https://orcid.org/0000-0002-7615-6779
Abstract
Bridging deliberative democracy and crisis management scholarship, we construct theoretical expectations about the role of deliberative minipublics in fostering public compliance with difficult political decisions. Our expectations are tested with a randomized cross-national survey experiment (United States and United Kingdom, N = 2088), in which respondents read a realistic news item depicting a political decision-making process leading to the extension of COVID-19 lockdown measures that follows either a (1) citizen deliberation, (2) public consultation, (3) politician deliberation, or (4) nothing. The findings show minipublics are unlikely to foster public compliance during a health crisis. On the contrary, reading about a minipublic could decrease compliance when individuals are distrustful of minipublics. This study has implications for citizen participation, deliberation, and leadership during future pandemics.
How do we achieve widespread public compliance with difficult political decisions? One possible solution, advanced in prior work, is involving lay citizens in political decision-making processes. Scholars argue that involving citizens can foster perceived democratic legitimacy among the citizenry (Barber, 1984; Dryzek et al., 2019; Pateman, 1970) and enhance public compliance with difficult decisions (e.g., Esaiasson, Gilljam, & Persson, 2012). A key assumption is that the decision-making mechanisms that include the public underpin public perceptions of the fairness of these processes. This may, in turn, improve people’s acceptance of unfavorable policy outcomes (for alternative mechanisms, see e.g., Esaiasson, Persson, Gilljam, & Lindholm., 2016; Marien & Kern, 2018).
Yet, most empirical studies testing this argument have focused on the small number of participating citizens—the individuals who partake in these processes. The influence of participatory mechanisms on public compliance among nonparticipating citizens, that is, individuals in the wider public, has been contested and understudied (Lafont, 2015; Curato, Vrydagh, & Bächtiger, 2020; Goldberg, 2021). It is not immediately clear whether and if so how participatory mechanisms can democratically prompt public compliance with difficult decisions among the broader citizenry.
An emerging body of recent work examines the upscaling effects of citizen-centered mechanisms by examining the extent to which they can improve voter knowledge (Már & Gastil, 2020), policy opinions (Boulianne, 2018; Ingham & Levin, 2018a), and individual capabilities (Suiter, Muradova, Gastil, & Farrel, 2020). In this article, we contribute to this nascent strand of work by evaluating how learning about a deliberative minipublic1—a body of randomly chosen lay citizens, tasked with considering policy issues at length and making policy recommendations (Smith & Setälä, 2018)—can increase people’s compliance with a difficult policy decision2. We build on Warren and Gastil’s (2015) argument that minipublics can act as trusted information proxies for citizens in the wider public; and the idea that people would comply with difficult political outcomes more if these outcomes were the result of deliberative decision-making processes involving other lay citizens (see also Esaiasson, Gilljam, & Persson, 2012). We depart from previous work by focusing on individuals’ (self-reported) compliance in the context of a crisis—“whose scale, timing, or unpredictability threatens to overwhelm routine capabilities” (Nelson, Lurie, Wasserman, & Zakowski, 2007, S11) and use a realistic news item depicting a real-world political decision-making mechanism.
To test our hypotheses, we conducted a cross-national survey experiment with 2,088 U.S. and U.K. respondents at the start of the COVID-19 crisis (April 2020). Political decision-making arrangements were manipulated with a news item, informing citizens of government decisions to extend COVID-19 lockdown measures. We manipulated four different kinds of institutional arrangements. We designed elite deliberation and public consultancy conditions to disentangle the participative and deliberative elements of a minipublic. We study the effect of such processes on citizens’ willingness to comply with the lockdown measures.
Our study makes several contributions to both the science and practice of deliberative democracy. First, it advances the growing literature on the upscaling effects of small-scale democratic experiments. We incorporate insights from crisis communication and theoretically elucidate the extent to which citizen deliberation can exert a pernicious effect on public compliance during a pandemic. Second, we make two methodological contributions. Our study design enables us to causally unpack the effects of the participative and deliberative elements of a deliberative minipublic. Furthermore, our easy-to-implement and realistic news item has high ecological validity and may interest scholars studying the realistic effects of informational cues about deliberative minipublics on public attitudes and behavior. Third, this study provides timely policy implications for the practice of participatory institutions during a pandemic. Against the background of the upsurge of participatory and deliberative projects during the pandemic (www.participedia.net; see also Elstub et al., 2021; Smith, Hughes, Adams, & Obijiaku, 2021, for a discussion), it is crucial to understand their function during these sorts of crises.
