Public support for ‘ soft ’ versus ‘ hard ’ public policies : Review of the evidence Research

his article reviews the literature studying the public support for ‘soft’ versus ‘hard’ behaviour change policies, and what factors drive such preferences. Policy instruments, the precise measures by which governments pursue their policy goals, vary in their governance principles and behavioural assumptions (Hood, 1986; Schneider & Ingram, 1990). Fundamentally, policy instruments can be differentiated based on whether they direct and mandate people to behave in certain ways, or indirectly encourage them to (Hood, 2007). Policy instruments can be understood on a spectrum of increasing government intrusiveness, as visualised in the Nuffield Intervention Ladder for public health policies (Nuffield, 2007). In this framework, policy measures range from “doing nothing”, to providing information for the public, to guiding people’s choices first through behavioural interventions and then financial incentives, up to the highest levels of intervention which restrict and eliminate personal choices.” Following on from this, we consider ‘hard’ policy instruments as those that restrict choice through laws, T I n tr o d u ct i o Abstract: This article reviews the literature on public support for ‘soft’ versus ‘hard’ policy instruments for behaviour change, and the factors that drive such preferences. Soft policies typically include ‘moral suasion’ and educational campaigns, and more recently behavioural public policy approaches like nudges. Hard policy instruments, such as laws and taxes, restrict choices and alter financial incentives. In contrast to the public support evidenced for hard policy instruments during COVID-19, prior academic literature pointed to support for softer policy instruments. We investigate and synthesise the evidence on when people prefer one type of policy instrument over another. Drawing on multi-disciplinary evidence, we identify perceived effectiveness, trust, personal experience and self-interest as important determinants of policy instrument preferences, along with broader factors including the choice and country context. We further identify various gaps in our understanding that informs and organise a future research agenda around three themes. Specifically, we propose new directions for research on what drives public support for hard versus soft behavioural public policies, highlighting the value of investigating the role of individual versus contextual factors (especially the role of behavioural biases); how preferences evolve over time; and whether and how preferences spillovers across different policy domains.

While this evidence is crucial to understand fast-changing evolving public opinions -especially during crises periods -the data is based on opinion polls and questionnaire surveys. Most studies, moreover, ask binary questions (yes/no; support/oppose) that may miss subtle differences in individual preferences, and may not fully represent the spectrum of public opinion. Notwithstanding these limitations, however, the overall picture this emerging body of findings is one of governments having been out of step with public appetite for hard policies in the early stages of the COVID-19 pandemic.
Indeed, these policy opinions during the pandemic has prompted us to look again at the question of what shapes the public's preferences for policy instruments (see Table 1 and Figure 1) in contrast to scholarly evidence before the pandemic which found that people tend to prefer softer instruments like nudges (Sunstein et al, 2018). This begs immediate questions -what drives the recent support for hard policies? Can they be easily explained by existing theories? Has COVID-19 prompted a deeper shift in citizens' underlying preferences for

Figure 1 Public Support for Harder Measures
Notes: *Support for imposing self-quarantine measures; **Support for self-quarantine measures. Authors' calculations drawing on data from Bhatia (2021) for India, Freitas (2020) for United Kingdom, Redfield & Wilton Strategies (2020) for France, Spain and Italy, Infogram (2020) for Singapore and Ipsos (2020) for China. Singapore* Support for extending Lockdown measures for at least two weeks or more Banerjee et al., 2021 4 types of health policy in the longer run? What about policies in related domains like the environment and climate change? Recent reviews in the public policy literature have considered the 'knowns and unknowns' of policy instrument choices Capano & Lippi, 2017), highlighting several issues including government capacity, implementation and routinisation of instrument choice. For instance, recent articles responding to policy making during the pandemic are concerned with 'national policy styles' as a means to understanding policy instruments:  set out to explore variations in the ways states have responded; Dunlop et al (2020) identify policy design and instruments as one of seven research themes for the COVID-19 era to investigate the convergence (or not) of national policy responses to the pandemic. Our focus, in contrast, is on whether there is public support for types of policy instruments -specifically soft and hard policies -and what factors may drive these preferences.
We make two contributions to the scholarly literature. We provide the first literature review of contemporary interdisciplinary research on public support for hard versus soft policy instruments. We synthesise these findings into individual and broader factors that explain the public's preferences (section 2) 2 . Our second contribution is to organise gaps in our knowledge into a research agenda (section 3), identifying research questions and arguing for greater methodological diversity to build on the limitations identified in the literature.

