure 2 presents results that only partially support H1.In particular, in the left panel of Fig. 2 we report the conditional MM of vignette's features on perceived legitimacy of measures when contrasting respondents with different ideological positions.In the right panel instead, we report the difference in such conditional MM between respondents who self-placed themselves in the ideological center (left) and those who self-placed themselves in the ideological right.As can be seen, right-wing respondents are significantly more prone than their left-wing homologues to support state interventions when the crisis concerns a terrorist threat (legitimacy among left-wing respondents: 44.1%; among right-wing respondents: 51%; difference: −6.9, p-value: .01),whereas no significant differences emerge for climate change (i.e., our 'progressive' crisis) and new pandemics (the 'control-group' crisis).Note that the significance of the difference with respect to terrorism between left-wing and right-wing is robust to alternative specifications of the ideological categories (see Appendix B).
Hence, and enriching the results of previous studies, it seems that ideology has some effect on the case of terrorism, which is consistent across the last decades.However, political ideology per-se does not directly influence the likelihood of accepting restrictive measures enforced during exceptional times for more salient, recent crises such as pandemics and climate change.In these cases, whether the crisis is framed in progressive and conservative terms only partially matters, thus not fully confirming our first hypothesis.
We now move to H2.In this respect, we first present results regarding the direct impact of political trust on the likelihood of accepting state-enforced curbing measures during crises.Confirming the large empirical literature in this regard (see Fig. 3), across almost all our conjoint attributes, individuals high in trust find measures consistently more legitimate than their lesstrusting fellows.This significant effect is absent only in the case of terrorism (even when we move to a less stringent 90% confidence interval), likely because, as said above, it is a less salient crisis in Italy nowadays compared to the other types of crises.According to our second hypothesis, however, we should expect the impact of political trust not being uniformly the same across the ideological positions of the respondents.In particular, the former should be magnified wherein it is supposedly 'needed' the most (i.e., among rightist respondents).To test this, we first created two distinct groups of respondents, namely rightist (according to our previously discussed operationalization: 36.4% of the total sample) or otherwise.
Note there is no systematic difference in how respondents in our two groups (right v. otherwise) have been exposed to our profile attributes and levels (see Figs. 4a and 5a in the Appendix).This is a welcome result, as experimental subgroups must be properly balanced in a sample to provide reliable estimates within a randomized conjoint design.The implication is that the differences we report below are not simply due to a distinct exposure of respondents to the conjoint attributes, for example that right-wing respondents were exposed to vignette types consistently different in terms of the severity of the crisis or its type.Conversely, such differences should reflect actual, specific characteristics of the subgroups.
Figure 4 shows the effect of high vs. low trust on support for state-enforced restrictions across the various attributes, distinguishing between the values of our moderating variable, i.e., rightwing respondents (left panel) and otherwise (right panel).In this respect, the results illustrate how political trust is a significant predictor of the likelihood of finding state-enforced measures during crises legitimate across all attributes, but overall in a statistically significant way solely for rightwing respondents.Hence, we find an indirect role of political ideology, which moderates the effect of political trust: since right inclining citizens are more risk-averse when it comes to state interventions, as previous literature has argued, they need the role of trust to find exceptional restrictive measures acceptable more than their left-wing homologues.