Deliberative political decision-making arrangements and broader public opinion
Proponents of deliberative democracy, a school of thought that places citizen deliberation at the heart of political decision-making (Bächtiger, Dryzek, Mansbridge, & Warren, 2018), argue that minipublics can have beneficial effects on the capabilities, attitudes, and behaviors of not only participating (Esaiasson, Gilljam, & Persson, 2012; Gastil, 2018; Mendelberg, 2002; Muradova, 2021) but also nonparticipating citizens—citizens in the wider public (e.g., Boulianne, 2018; Már & Gastil, 2020). One prominent mechanism of influence of small-scale citizen forums on nonparticipating citizens is for deliberative minipublics to act as trusted information proxies for citizens in the broader public (MacKenzie & Warren, 2012; Warren & Gastil, 2015). The assumption is that individuals in democracies cannot possibly engage with all the information concerning the sociopolitical issues they are affected by, given the complexity of contemporary politics and the cognitive cost of this engagement. Instead, citizens tend to trust others to decide things on their behalf (Parkinson & Mansbridge, 2012). Deliberative minipublics can potentially be the kind of institution that individuals can trust for (at least) two reasons (MacKenzie & Warren, 2012; Warren & Gastil, 2015). First, well-organized minipublics involve the random selection of lay citizens from different walks of life. As such, a minipublic constitutes a microcosm of the citizenry in broader society (Fishkin, 2018). Seeing themselves reflected in the membership of a minipublic (e.g., lay citizens who are independently motivated and represent diverse sets of needs and preferences), citizens may perceive the political process involving a minipublic as more trustworthy and legitimate (Warren & Pearse, 2008). Pow et al. (2020, 45) argue that citizens in the wider public can form a psychological attachment to minipublics via a similarity mechanism. Individuals can perceive the participants of the minipublic to be “like them” with the similarity defined not only along sociodemographic lines, but also by life experiences, needs, and perspectives. For this mechanism to work, respondents in the wider public are supposed to be informed about the members of the institution. The authors test their argument in the political context of Northern Ireland and find that respondents’ perceived similarity constitutes an important predictor of perceptions about the legitimacy of minipublics. Second, individuals in minipublics accrue competence by learning from experts and each other through deliberation (Warren & Gastil, 2015). Citizens may perceive the members of the minipublic as more enlightened versions of themselves. The resulting trust could prompt citizens to use the recommendations of minipublics as trustworthy cognitive shortcuts and align their beliefs with those of the minipublic.3,4
Prior research provides some empirical support for the theorized effects of deliberative minipublics on people’s attitudes and behavior. Reading about a randomly chosen group of lay citizens who have discussed a policy issue at length and made recommendations can expand and improve public engagement in politics, increase public trust and democratic legitimacy, and make citizens’ political attitudes and behavior more reflective (Esaiasson, Gilljam, & Persson, 2012; Boulianne, 2018; Ingham & Levin, 2018a; Knobloch, Barthel, & Gastil, 2019; Suiter, Muradova, Gastil, & Farrell, 2020; Suiter & Reidy, 2020). Ingham and Levin (2018a) show that learning about a deliberative minipublic alters respondents’ policy opinions, bringing them toward the opinions of the participants in the minipublics, while Suiter and Reidy (2020) find systematic improvements in democratic outcomes such as alignment between values and the actual vote. Suiter, Muradova, Gastil, and Farrell (2020) corroborate that reading a report from a minipublic can strengthen the deliberative capacity of an electorate, by making individuals more empathetic and knowledgeable about the discussed issue. Knobloch, Barthel, and Gastil (2019) find increases in individuals’ internal and external efficacies. López-Moctezuma, Wantchekon, Rubenson, Fujiwara, and Pe Lero (2022) show that one form of minipublics, town-hall meetings, can encourage changes in citizens’ voting behavior (see also Esterling, Neblo, & Lazer, 2011). Finally, Werner and Marien (2022) in the context of experiments fielded in the Netherlands find that hearing about a deliberative minipublic can enhance decision acceptance on contentious issues, even among the losers from the decision.
Building on this body of work, and the idea that people would comply with political outcomes more if these outcomes are the result of deliberative decision-making processes involving other lay citizens (Warren & Gastil, 2015), we hypothesize that:
H1: Learning that a minipublic is a part of the political decision-making process leading to a difficult policy outcome will increase people’s compliance with this outcome.
To the extent that trust in minipublics plays an important role in this relationship, individuals’ baseline trust in minipublics would moderate the relationship, with trustful citizens being persuaded by minipublic recommendations more than the distrustful ones. Therefore, we advance that:
H2: The effect of a deliberative minipublic will be stronger among citizens who trust in the capacity of minipublics to participate in political decision-making.
Yet, existing empirical findings about the relationship between reading/hearing about a minipublic and the potential transformation of nonparticipating citizens’ attitudes and behavior are not always uniformly supportive. For instance, Boulianne’s (2018) experiments provide mixed empirical evidence. In her studies, while exposure to minipublic information exerts a positive and statistically significant effect on public opinion on policies concerning the promotion of energy-efficient buildings, it fails to do so on attitudes about the construction of buildings. The author conjectures that her inconsistent findings may be due to the differing features of the described minipublics (e.g., the length of deliberation) in the study. Similarly, Ingham and Levin (2018b) find only a partial support for the effect of learning about a minipublic on public support for two different policies (retirement age and the cap on payroll taxes). They conclude that this difference may depend on the policy issue in question and the strength of people’s prior attitudes (see also a review by van der Does & Jacquet, 2021).
Furthermore, all of these studies have been conducted in “normal” times. The literature has been both theoretically and empirically mute on the upscaling effects of deliberative institutions on public compliance in crisis situations, some of which require unprecedented, fast, and more centralized modes of governance. Therefore, it is not immediately clear whether deliberative instruments can prompt public compliance among the broader citizenry during such crises. In the subsequent section, we build on the crisis management literature to advance a set of hypotheses contradictory to the above expectations.
Public health crises: alternative hypotheses
Crises are situations that are “rare, significant, high impact, ambitious, urgent and involve high stakes” (Serra Gorpe & Zamoum, 2018, 204). Even though it has been defined differently across different fields of research, a commonly accepted definition of a crisis is “a threat that is perceived to be existential in one way or another” (Boin, ‘t Hart, & Kuipers, 2018, 24). Whether a war, a pandemic, or a natural or financial disaster, crises can cause insecurity, uncertainty, and fear in the public. Yet, there is divergence in the type, speed, and scale of government responses to different crises. Some crises trigger immediate and strong responses, while others get slow-moving reactions from the government. According to van der Ven and Sun (2021), their differing characteristics explain the variation in political responses to different global crises. They advance a conceptual framework based on eight different dimensions of the crises (immediacy, transience, visibility, proximity, accountability, universality, expertise, and legibility).5 The crises that score high on these features, according to this framework, receive far-reaching and rapid government responses (and vice versa).