Review of Public Support for Hard vs. Soft Policy Instruments: Lessons and Gaps
We conducted a review of the literature, focusing on factors affecting public preferences for policy instruments, and how these preferences might change 3 . We used Scopus and Google Scholar to search for peer-reviewed articles, using the search terms 'policy preferences', 'instruments', 'tools', 'nudge', and 'public attitudes'. We focused on research published during the past five years of research and narrowed our results to those that examine public policy preferences for at least one type of policy instrument. A total of 29 articles were selected for the review through a process of agreement between two authors including snowballing. They are summarised in Table 2.
The articles were drawn from multiple disciplines including public policy and political science, psychology, and behavioural economics. They use a range of approaches including experimental, observational, and qualitative methods. Empirical applications focused largely on health and pro-environmental behaviours. The behaviours and challenges in these policy domains are especially pertinent to the COVID-19 context: the need for collective action, the potential disjuncture between communal gains and individual costs, and the problem of intertemporal costs where action involves short run costs for long run benefits.
The literature can be organised into three strands. The first strand summarises findings on public preferences, specifically, do people prefer soft or hard policy instruments? The second strand looks at individual-level factors affecting preferences for policy instruments, including socio-demographics, traits and preferences, and behavioural factors. The third strand considers broader factors, beyond individual attributes, such as the context in which individuals receive information about policies, and country-level variables.

Public Preferences for Soft Versus Hard Policy Instruments
In contrast with very recent polling data discussed earlier, which pointed to support for harder measures to tackle COVID-19, the broad lesson from the past literature is that people prefer softer instruments. Several studies by Sunstein and colleagues (2016aSunstein and colleagues ( , 2016bSunstein and colleagues ( , 2018aSunstein and colleagues ( , 2018bSunstein and colleagues ( , 2018c report that a majority of citizens unconditionally supported and approved of nudges in fifteen countries based on nationally representative samples. Recent evaluations, moreover, show that nudges can be cost-effective (Benartzi et al., 2017). High acceptability for softer policy instruments has often been compared to public resistance towards taxes or mandates (see Dubov & Phung, 2015;Goldin & Lawson, 2016).
Such support has been interpreted as a preference for policies that were less restrictive for personal freedom and choices. For instance, Diepeveen et al., find that "support was generally higher for interventions perceived as less intrusive" based on a review of 200 studies of health-related behaviours (2013, p. 4). Warning labels in health education campaigns are a classic example of a soft policy instrument since they preserve liberty while altering the context for the choice; and were found to be more acceptable than harder policy instruments such as 'sin' taxes. This finding was consistent across the four health behaviours studied -diet, physical activity, alcohol consumption and smoking. Mazzocchi and colleagues also corroborate the idea there is "on average relatively lower support for more intrusive actions like bans or taxation", based on survey about healthy eating policies from five European countries, namely, Belgium, Poland, Denmark, Italy and the UK (2014, p. 286).
Similar evidence of public preferences leaning towards soft measures are reported for pro-environmental behaviours as well. Kantenbacher et al., investigated preferences across 14 types of hard and soft aviation climate policy measures (2018, p. 47). Hard measures such as limits on flights and sanctions on cruises had the least support; the most support was recorded for soft measures such as providing information about environmental impacts and developing public transport. This is in line with qualitative research on attitudes towards tourist travel, which suggests that harder regulatory approaches are perceived as imposing on individual freedoms (Higham, Cohen, Cavaliere, Reis, & Finkler, 2016). In their study of acceptance levels and preferences across 24 energy-saving strategies in China, Jia, Xu and Fan (2018) report that those relying on voluntary behaviour change (like turning off lights in unused rooms) rather than technical solutions that involve an upfront financial cost (like installing solar panels) are the most popular (Jia et al., 2018, p. 491). Thus, the broad message from past studies on preferences for health and environmental policies is that the public prefers softer approaches.
However, people's support for soft policies can be contingent on trust in policymakers and governments, and the perceived effectiveness of the policy itself. Sunstein (2016) identifies various situations that could diminish this support. Sunstein (2016) reports that people need to trust the choice architects and will oppose nudges that promote what they perceive as illicit goals or are perceived as inconsistent with the interests or values of most choosers. For example, people do not want choice architects to produce economic or other losses by using people's inertia or inattention against them; a finding which may reduce support for defaults, as an example of a soft policy instrument. Bang, Shu and Weber (2020) also report varying levels of acceptability amongst soft policy instruments, related to their perceived effectiveness and who the choice architect was. For example, nudge policy decisions that benefit the choice architect are less acceptable than those instigated for health or sustainability reasons (Bang et al., 2020, p. 18). Although this emerging empirical research focus solely on nudges, they are nonetheless illustrative of the way people evaluate policy instruments and nuance the idea of there being a preference for softer policy instruments.