Conclusions
Previous literature produced contrasting results on how ideology influences public perceptions on the legitimacy of exceptional measures.With this in mind, our investigation aimed at shedding a new light on this process, by looking at the role of the specific situation (type of crisis) and underlying levels of individual trust.Our results show that ideology has a direct effect solely in the case of terrorism, whereby right-wing individuals are more consistently prone to support state interventions to curb the crisis.On the other hand, and in the more salient cases of climate change and pandemics, ideology has no direct effect per-se.We tested whether such effect could be indirect, testing that on the uncontroversial role political trust plays in this context.Our key result here is that ideology matters in moderating the effect of trust on the likelihood of accepting restrictions during crises.In particular, political trust is significant as a variable only for right-wing individuals who, being traditionally more risk-averse, need the role of trust to find exceptional restrictive measures acceptable more than their left-wing homologues.Rather than having a direct effect on support for state-enforced measures during crises, hence, ideology has an indirect effect by moderating the impact of political trust.
Our findings provide engaging insights for the currently significantly relevant literature on public attitudes during crises, shedding light on previous controversial results.Indeed, previous research provided clear evidence regarding the positive effect of political trust on compliancerelated attitudes and behaviors but provided few insights about how this effect varies across individuals characterized by different ideological orientations.We hypothesized and explained the moderating role of ideology on theoretical grounds, stressing the connection between perceived threats during large-scale crises and ideological beliefs.In particular, we built upon previous studies according to which political conservatism is usually associated with risk aversion (Jost et al., 2003;Crawford, 2017).Hence, the positive effect of political trust on the perceived legitimacy of state-enforced measures during crises is not homogenous within the population given that individuals differ in risk aversion and such cognitive biases underpin differences in political ideology.
More broadly, this research can talk to both political science and political psychology, underlying the possible linkage between values, political beliefs, cognitive biases, and personality traits on the one hand and attitudes toward public policies on the other.Interestingly, our results resemble those of other studies showing that the effects of political trust on support for fiscal policies are moderated by ideology (Rudolph and Evans 2005;Rudolph, 2009).In particular, the theoretical expectations of these studies are derived from Hetherington's (2005) sacrifice-based theory of political trust, which posits that political trust gets activated when citizens are asked to sacrifice their own self-interest for the sake of others.When the concept of sacrifice is broadened to include considerations of ideological interest, evidence shows that the effect of political trust is significantly more pronounced among citizens who pay a higher ideological cost for supporting these policies, namely conservatives as regards government spending (Rudolph and Evans, 2005) and liberals as for tax cutting (Rudolph, 2009).The mechanism is similar to that argued in our study: political trust is especially relevant for right-wing citizens who need it the most to reduce the perception of the risk associated with threat management by public authorities during crises.State-enforced restrictions, indeed, impose different cognitive burdens on progressives and conservatives, as well as fiscal policies do not impose equal ideological costs on people of different ideological orientations.
Besides its academic contributions, this article could provide fresh insights for policy-makers in need to justify state restrictions during crises.Our results provide novel ways of understanding how to leverage political trust and political ideology to defend the idea that measures are needed and legitimate.More importantly, our results should encourage civil activists and NGOs that monitor government action and public support during crises to more closely focus on the role of ideology and trust in their campaigns, for instance targeting specific groups when trying to diffuse awareness about the effects of state-enforced restrictions.
Finally, our findings can have significant and ambivalent political implications.On the one hand, we know that in democracies cooperation and compliance with public regulations require citizens' positive attitudes (Van den Bos et al., 1998;Zmerli and Newton, 2008).By positively affecting compliance-related attitudes (Song et al., 2018), political trust is thus an essential resource in times of crisis.On the other, the heterogenous effect of political trust among subgroups of different ideological orientations implies that declining trust may erode public support for governmental measures to curb crises both in general and especially among certain segments of the electorate.This could be particularly problematic for the effectiveness of crisis management whether both ideological polarization is high and political trust strongly varies according to citizens' ideological orientations.
To be sure, our article is limited in terms of context, historical time, and the focus on specific crises.Future studies could replicate our survey analysis in other countries and time periods to see whether the same patterns are valid in different contexts.In addition, scholars could focus on other crises (e.g., war, financial crashes, immigration, ethnic protests, democratic backsliding) and other characteristics of the crises (e.g., length, origin) to test whether ideology moderates trust in the same way to explain public support.Finally, new analyses could check whether other subgroup features (e.g., specific political attitudes, cognitive biases, emotions) do interact with ideology in shaping how citizens react to state-enforced restrictions.
Assuming that state interventions to curb crises exist to address violations of core valuesfor instance, limiting freedom of movement of selected individuals to ensure the personal safety of the publicthen support for specific state interventions should be driven jointly by the nature of the interventions (i.e., the crisis) and the values to which individuals adhere.Progressives should thus intuitively be more likely to support state interventions to curb threats to post-materialist values, and conservatives should be more likely to support state interventions to curb threats to conservation values.
H1. Left-wing respondents are more likely to support state interventions to curb threats to postmaterialist values (climate change), whereas right-wing respondents are more likely to support state interventions to curb threats to conservation values (terrorism).

Figure 1 .
Figure 1.Effects of attributes on the perceived legitimacy of measures (n = 1,000).

Figure 2 .
Figure 2. Conditional marginal means (left panel) and differences in conditional marginal means (right panel) of vignette's features on perceived legitimacy of measures, by the ideological position of the respondents (n left in our sample: 382; n center: 254; n right: 364).

Figure 3 .
Figure 3. Conditional marginal means (left panel) and differences in conditional marginal means (right panel) of vignette's features on perceived legitimacy of measures, by political trust (n low political trust in our sample: 631; n high political trust: 369).

Figure 4 .
Figure 4. Difference in conditional marginal means of vignette's features on perceived legitimacy of measures, by political trustideological right set (left panel) vs. ideological not right set (right panel) (left panel: n low political trust in our sample: 223; n high political trust: 141; right panel: n low political trust in our sample: 408; n high political trust: 228).