We borrow this reasoning and apply it to studying public reactions to different political decision-making processes. We argue that the public perceives pandemics, or global health crises, as particularly severe. Individuals find immediate and centralized government responses essential and urgent when faced with such crises. This happens because of three main characteristics of the global health crises, exemplified by the COVID-19 pandemic. First, responding to COVID-19, governments worldwide legitimized the use of lockdown measures with an appeal to the immediacy of the situation. The perceived immediacy of the crisis contributes to the level of salience of response among individuals, “[keeps] the crisis in the public mind and reduce[s] opposition to drastic responses” (van der Ven & Sun, 2021). COVID-19 as a global health crisis thus scores highly on the immediacy feature of crises. Second, during the early stages of the crisis, the most important public health strategy for controlling the pandemic was flattening the curve, which entailed a short-term delay of the peak of the outbreak so that the health systems could cope with the demand piling up on their services. As a result, people may have been initially convinced that the crisis was likely short term in nature. A perception of the crisis as being short term may have encouraged citizens to temporarily live with the lockdown to control the spread of the pandemic. As van der Ven and Sun (2021) argue, this perceived transience may persuade the public to be willing to put up with more centralized modes of policy-making. Third, the proximity of this crisis, “the degree to which a crisis appears near to an observer” (van der Ven & Sun 2021), could have rendered swift government responses, essential in the public eye.
Thus, the combination of high immediacy, proximity, and perceived transience could ensure the pandemic is perceived as especially severe in the eyes of the citizens, which would contribute to individuals’ preference for immediate and centralized decision-making. This is consistent with the literature on authoritarianism which posits that extreme crises impose immense threats on citizens and when faced with such crises, citizens are willing to trade off their democratic preferences in return for safety and security (Altemeyer, 1996; Merolla & Zechmeister, 2009; Amat, Arenas, Falcó-Gimeno, & Munoz, 2020). As Miller (2017) argues, this is in line with the Hobbesian insight that the desire for security is the most reliable and rational desire of human nature, as people are driven first by fear of their mortality, which is exacerbated when there is no sovereign government capable of regulating conduct (Wrong, 1994). Anxiety induced by increased perceptions of risk encourages people to embrace more risk-averse behavior and, often more conservative policies in order to eliminate or prevent the sense of heightened threat (Vasilopoulos & Marcus, 2018). A considerable body of research has repeatedly shown that rising threat levels lead to more authoritarian preferences (Feldman, 2003; Huddy, Feldman, Taber, & Lahav, 2005). Recent research has also found that one type of threat, the prevalence of infectious diseases, causes more authoritarian modes of governance, particularly if the diseases inflict elevated stress on the population’s welfare (Murray, Schaller, & Suedfeld, 2013).
Furthermore, prior research shows that perceived crisis severity, defined as “the amount of damage generated by a crisis including financial, human, and environmental damage” (Coombs & Holladay, 2002, 169), is an important predictor of the adoption of precautionary health behaviors (Trkman, Popovič, & Trkman, 2021) and can determine the outcome of a crisis (see Zhou, Eyun-Jung, & Brown, 2019, for a review). Convening minipublics requires time and resources. Learning that the government has delayed decision-making to convene a deliberative minipublic may negatively affect citizens’ perceptions about the seriousness of the crisis, leading to greater reluctance to comply with the required measures.6 Building on this reasoning, we advance a set of alternative hypotheses:
H3: Learning that a deliberative minipublic is part of the political decision-making processes leading to policy outcomes will decrease people’s compliance with these outcomes.
To the extent that authoritarianism is part of the equation, we expect the negative effect of the minipublic to be stronger among those who hold more authoritarian preferences in times of a global health crisis.
H4: The negative effect of a deliberative minipublic will be stronger among those with more authoritarian and centralized preferences toward managing the crisis.
Research design
To test our hypotheses, we apply a cross-national, between-subject, and preregistered survey experiment.7 The experiment was fielded in the United States (N = 1047) and the United Kingdom (N = 1041) at the start of a global health crisis, COVID-19 (April 2020).
Case selection
This study focuses on deliberative minipublics in majoritarian, Anglo-Saxon countries: United Kingdom and United States. Both countries were at a similar stage of the COVID-19 pandemic at the time of the study. By 27 March, people in the United Kingdom could not leave their homes except under limited circumstances. In the United States, the President issued nonbinding guidelines, and at state level, the governors issued stay-at-home edicts. During the first weeks of the pandemic, most U.S. states reacted similarly, issuing “shelter-at-home” measures. Boris Johnson and Donald Trump, right-wing populists who governed both countries, were up to that time openly dismissive of the seriousness of the virus and considering a herd immunity strategy.
Both countries have some experience with minipublics. According to the Participedia database, some 24 minipublics were convened in the United Kingdom in 2019, one of which was an important U.K.-wide citizens’ assembly on climate change commissioned by several Westminster parliamentary committees.8 The most prominent example of U.S. minipublics is the Citizen Initiative Review in Oregon (Gastil & Knobloch, 2020) and the Deliberative Polls led by James Fishkin (2018). Fishkin has conducted at least three national projects with the first “National Issues Convention” broadcast widely on PBS in 1996 and America in One Room (Fishkin, Siu, Diamond, & Bradburn, 2021) as well as numerous state-wide deliberative polls.
When it comes to social-distancing measures, there were country-level differences. According to the YouGov-Imperial College London COVID-19 Behaviour Tracker (2020), as of April 6–12 (at the time of our study), only 72% of surveyed U.S. and 82% of surveyed U.K. respondents reported avoiding going out in general, either always or frequently. Having two country cases enable us to examine whether the observed effects are not specific to a single-country context with a specific experience with minipublics and the pandemic.