Individual-level Factors
Personal lifestyle, prior beliefs and experience, self-interest as well as concern for others, have been linked to preferences for softer policy instruments 4 . These factors have largely been measured through stated responses to questionnaires about initial attitudes and behaviours. For instance, those with pro-environmental beliefs are more likely to support all types of policy action for curbing air travel; those who are regular flyers are significantly less likely to support harder measures such as a frequent flyer tax or limits on the number of flights (Kantenbacher et al., 2018, p. 50). Individuals who have recently experienced extreme weather such as heat waves or drought "are more likely to support (climate change) laws…even if it means restricting individual freedoms" (Owen, Conover, & Videras, 2012). In a study of carbon taxes in Europe, Umit and Schaffer (2020) find support for the idea that self-interest matters, with lower support for carbon taxes amongst people with high energy dependency (Umit & Schaffer, 2020, p. 5).
Similar findings are reported in the health domain. For example, people who report high levels of personal consumption of fast food and pre-packaged food had the least support for advertising bans and fiscal measures to promote healthy eating (Mazzocchi, Cagnone, Bech-Larsen, Niedźwiedzka, Saba, Shankar, Verbeke, & Traill, 2014, pp 281-282). Diepeveen et al (2013) also report evidence consistent with self-interest theory, where people not engaging in the target behaviour "appear more willing to advocate interventions that restrict the behaviour of others" (Diepeveen et al., 2013, p. 7).
Political ideology has also been found to be significantly associated with preferences for policy instruments. In the USA, those with strong affiliations to the Democratic Party were found to be more likely to support laws to protect the environment (Owen et al., 2012, p. 564); and those with left-leaning political views were more likely to support government interventions for obesity including fiscal policy measures (Mazzocchi et al., 2014, p.280). Haselswerdt and Bartels (2015) similarly found that US conservatives prefer tax breaks to direct spending.
Age, gender and education have also been investigated, but findings are inconclusive and can depend on the policy instrument and issue in question. Diepeveen et al (2013, p. 7) find that older people and women are more likely to endorse hard policy instruments. They speculate that age may be related to trust in government and women may prefer stronger preventative measures since they are more likely to act as informal care providers. Kantenbacher et al (2018), in contrast, find no effects from gender, age or employment on aviation policy. Sunstein, Reisch and Rauber (2018) also suggest that the significance of demographic characteristics can vary with the policy instrument.
Women are in general more likely to support nudges than men, although the authors suggest this may be more to do with the particular goals of the nudges they surveyed (such as heath and pro-environmental outcomes). They report nuanced effects from education, with more education associated with higher approval for government-mandated information campaigns, but lower approval for subliminal advertising; suggesting that the more educated find information as a regulatory tool more acceptable (Sunstein et al., 2018, p. 11). Similarly, Mazzocchi et al (2014) report some significant associations between gender, age, and education and policy instrument preferences, but these are of low magnitude and vary with the policy instrument in question. For example, men are more likely to support soft policies around information for healthy eating compared to women, but there are no gender effects relating to fiscal measures or advertising bans. Sunstein (2016) warns against the notion that people have dogmatic views one way or another about the use of nudges. Rather, public support "turns on whether they approve of the purposes and effects of particular nudges" (Sunstein, 2016, p. 118). This in turn depends on what information is available to them at the time they are being asked to express their preferences. The literature has begun to explore how types of information, the way information is framed, and how the policy choices are presented, can affect support towards policy instruments. Haselswerdt and Bartels (2015) investigate support for different types of welfare policy instruments. They test the way information context affects preferences, by randomly assigning whether respondents receive information on the costs of a program. They report that having this information does lower support, corroborating the notion of self-interest affecting policy instrument preferences. They go on to highlight another contextual factor playing a role in policy instrument evaluations in the form of status quo bias, with respondents demonstrating "a preference for familiar forms of government intervention" (Haselswerdt & Bartels, 2015, p. 617). Bang, Shu and Weber (2020) also report evidence that people, when presented with repeated choices, attempt to stay consistent with their response to the first frame. These studies suggest that the evidence may reflect anchoring, status quo bias or the tendency to mitigate cognitive dissonance.