Sample
We administered our survey experiment via the widely used online platform Prolific, which provides researchers with diverse, but nonprobability samples.9 Although this subject pool is not representative and there are some concerns about self-selecting in these samples, prior research has found that treatment effects obtained from such opt-in platforms do not differ substantively from treatment effects found in other more diverse subject pools (Berinsky, Huber, & Lenz, 2012, 366; Huff & Tingley, 2015). Our samples are biased toward young, highly educated, and more liberal Americans; and female, young, and highly educated U.K. respondents (see Online Appendix B for the sample descriptive statistics and population reference figures). We control for the main sociodemographic variables and political attitudes in some of our models to account for potential bias.
Experimental design and stimuli
The experiment had the following structure (Figure 1). First, respondents took a pre-experimental survey measuring a set of variables (e.g., sociodemographics, political attitudes). Next, they were randomly assigned to one of four experimental conditions (control, citizen deliberation, politician deliberation, and public consultation). Finally, they completed a short survey measuring our outcome variables.
Experimental design.
In designing our experimental stimuli (i.e., a minipublic), we attempted to strike a balance between realism and parsimony. Existing studies primarily rely on designs where respondents are exposed to compendious information about a deliberative minipublic in the form of a vignette, devoid of any other information. Such exposures, however, do not reflect how citizens typically receive information about minipublics. People usually read (or hear) about such processes within news reports, which is significantly more ambiguous than receiving short vignettes. Our informational exposure consists of a realistic news item built on the real-world situations in each country and thus enjoys a high ecological validity (Figure 2).10
An example of news item (United Kingdom).
The news item informed participants about the government’s decision to extend the lockdown measures until May 15 (a month hence), which was constant across all experimental conditions. The second part of the news item introduced the stimuli (Table 1 presents the exact wording of our deconstructed stimuli). There, the government decision followed either citizen deliberation, politician deliberation, public consultation, or nothing. The politician deliberation condition was aimed at comparing the effects of citizen deliberation to the elite deliberation, while the objective of the public consultation condition was to investigate whether mere citizen participation (without its deliberative component) has a similar effect. Our design entailed a 20-s forced exposure to the news item. We use the term “citizens’ assemblies” for a deliberative minipublic, a more commonly used term and more likely to be recognized by people. We know that capturing the idea of a deliberative minipublic with a short-text exposure is challenging. We chose to concentrate on the following aspects of a minipublic, consistent with the trusted information proxies theory. First, to signal the similarity, we emphasized that the citizens of the minipublic were chosen randomly and were representative of the larger population. Second, we mentioned that the minipublic heard from the experts and deliberated for an extended time to signal their competence.
Experimental Treatments (U.K. sample)
| Conditions | A | B | C |
|---|---|---|---|
| Control | The Government has decided to extend the COVID-19 lockdown until May 15 | [nothing here] | Under the restrictions everybody must continue to stay at home and only leave: to exercise only within 1 mile of home—either alone, or with members of your household; shopping for basic necessities (the guidelines say this should be done as infrequently as possible); any medical need, including to donate blood, avoid or escape risk of injury or harm, or to provide care or to help a vulnerable person; to travel to or from work, but only if it is deemed essential. Other valid reasons include attending the funeral of a close family member, key workers taking children to childcare, and because of access arrangements for a child. |
| Citizen Deliberation | The Government has decided to extend the COVID-19 lockdown until May 15 | …following a recommendation from a Citizens’ Assembly. The Assembly was made up of 1000 citizens, randomly chosen like a jury and demographically balanced to represent the wider public. The assembly met online for five consecutive days to discuss COVID_19 with medical experts and economists to weigh evidence before issuing their recommendation. | Under the restrictions everybody must continue to stay at home and only leave: to exercise only within 1 mile of home – either alone, or with members of your household; shopping for basic necessities (the guidelines say this should be done as infrequently as possible); any medical need, including to donate blood, avoid or escape risk of injury or harm, or to provide care or to help a vulnerable person; to travel to or from work, but only if it is deemed essential. Other valid reasons include attending the funeral of a close family member, key workers taking children to childcare, and because of access arrangements for a child. |
| Politician Deliberation | The Government has decided to extend the COVID-19 lockdown until May 15 | …following a recommendation from a Select Committee. This cross-party Select Committee was made up of MPs from all major parties. The committee called witnesses during public hearings before writing and voting on their final report. | Under the restrictions everybody must continue to stay at home and only leave: to exercise only within 1 mile of home – either alone, or with members of your household; shopping for basic necessities (the guidelines say this should be done as infrequently as possible); any medical need, including to donate blood, avoid or escape risk of injury or harm, or to provide care or to help a vulnerable person; to travel to or from work, but only if it is deemed essential. Other valid reasons include attending the funeral of a close family member, key workers taking children to childcare, and because of access arrangements for a child. |
| Public Consultation | The Government has decided to extend the COVID-19 lockdown until May 15 | …following a public consultation. This online consultation process permitted anyone to send a comment to the Government about the situation or to respond to other people’s suggestions. | Under the restrictions everybody must continue to stay at home and only leave: to exercise only within 1 mile of home – either alone, or with members of your household; shopping for basic necessities (the guidelines say this should be done as infrequently as possible); any medical need, including to donate blood, avoid or escape risk of injury or harm, or to provide care or to help a vulnerable person; to travel to or from work, but only if it is deemed essential. Other valid reasons include attending the funeral of a close family member, key workers taking children to childcare, and because of access arrangements for a child. |