Broader Factors: Choice and Country Context
But whether these tendencies induce stability in preferences for types of policy instruments is largely unknown. Indeed, perhaps one of the most under-attended questions is how preferences for policies change over time. Diepeveen et al (2013), for instance, suggests support for a policy instrument may grow after implementation stage, since the smoking ban was considered more acceptable after it was implemented in the UK. Counterexamples exist, however, where hard policy instruments have grown less popular after their introduction such as the 'fat tax' in Denmark (Bødker, Pisinger, Toft, & Jørgensen, 2015), and the 'bedroom tax' in the UK (Gibb, 2015). The reliance on cross-sectional surveys and the lack of longitudinal research means there is little to explain to what extent and how preferences within a political system might shift as time goes on within an individual across time, for example through policy learning or feedback loops.
A number of studies confirm the idea that the framing of policy instruments -how they are presented to the public -matters for preferences. One example of framing is presenting the policy as having an effect on the wider public (the third-person perspective, referring to 'people'), or the individual (the second-person perspective, referring to 'you'). Cornwell and Krantz (2014) find that fiscal proposals garner more support when they are framed in the third-person perspective, and this applied across four policy domains covering fuel taxes, criminal penalties, savings incentives, and tax incentives to encourage community service.
Frames may affect different individuals differently, making it difficult to generalise. For example, in their study of US respondents, Jung and Meller (2016) find that more 'empathetic' respondents were more likely to favour nudges overall and responded more positively when nudges were framed as having societal effects. Conversely, participants identified as 'individualists' were unaffected by frames and more likely to be opposed to nudges overall 5 . Although this study focus on types of soft instruments, particularly nudges, rather than comparing soft and hard policy instruments, the findings may be illustrative of factors that can affect preferences for policy instruments more generally. Davidai and Shafir (2018) investigate through a series of experiments whether the presentation of softer policy instruments alongside harder policy instruments when a set of choices is presented, affects people's evaluations and preferences. Survey respondents were given information about changing defaults in order to change financial behaviours (including health insurance and retirement savings). They find that "attitudes are malleable and can be influenced by the method of evaluation whether policies are evaluated separately or jointly" (Davidai & Shafir, 2018, p. 13). The type of policy also matters, with non-deliberative nudges found to be relatively appealing when presented and evaluated separately than when they are presented alongside measures that involved more deliberation. The framing of the policy instrument with or without alternatives may be particularly important where the individual has milder preferences to begin with or knows less about the alternative policies that they might otherwise use as a benchmark for their evaluations. Similarly, Hagmann et al (2019) consider attitudes when nudges are presented jointly with harder policy instruments to address climate change and find that the availability of a green nudge reduces support for a carbon tax. These findings, too, suggest that preferences may be flexible and depend on how a policy instrument is framed, and what alternatives are included in a choice set.
Apart from the choice context, an emerging set of cross-country comparison highlight that public attitudes vary across nations, suggesting that country-level context impacts preferences (Higham et al., 2016). Asked to consider policy options to address obesity, Mazzocchi et al (2014) find that Belgian respondents supported advertising bans more than British respondents; and Danish respondents favoured fiscal measures more than British respondents. These descriptive findings are not explained in detail, although the authors assert that "policy preferences vary across cultures and over time" (Mazzocchi et al., 2014, p. 269). Hagman et al (2015) report that public opinion towards nudges was more favourable amongst Swedish respondents, with 73% finding them acceptable on average, compared to 66% of US respondents. This is attributed to the contrasting norms embodied by Sweden's welfare state and its emphasis on "collective and uniform solutions", and discourse in the US which emphasises more individualistic values and the freedom of choice (Hagman et al., 2015, p. 459).
Some studies attempt to uncover differing levels of support for nudges by groups of countries. Sunstein et al. (2018) contrast industrialised Western democracies and some BRICS nations with less nudge-enthusiastic countries comprising Denmark, Japan and Hungary. They speculate that varying levels of trust in government may account for differences in support for nudge instruments. A similar argument is put forward by Umit and Schaffer (2020) who find political trust and efficacy are positively associated with harder carbon policy measures. As neither study compares soft and hard policy measures while assessing the interaction with political trust, it is unclear whether it operates consistently in support for any government action, or whether trust is particularly important for supporting some policy instruments, in some domains. These explanations imply, more broadly, that the support for hard versus soft policies can vary by political culture. However, measures of political culture such as trust and ideology are not always comparable across studies or even within cross-country studies (Sunstein et al., 2018, p. 7).