| Conditions | A | B | C |
|---|---|---|---|
| Control | The Government has decided to extend the COVID-19 lockdown until May 15 | [nothing here] | Under the restrictions everybody must continue to stay at home and only leave: to exercise only within 1 mile of home—either alone, or with members of your household; shopping for basic necessities (the guidelines say this should be done as infrequently as possible); any medical need, including to donate blood, avoid or escape risk of injury or harm, or to provide care or to help a vulnerable person; to travel to or from work, but only if it is deemed essential. Other valid reasons include attending the funeral of a close family member, key workers taking children to childcare, and because of access arrangements for a child. |
| Citizen Deliberation | The Government has decided to extend the COVID-19 lockdown until May 15 | …following a recommendation from a Citizens’ Assembly. The Assembly was made up of 1000 citizens, randomly chosen like a jury and demographically balanced to represent the wider public. The assembly met online for five consecutive days to discuss COVID_19 with medical experts and economists to weigh evidence before issuing their recommendation. | Under the restrictions everybody must continue to stay at home and only leave: to exercise only within 1 mile of home – either alone, or with members of your household; shopping for basic necessities (the guidelines say this should be done as infrequently as possible); any medical need, including to donate blood, avoid or escape risk of injury or harm, or to provide care or to help a vulnerable person; to travel to or from work, but only if it is deemed essential. Other valid reasons include attending the funeral of a close family member, key workers taking children to childcare, and because of access arrangements for a child. |
| Politician Deliberation | The Government has decided to extend the COVID-19 lockdown until May 15 | …following a recommendation from a Select Committee. This cross-party Select Committee was made up of MPs from all major parties. The committee called witnesses during public hearings before writing and voting on their final report. | Under the restrictions everybody must continue to stay at home and only leave: to exercise only within 1 mile of home – either alone, or with members of your household; shopping for basic necessities (the guidelines say this should be done as infrequently as possible); any medical need, including to donate blood, avoid or escape risk of injury or harm, or to provide care or to help a vulnerable person; to travel to or from work, but only if it is deemed essential. Other valid reasons include attending the funeral of a close family member, key workers taking children to childcare, and because of access arrangements for a child. |
| Public Consultation | The Government has decided to extend the COVID-19 lockdown until May 15 | …following a public consultation. This online consultation process permitted anyone to send a comment to the Government about the situation or to respond to other people’s suggestions. | Under the restrictions everybody must continue to stay at home and only leave: to exercise only within 1 mile of home – either alone, or with members of your household; shopping for basic necessities (the guidelines say this should be done as infrequently as possible); any medical need, including to donate blood, avoid or escape risk of injury or harm, or to provide care or to help a vulnerable person; to travel to or from work, but only if it is deemed essential. Other valid reasons include attending the funeral of a close family member, key workers taking children to childcare, and because of access arrangements for a child. |
Note: Texts in bold are treatments.
Experimental Treatments (U.K. sample)
| Conditions | A | B | C |
|---|---|---|---|
| Control | The Government has decided to extend the COVID-19 lockdown until May 15 | [nothing here] | Under the restrictions everybody must continue to stay at home and only leave: to exercise only within 1 mile of home—either alone, or with members of your household; shopping for basic necessities (the guidelines say this should be done as infrequently as possible); any medical need, including to donate blood, avoid or escape risk of injury or harm, or to provide care or to help a vulnerable person; to travel to or from work, but only if it is deemed essential. Other valid reasons include attending the funeral of a close family member, key workers taking children to childcare, and because of access arrangements for a child. |
| Citizen Deliberation | The Government has decided to extend the COVID-19 lockdown until May 15 | …following a recommendation from a Citizens’ Assembly. The Assembly was made up of 1000 citizens, randomly chosen like a jury and demographically balanced to represent the wider public. The assembly met online for five consecutive days to discuss COVID_19 with medical experts and economists to weigh evidence before issuing their recommendation. | Under the restrictions everybody must continue to stay at home and only leave: to exercise only within 1 mile of home – either alone, or with members of your household; shopping for basic necessities (the guidelines say this should be done as infrequently as possible); any medical need, including to donate blood, avoid or escape risk of injury or harm, or to provide care or to help a vulnerable person; to travel to or from work, but only if it is deemed essential. Other valid reasons include attending the funeral of a close family member, key workers taking children to childcare, and because of access arrangements for a child. |
| Politician Deliberation | The Government has decided to extend the COVID-19 lockdown until May 15 | …following a recommendation from a Select Committee. This cross-party Select Committee was made up of MPs from all major parties. The committee called witnesses during public hearings before writing and voting on their final report. | Under the restrictions everybody must continue to stay at home and only leave: to exercise only within 1 mile of home – either alone, or with members of your household; shopping for basic necessities (the guidelines say this should be done as infrequently as possible); any medical need, including to donate blood, avoid or escape risk of injury or harm, or to provide care or to help a vulnerable person; to travel to or from work, but only if it is deemed essential. Other valid reasons include attending the funeral of a close family member, key workers taking children to childcare, and because of access arrangements for a child. |
| Public Consultation | The Government has decided to extend the COVID-19 lockdown until May 15 | …following a public consultation. This online consultation process permitted anyone to send a comment to the Government about the situation or to respond to other people’s suggestions. | Under the restrictions everybody must continue to stay at home and only leave: to exercise only within 1 mile of home – either alone, or with members of your household; shopping for basic necessities (the guidelines say this should be done as infrequently as possible); any medical need, including to donate blood, avoid or escape risk of injury or harm, or to provide care or to help a vulnerable person; to travel to or from work, but only if it is deemed essential. Other valid reasons include attending the funeral of a close family member, key workers taking children to childcare, and because of access arrangements for a child. |