Directions for Future Research
A review of the past evidence, especially from the health and the environmental policy domains, suggests that the public has preferred softer policy measures like nudges to harder interventions. When we examined how studies have explained what drives support for hard versus soft policies, scholars have pointed to the influence of many individual, contextual, and country-level factors. Indeed, this body of literature is still nascent, relatively sparse and contains several inconsistent findings about the role of different factors and has a number of outstanding gaps. While it employs several methodological approaches, most studies rely on stated responses to survey questionnaires and interviews.
We identify three promising avenues for future research to unpack what drives people's preferences for hard versus soft public policies (Table 2): first, probing how and when individual-level versus contextual factors matter (especially the role of behavioural biases and attitudes); second, how preferences for hard versus soft policies evolve over time; and third, if and when there are policy preference spillovers within and between policy domains, including the role of shocks like COVID-19.
First, we turn to the need for research to unpack how and when individual-level versus contextual factors matter (especially the role of economic preferences and attitudes). Our findings highlighted that people are more likely to support nudges when they are in their self-interest (e.g., it is cheaper to do so), or have prior attitudes and personal experiences congruent with the issue at hand (e.g., have pro-environmental beliefs, and if the nudge is aligned with past experiences and behaviour). There is less clarity about whether preferences for policies differ systematically by standard attributes like age, gender, income and educational background. The small set of studies modifying the choice context in which preferences are elicited show that preferences might be malleable (e.g., research on framing), but that they may also exhibit status quo bias, suggesting some resistance to change.
Some important research gaps here are the lack of empirical evidence from across countries about how behavioural biases (e.g., status quo bias, loss aversion, present bias), economic preferences (e.g., trust, social, time and risk preferences), psychological attributes and attitudes (e.g., Big-5 personality traits, fatalism), noncognitive abilities (e.g. grit, self-efficacy), and emotions or affect (e.g. fear, anger, optimism) influence preferences for hard versus soft policies. Another important gap is the need for greater clarity about if and how basic socio-demographic factors matter, given the mixed findings so far. The literature is conspicuously silent about the role of socio-economic status (which ought to consider class and wealth rather than just income) and ethnicity (such as by race, region, and caste). Apart from "trust", other country-level cultural and political dimensions which have not been studied include socio-cultural beliefs, norms, religion and values, and historical and current experiences of colonialism, war, and extreme weather and health events (e.g., in Acemoglu, Johnson, & Robinson, 2001).
One way to address this gap is to collect data from different countries about public preferences for hard and soft policies along with pre-validated indicators about individual attributes, including affect, attitudes and values, such as those identified in the Global Preferences Survey (Falk et al., 2018). The survey reveals substantial heterogeneity in preferences across countries; but even larger within-country heterogeneity, relating preferences to biogeographic and cultural variables, such as agricultural suitability, language structure, and religion, as well as from economic outcomes and behaviours.
Another avenue of research is to elicit more robust metrics on public preferences, given the reliance on primary stated responses to surveys in existing literature. This could be achieved by trialing preference valuation methods, including contingent and discrete choice approaches. These methods have long been applied to valuing policy preferences in health and environmental policy (de Bekker Grob et al., 2014;Hoyos, 2010). Such approaches could serve both to validate existing insights from stated preference studies, and to shed new insights into other metrics such as willingness to pay for or accept changes to different public policies. They could be combined with controlled experimentation to unpack how behavioural biases, like status quo bias, affect preferences for soft versus hard public policies. These can also be employed to test how framing information in certain ways may have systematic impacts on preferences for hard versus soft policies.
Second, more research is needed to explore how these public preferences evolve over time both within and across individuals and countries, along with the related question of what factors that might be responsible for changes over time. As noted by Diepeveen et al (2013, p. 8), the "stability of public attitudes…is little studied", since most research notes a range of possibilities rather than definitive conclusions based on empirical data. While there are some examples of hard policies being unpopular after their introduction (e.g., Denmark's 'fat' tax and the UK's bedroom tax), in other cases bans may have become more popular after they were introduced (e.g., the UK's smoking ban). The latter insight aligns with the emerging body of evidence documenting 'hysteresis' i.e., the tendency for policies to affect household behaviours even when they are no longer in place (Costa & Gerrard, 2018). Similarly, other studies have pointed out that nudges can have persistent effects on habitual energy behaviours after they are removed (Allcott & Rogers, 2014). It is also in line with evidence reviewed above that personal past experiences matter and can shift people's preferences towards different types of policies.
An important gap is the lack of longitudinal and repeated cross-sectional data, which can help explain how policy preferences evolve over time. Addressing this gap will enable learning on the stability of preferences for hard versus soft policies either within the individual across time, or at least at a regional level (e.g., within and across countries). Despite its limitations, there are some lessons which could be learnt from the literature studying temporal stability of time and risk preferences. For instance, Chuang and Schechter (2015) use panel data from rural Paraguay over almost a decade to find that social preference survey questions are relatively stable but note that experimental measures are not. Similarly, Meier and Sprenger (2015) present evidence from a large field study conducted over two years, with around 1,400 individuals using incentivized intertemporal choice experiments. They found that although most people's time preferences were unchanged over the study period, there were shifts amongst other individuals (which were unrelated to socio-demographic characteristics). While these studies examine individual attributes rather than policy preferences, they nevertheless point to the value of methodological approaches like longitudinal surveys that can be used to both study if policy preferences shift within the individual, but also ways to elicit the underlying mechanisms for such changes.
The question of how stable people's preferences are for hard versus soft policies is especially important to understand the long-term effects of unexpected shocks like COVID-19. Findings from our review shows that self-interest and first-hand experience can influence preferences for policy instruments, which suggests that whether people support harder (or softer) measures may depend on the extent to which they can benefit and have benefitted from such measures being in place. For instance, it is possible that an increased preference for soft policies may emerge if compliance with (hard) lockdown measures have negative psychological effects like post-traumatic stress symptoms, confusion, and anger (Brooks, Webster, Smith, Woodland, Wessely, Greenberg, & Rubin, 2020). The influence of the public's experience with lockdown and social distancing may also depend on how effective these policies were perceived to be in curbing the pandemic, how they were paired with other policy actions within the health domain (e.g., contact tracing and testing), how particular populations fared (e.g. children, non-white minorities and older populations) or even other policies in related domains (e.g. wage protections, mental health support).
Third, building on the need to investigate the temporal stability for types of policies in a given domain, future research can also examine if and when there are policy preference spillovers within and between different domains and issues. In particular, we see significant potential for work at the intersection of health and environmental policy. It is unclear whether and how people's experiences and understanding from COVID-19 might affect preferences for types of policies in other domains. For instance, the lockdown has directly affected work-related travel and commuting behaviours for some groups, which in turn has improved air quality. This direct experience could have saved time and costs for both employers and employees, which may shift preferences towards harder policy instruments that aim to influence work-related travel behaviours and environmental quality. Along these lines, Kahn et al (2020) found that residents of Chinese cities with high pollution sensitive populations, who experienced sharply improved air quality due to lockdowns during the pandemic, have increased their online discussions focused on environmental protection, and local officials are incorporating "green" industrial subsidies into post-COVID stimulus policies.
The swift and interventionist responses to COVID-19 may also enable decision action on climate change through growing preferences for harder climate and biodiversity policies like taxes and regulation. For instance, Shreedhar and Mourato (2020) found that people expressed higher support for commercial wildlife trade bans -a 'hard' conservation policy tool with wide-ranging implications -when they were informed that infectious zoonotic diseases, of which COVID-19 may be one, are linked to anthropogenic environmental change. Apart from enabling a swift reaction to other crises, experiencing the COVID-19 pandemic may lead to greater awareness, which in turn could result in policy spillovers. For instance, Rousseau and Deschacht (2020) found that the pandemic led to a positive shift in public awareness of and attention to nature and biodiversity-related topics (e.g., birds, forest, nature) in the UK. As before, studies on policy preference spillovers within and across domains could be conducted using either cross-sectional or longitudinal surveys if we want to examine how persistence spillover effects are. They can also incorporate experimental and valuation techniques as previously suggested.