| Conditions | A | B | C |
|---|---|---|---|
| Control | The Government has decided to extend the COVID-19 lockdown until May 15 | [nothing here] | Under the restrictions everybody must continue to stay at home and only leave: to exercise only within 1 mile of home—either alone, or with members of your household; shopping for basic necessities (the guidelines say this should be done as infrequently as possible); any medical need, including to donate blood, avoid or escape risk of injury or harm, or to provide care or to help a vulnerable person; to travel to or from work, but only if it is deemed essential. Other valid reasons include attending the funeral of a close family member, key workers taking children to childcare, and because of access arrangements for a child. |
| Citizen Deliberation | The Government has decided to extend the COVID-19 lockdown until May 15 | …following a recommendation from a Citizens’ Assembly. The Assembly was made up of 1000 citizens, randomly chosen like a jury and demographically balanced to represent the wider public. The assembly met online for five consecutive days to discuss COVID_19 with medical experts and economists to weigh evidence before issuing their recommendation. | Under the restrictions everybody must continue to stay at home and only leave: to exercise only within 1 mile of home – either alone, or with members of your household; shopping for basic necessities (the guidelines say this should be done as infrequently as possible); any medical need, including to donate blood, avoid or escape risk of injury or harm, or to provide care or to help a vulnerable person; to travel to or from work, but only if it is deemed essential. Other valid reasons include attending the funeral of a close family member, key workers taking children to childcare, and because of access arrangements for a child. |
| Politician Deliberation | The Government has decided to extend the COVID-19 lockdown until May 15 | …following a recommendation from a Select Committee. This cross-party Select Committee was made up of MPs from all major parties. The committee called witnesses during public hearings before writing and voting on their final report. | Under the restrictions everybody must continue to stay at home and only leave: to exercise only within 1 mile of home – either alone, or with members of your household; shopping for basic necessities (the guidelines say this should be done as infrequently as possible); any medical need, including to donate blood, avoid or escape risk of injury or harm, or to provide care or to help a vulnerable person; to travel to or from work, but only if it is deemed essential. Other valid reasons include attending the funeral of a close family member, key workers taking children to childcare, and because of access arrangements for a child. |
| Public Consultation | The Government has decided to extend the COVID-19 lockdown until May 15 | …following a public consultation. This online consultation process permitted anyone to send a comment to the Government about the situation or to respond to other people’s suggestions. | Under the restrictions everybody must continue to stay at home and only leave: to exercise only within 1 mile of home – either alone, or with members of your household; shopping for basic necessities (the guidelines say this should be done as infrequently as possible); any medical need, including to donate blood, avoid or escape risk of injury or harm, or to provide care or to help a vulnerable person; to travel to or from work, but only if it is deemed essential. Other valid reasons include attending the funeral of a close family member, key workers taking children to childcare, and because of access arrangements for a child. |
Note: Texts in bold are treatments.
Measures
Our dependent variable is individuals’ compliance with lockdown measures.11 It is measured with a self-reported survey item.12 Respondents were asked: “Are you willing to comply with the coronavirus (COVID-19) lockdown measures for as long as it takes?” The response options range from “extremely unwilling” to “extremely willing” on an 11-point scale. Mean public compliance in both countries is high, with the highest among the U.K. respondents (MUS = 8.35; SDUS = 2.04; MUK = 9.03; SDUK = 1.48)13.
Trust in minipublics is measured with the following item: “Do you think in general citizens’ assemblies can be trusted to make informed decisions about a range of policy issues?” (MUS = 5.69; SDUS = 2.03; MUK = 5.9; SDUK = 1.9; response range: 0–10).14
Authoritarian attitudes in times of crises is operationalized using a five-item battery which includes the following statements: “In times of crises, such as coronavirus (COVID-19), (a) individual freedoms become less important; (b) heavy-handed measures must be imposed by the government, (c) governments must act without delay; (d) governments should respect individual freedoms as much as possible, (d) citizens should be consulted before imposing strict restrictions on freedoms.”15 The response options range from “strongly agree” to “strongly disagree” on a 7-point Likert scale. For our analyses, we reversed the scale so that the greater scores reflect more authoritarian attitudes. The validity and reliability of the battery were pretested in a pilot study (N = 60). The items were summed (α = 0.77 in the U.S. and α = 0.76 in the U.K. samples) and rescaled to a 0–10 scale (MUS = 6.13; SDUS = 1.82; MUK = 6.4; SDUS = 1.85).
Controls
A series of control variables were used in some of our analyses, including individuals’ threat perceptions about COVID-19, their political partisanship, political ideology, trust in scientists, interpersonal trust, awareness of minipublics, trust in politicians, age, gender, and education (see the Online Appendix C).
Results
To test our hypotheses, we start by estimating the main effect of minipublic deliberation on public compliance with COVID-19 lockdown measures. Table 2 shows the results of six OLS regression models. The Pooled Model 1 estimates the effect of the treatments compared to the control condition. In Models 2 and 3, we rerun the same analyses per country separately. The minipublic treatment (versus control) exerts a null effect and there are no country differences. In other words, reading that a citizen deliberation mechanism was part of the decision-making processes leading to the government’s decision to extend the COVID-19 lockdown measures makes people neither more nor less likely to express their willingness to comply with the lockdown measures for as long as it takes. These findings are inconsistent with both H1 and H3. Further, neither politicians’ deliberation nor public consultation influences people’s willingness to comply with the lockdown measures. When we compare the effect of minipublic to that of politician deliberation and public consultation (Model 4), we observe no statistically significant differences. Finally, the treatments do not explain much in the variation of the public compliance (e.g., R2 is extremely low).
The Effect of Minipublic Deliberation on Public Compliance with COVID-19 Lockdown Measures
| Pooled-M1 | US-M1 | UK-M1 | M2 | M3 | M4 | |
|---|---|---|---|---|---|---|
| compliance | compliance | compliance | compliance | compliance | compliance | |
| Control | 0 | 0 | 0 | 0.08 | 0.00 | 0.19 |
| (.) | (.) | (.) | (0.11) | (0.11) | (0.11) | |
| Politician Del | −0.08 | −0.21 | 0.05 | 0 | −0.07 | 0.12 |
| (0.11) | (0.18) | (0.13) | (.) | (0.11) | (0.11) | |
| Public Cons | −0.00 | −0.00 | −0.01 | 0.07 | 0 | 0.19 |
| (0.11) | (0.18) | (0.13) | (0.11) | (.) | (0.11) | |
| Citizen Del | −0.19 | −0.31 | −0.08 | −0.12 | −0.19 | 0 |
| (0.11) | (0.18) | (0.13) | (0.11) | (0.11) | (.) | |
| _cons | 8.76*** | 8.48*** | 9.04*** | 8.68*** | 8.76*** | 8.56*** |
| (0.08) | (0.13) | (0.09) | (0.08) | (0.08) | (0.08) | |
| N | 2086 | 1046 | 1040 | 2086 | 2086 | 2086 |
| R2 | 0.002 | 0.004 | 0.001 | 0.002 | 0.002 | 0.002 |
| adj. R2 | 0.000 | 0.001 | −0.002 | 0.000 | 0.000 | 0.000 |
| Pooled-M1 | US-M1 | UK-M1 | M2 | M3 | M4 | |
|---|---|---|---|---|---|---|
| compliance | compliance | compliance | compliance | compliance | compliance | |
| Control | 0 | 0 | 0 | 0.08 | 0.00 | 0.19 |
| (.) | (.) | (.) | (0.11) | (0.11) | (0.11) | |
| Politician Del | −0.08 | −0.21 | 0.05 | 0 | −0.07 | 0.12 |
| (0.11) | (0.18) | (0.13) | (.) | (0.11) | (0.11) | |
| Public Cons | −0.00 | −0.00 | −0.01 | 0.07 | 0 | 0.19 |
| (0.11) | (0.18) | (0.13) | (0.11) | (.) | (0.11) | |
| Citizen Del | −0.19 | −0.31 | −0.08 | −0.12 | −0.19 | 0 |
| (0.11) | (0.18) | (0.13) | (0.11) | (0.11) | (.) | |
| _cons | 8.76*** | 8.48*** | 9.04*** | 8.68*** | 8.76*** | 8.56*** |
| (0.08) | (0.13) | (0.09) | (0.08) | (0.08) | (0.08) | |
| N | 2086 | 1046 | 1040 | 2086 | 2086 | 2086 |
| R2 | 0.002 | 0.004 | 0.001 | 0.002 | 0.002 | 0.002 |
| adj. R2 | 0.000 | 0.001 | −0.002 | 0.000 | 0.000 | 0.000 |
Note: Standard errors in parentheses.
p < .05, **p < .01, ***p < .001.
The Effect of Minipublic Deliberation on Public Compliance with COVID-19 Lockdown Measures
| Pooled-M1 | US-M1 | UK-M1 | M2 | M3 | M4 | |
|---|---|---|---|---|---|---|
| compliance | compliance | compliance | compliance | compliance | compliance | |
| Control | 0 | 0 | 0 | 0.08 | 0.00 | 0.19 |
| (.) | (.) | (.) | (0.11) | (0.11) | (0.11) | |
| Politician Del | −0.08 | −0.21 | 0.05 | 0 | −0.07 | 0.12 |
| (0.11) | (0.18) | (0.13) | (.) | (0.11) | (0.11) | |
| Public Cons | −0.00 | −0.00 | −0.01 | 0.07 | 0 | 0.19 |
| (0.11) | (0.18) | (0.13) | (0.11) | (.) | (0.11) | |
| Citizen Del | −0.19 | −0.31 | −0.08 | −0.12 | −0.19 | 0 |
| (0.11) | (0.18) | (0.13) | (0.11) | (0.11) | (.) | |
| _cons | 8.76*** | 8.48*** | 9.04*** | 8.68*** | 8.76*** | 8.56*** |
| (0.08) | (0.13) | (0.09) | (0.08) | (0.08) | (0.08) | |
| N | 2086 | 1046 | 1040 | 2086 | 2086 | 2086 |
| R2 | 0.002 | 0.004 | 0.001 | 0.002 | 0.002 | 0.002 |
| adj. R2 | 0.000 | 0.001 | −0.002 | 0.000 | 0.000 | 0.000 |
| Pooled-M1 | US-M1 | UK-M1 | M2 | M3 | M4 | |
|---|---|---|---|---|---|---|
| compliance | compliance | compliance | compliance | compliance | compliance | |
| Control | 0 | 0 | 0 | 0.08 | 0.00 | 0.19 |
| (.) | (.) | (.) | (0.11) | (0.11) | (0.11) | |
| Politician Del | −0.08 | −0.21 | 0.05 | 0 | −0.07 | 0.12 |
| (0.11) | (0.18) | (0.13) | (.) | (0.11) | (0.11) | |
| Public Cons | −0.00 | −0.00 | −0.01 | 0.07 | 0 | 0.19 |
| (0.11) | (0.18) | (0.13) | (0.11) | (.) | (0.11) | |
| Citizen Del | −0.19 | −0.31 | −0.08 | −0.12 | −0.19 | 0 |
| (0.11) | (0.18) | (0.13) | (0.11) | (0.11) | (.) | |
| _cons | 8.76*** | 8.48*** | 9.04*** | 8.68*** | 8.76*** | 8.56*** |
| (0.08) | (0.13) | (0.09) | (0.08) | (0.08) | (0.08) | |
| N | 2086 | 1046 | 1040 | 2086 | 2086 | 2086 |
| R2 | 0.002 | 0.004 | 0.001 | 0.002 | 0.002 | 0.002 |
| adj. R2 | 0.000 | 0.001 | −0.002 | 0.000 | 0.000 | 0.000 |
Note: Standard errors in parentheses.
p < .05, **p < .01, ***p < .001.