Conclusions
In this article, we aimed to explore the academic literature on public support for hard versus soft policies, and what factors drive these policy preferences. Indeed, much of the extant scholarly evidence has found that people tend to prefer softer instruments like nudges (e.g., Sunstein et al., 2018). This seems to contrast with the emerging evidence from several opinion polls throughout the Covid-19 pandemic. A majority of the public -across numerous countries -have supported hard policy instruments including laws and physical restrictions in the pandemic response. Are there policy preference spillovers between different domains? Scarce and mixed evidence on the direction of behavioural spillovers.
Policy preference spillovers arising from the experience of a particular event or public policy in the same or different domain.
A systematic study of hard versus soft policy instruments in causing policy preference spillovers within and between different domains and issues; for e.g., health and environment.
In order to explain this apparent paradox, the question of what shapes the public's preferences for policy instruments is crucial. Notwithstanding the limitations of our narrative literature review, including the lack of representation across all policy instrument types and domains, we found that both individual-level attributes, as well as choice-and country-level factors, are influential. These factors offer some explanation for the marked preferences for hard policies. For instance, studies have pointed to the role of perceived effectiveness and past policy experiences (Bang, Shu, & Weber, 2020).
However, our review also highlights how narrow the current evidence base is, and numerous gaps in the literature. For instance, few studies systematically examine if emotions, like fear and risk perception, or indeed biases, attitudes, norms and other behavioural factors affect public support for hard versus soft policies (Maekelae, Reggev, Dutra, Tamayo, Sil-va-Sobrinho, Klevjer, & Pfuhl, 2020). We synthesise the existing findings and craft an agenda for future scholarship around three goals, to understand: how and when individual-level and contextual factors matter; how preferences for hard and soft policies evolve over time; and lastly, if there are policy preference spillovers between different domains.
In sum, while past research shows a tendency for the public to support softer, behavioural public policy measures, public preferences depend on a wide range of factors including trust, framing, and perceived effectiveness, spanning individual and contextual factors. These factors collectively highlight the need for subtlety in unpacking preferences for soft and hard instruments; and offer reasons why the apparent appetite for hard policy instruments during the COVID-19 pandemic may not be the paradox it first seems. More evidence is needed about how stable people's preferences to types of policy instruments are, and what factors underpin these preferences, how public support for policies evolves over time, and how it may spillover across different domains. Future research could focus on these questions to systematically study what drives people's support for hard versus soft public policies. (2020) is based on a survey of 2250 residents in the United Kingdom, jointly conducted by King's College London and Ipsos Mori. These are opinion polls and we must exercise some caution in generalising these findings over time. 2. As such this review is a more specific discussion on an aspect of instrument choice, public preferences, which relates to 'legitimacy' in instrument choice raised by Capano and Lippi (2017), and the role of policy feedback raised by . 3. While this is not a systematic review, we share our literature search and selection strategy for transparency. 4. Self-interest implies that if people try to minimize costs arising from policy, their preferences will be influenced by how the policy affects them personally. 5. Jung and Meller (2016) and other scholars distinguish between nudges that target system 1 and system 2, but for the purposes of this article it is the preference for nudges in general that is relevant. (1) Broad public support for nudges by a majority of people, varying by level of intrusiveness of the nudge policy and the type of behaviour being targeted.

The study by Duffy and Allington
(2) Age (+), males (-), political valence (populists and liberals lower than conservatives) (3) Intercountry differences found (Germany, Hungary and Denmark differ significantly from others) Shows a divergence in public attitudes and palpability towards nudges, social marketing and harder paternalistic measures. While respondents in UK, Germany and Australia favour nudges and social marketing, Norwegians prefer harder paternalistic measures through taxation over softer ones. N/A Demand for structures of provision to change, cross-subsidisation, research and adopt technologies (+), within and between sample divergence in attitudes.

DV=stated 5 Carter (2015)
Making the blue zones: neoliberalism and nudges in public health promotion No; evaluates the ideological and political origins of a specific programme and recommends using nudges.

The author analyses the Blue Zones
Programme (BZP) launched as part of its healthiest state initiative (HIS) in 2016. This BZP marked a clear shift towards neoliberal governmentality, the idea that people can "perceive, problematize and govern their own health" (p375, 2015). However, whereas the general idea of neoliberal governmentality is based on a policy neoliberalism attitudes, the BZP takes a deep neoliberalism perspective in that it embraces "technologies of the self" (p379, 2015) i.e. relies on libertarian paternalism to influence the choice of the agents by bringing in market rationality in their decision-making processes. No; reviews evidence and the theoretical basis for nudges as a tool to change health behaviour.
The authors find evidence in the favour of efficacy of nudges. They suggest that nudges should be used as complements rather than substitutes to other interventions: legislation and taxation, pertinently. Socio-economic status for antismoking campaigns (low and middle +). N/A Mixed-methods approach (including an extensive literature review which were refined through focus group testing and interviews with senior policy makers and behavioural scientists). N=N/A DV= stated

# Author (Year) Title
Hard vs soft policies?
Yes, the author finds growing evidence in favour of S2 nudges. Partisan affiliations (+) matter as democrats are more inclined to favour system 1 nudges in the contexts of green energy and water conservation while democrats and republicans prefer S2 nudges to reduce abortions Within-subjects differences noted in acceptability (at least one third prefer S2 nudges). Preferences are not always stable; for example, framing matters, as when provided information in a neutral condition people tend to prefer S2 nudges more than S1 nudges. With additional information on effectiveness, preferences shift towards S1 nudges.
Survey N= (>) 2800 Americans administered on Sampling Survey International (followed up with 400 surveys on MTurk).

DV=stated 8 Junghans, Cheung and De Ridder (2015) Under consumers' scrutiny -an investigation into consumers' attitudes and concerns about nudging in the realm of health behaviour
No; examines consumers' attitudes to nudging in the realm of health behaviour.
The authors find evidence of UK consumer's support of nudges in general and in the health domain specifically. N/A While most consumers included in the study are largely unaware of the presence of nudges, on being provided with additional information, these people are highly in favour of the nudges. While it does not matter who delivers the nudge, the uptake and the approval of the nudge depends on whether the nudge is aligned to the personal goals of the consumers and is set out with 'good' intentions.
N=21 interviewees (excluding 1 for English proficiency reasons). Broad public support for nudges by a majority of people. Males (-), age (-) as older approve less intrusive, government mandates, education (+) as high approval for governmentally mandated information nudges, political valence (-ve for conservatives over liberals). Country-specific effects presents; three clusters identified: a) Western countries (liberal democracies) where a majority support the nudges as long as they are applied in the best interest of the citizens and they align with the respondents' goals (b) Confucian Asian countries that approve of all nudges irrespective of anything else, and (c) Outliers (like Japan, Denmark and Hungary) which do not have majority approval rates for the nudges. Survey N=7927 (unweighted) respondents recruited via Qualtrics from Australia, Brazil, Canada, China, Japan, Russia, South Africa and South Korea. DV=stated 11 Sunstein (2016) Do people like nudges?
It assesses the acceptability of nudges and then compares nudges with mandates in terms of their public support.
A majority of people tend to prefer nudges in general and over mandates. The nudges are favoured as long as they are not illicit, and they match or align with the interests of the respondents. The support for nudges, contingent on its legitimacy, extends unanimously across partisan affiliations. There arises reactance to nudges and some defaults fail to work. They find that S2 nudges are more preferred than S1 nudges while informing people of nudges leads to no change in efficacy. Nudges enjoy a majority support in these countries. Age (+), gender (+), trust in institutions (+), higher formal education (-), city dwellers over villages (+), number of children (+), environmental concern (+) and belief in markets (-). Level of state intervention in countries (-), framing (+ if nudge is targeted to others than one's self), cross-country differences noted as well. Yes, within soft instrumentscompares system 1 and system 2 nudges. Support for nudges also considered. Support for nudges found except for defaults, visual illusions and one-click donations (in study 1) and defaults (in study 2). Individual dispositions matter; for example, empathy (+), individualism (-), conservativism (-), desire for control (-for S1 nudges). Perceptions matter (S1 more paternalistic over S2). Framing effects insignificant, likeability of company policies (+). Survey N=250 respondents (Study 1) and 800 respondents (Study 2) recruited via Qualtrics.

# Author (Year) Title
Hard vs soft policies? responsiveness to nudges, mandates and taxes.

Results Methods
The authors use a decision-making model to understand if policy instruments are better than one another. They find that defaults are better than mandates unless the active choosers (who decide for themselves) are making suboptimal choices for their selves. Furthermore, they also find that a combination of defaults with taxes is better than each individually and preferred to the mandates. N/A N/A Partly -survey of 2 countries but focuses on soft policy measures. Nudges classified as pro-self and pro-social.
Acceptance for nudges was generally high. Paradoxically, nudges were still seen as intrusive to freedom of choice. Pro-self nudges had higher acceptability than prosocial nudges. Individualism (-) on support, analytical thinking (-) on intrusiveness. Cross-country differences notes (Swedish respondents more supportive of nudges than US citizens). Six areas of 'knowns' identified, including basic typologies of policy tools and motivations for policy makers choosing amongst them N/A Mentions policy taker behaviour and compliance, which we would concur with. Also mentions the impact of sequencing and speed considerations -the dynamics of time in policy instrument choicebut not really from the point of view of public preferences. Yes, nudges were framed differently either in the second-or thirdperson plural.
The support for policies is more when the justification is framed and points to others in general than the individual. In Study 1 (third person effects): Having beliefs about unintended policy consequences and achieving policy goals (+) with support. In Study 2 (replication study with more controls and increased policy options), the above correlations (+) hold. Agency (-) with policy support in the second person framing condition. The results find evidence in favour of third person effects in the literature of nudging. Yes, level of transparency was increased in different choice architecture interventions to see effect on perceived effectiveness and acceptability.
In Study 1: Perceived effectiveness was larger for others than one's own self. Higher perceived effectiveness explained higher acceptability of the choice architectural manipulation. N/A In Study 2: Framing effects found; for example, riskier option chosen significantly more in loss frame. Perceived intentions also differed by source; for example, government received more responses as its 'sustainability' motive was trusted more relative to airline companies. Third party sources (-) saw lower effect compared to friends. No, the paper outlines a research agenda that must be followed in any public policy administration discourse; of the seven broad themes identified, one includes the design of policies and the instruments.

Survey
The authors identify seven broad themes that they urge public policy makers to consider when designing policies relevant to social problems that we face; these include policy design and instruments, policy learning, public service and its publics, organisational capacity, public governance, administrative traditions and public sector reforms in multi-level governance. The themes identified are, by no means, prescriptive and are a tool to help policymakers in thinking out loud in their considerations.
Thematic analysis N= N/A DV= N/A