The bivariate relationship can hide potentially important heterogeneous effects. To test this and examine the H2, we add trust in minipublics (measured pretreatment) and its interaction term (with the treatment) to the main model (Figure 3). Descriptive statistics show that mean trust in minipublics among the U.S. and U.K. respondents is not high (MUS = 5.7, SDUS = 2.0 and MUK = 5.9, SDUK = 1.9, range: 0–10).
The effect of minipublic moderated by trust in minipublics.
As Figure 3 shows, there is a statistically significant interaction effect between the exposure to minipublics and trust in minipublics. Inclusion of trust in minipublics increases the goodness of fit of our model to R2 = 0.03. To improve the interpretability of the interaction effect, we calculate predicted public compliance for every value of trust in minipublics for each treatment condition (Figure 4).
Margins plot for treatments # trust in minipublics.
The results show that the minipublic intervention negatively affects compliance among citizens who distrust the ability of citizens’ assemblies to make informed decisions about a range of policy issues. The predicted margin of public compliance within the minipublic condition equals to 7.09 (95% CI: 6.60–7.58) when trust is the lowest and to 9.61 (95% CI: 9.25–9.98) when trust is the highest. This difference equals to 2.52 on an 11-point scale. The predicted margin of public compliance within the control condition at a low trust level is 7.91 (95% CI: 7.42–8.39), whereas it equals to 8.09 (95% CI: 7.63–8.55) and 7.97 (95% CI: 7.48–8.47) within the politician deliberation and public consultation conditions, respectively. When trust in minipublics is high, the difference in mean compliance between experimental conditions narrows down. For example, at the highest level of trust (10 in a scale of 0–10), mean compliance in the minipublic condition is 9.61 (95% CI: 9.25–9.98), whereas this equals to 9.36 (95% CI: 9.06–9.65), to 9.11 (95% CI: 8.76–9.46) and to 9.32 (95% CI: 8.95–9.7) in the control, politician deliberation and public consultation conditions, respectively. Two main observations are worth interpreting. First, trust in minipublics has a significant moderating effect on public compliance, with individuals espousing low trust in minipublics being negatively affected by exposure to minipublic and individuals with high trust in minipublics reporting increased levels of willingness to comply. Yet, the effect is modest (0.82 on an 11-point scale for low trust × minipublic; and 0.25 for high trust × minipublic, when the baseline condition is control) and thus should be interpreted with caution. Second, trust in minipublics seems to condition not only the effect of minipublic, but also the politician deliberation and public consultation, but to a lesser degree. For instance, when this trust is high, mean public compliance is at its highest across all experimental conditions and the difference between them is small. The study was fielded in times of political polarization surrounding the COVID-19 crisis; as such it is plausible that individuals’ distrust in minipublics is driven by their lack of general trust in strangers. To empirically examine this possibility, we run bivariate correlation analysis between trust in minipublics and interpersonal trust. The results of the pairwise correlation show that interpersonal trust and trust in minipublics are positively correlated, although this correlation is weak (r = 0.26). Exploratory analyses show that including this variable into the model or replacing the variable “trust in minipublics” with the variable “interpersonal trust” does not change the results substantially (Online Appendix M).
To test H4, we add an individual-level authoritarian attitudes variable and its interaction term to the model (Figure 5). There is no significant interaction between politician deliberation and authoritarian attitudes, or public consultation and authoritarian attitudes. Yet, citizens with less authoritarian attitudes toward the crisis are even less likely to comply when they read that it is the recommendation of their fellow citizens.
The effect of minipublic moderated by authoritarian attitudes.
The interaction term is statistically significant. Plotting the predicted margins (Figure 6) shows that when individuals’ authoritarian attitudes in times of a crisis are low, the minipublic exerts a negative effect on public compliance. Consider this: The predicted margin of public compliance within the minipublic condition equals 5.46 (95% CI: 4.98–5.93) when authoritarian preferences are at their lowest and 10.5 (95% CI: 10.2–10.8) when they are at their highest. This is a jump of 5 points on an 11-point scale. The predicted margin of public compliance within the control condition at low authoritarianism level is 6.26 (95% CI: 5.78–6.75). The difference between control and minipublic conditions in mean compliance when the authoritarian preferences is modest (0.80 on an 11-point scale). But as seen from the graph, this difference disappears at higher levels of authoritarian preferences. Overall, these differences may not be substantively meaningful.
Margins plot for the effect of minipublic deliberation on compliance moderated by authoritarian attitudes.
Further, we rerun our main model, this time controlling for covariates, such as COVID-19 threat perceptions, trust in experts, interpersonal trust, trust in politicians, gender, age, and people’s awareness about citizens’ assemblies (Figure 7). All predictors are entered simultaneously. The addition of the covariates makes the main effect of minipublic deliberation negative and significant. Similarly, the effect of politician deliberation also becomes statistically significant. Both deliberation conditions exert a negative and statistically significant effect on people’s self-reported willingness to comply with the COVID-19 lockdown measures, with public consultation having a null effect. Yet, the effects are small (β = −0.19 and β = −0.20 for politician deliberation and citizen deliberation, respectively) and rerunning the analyses per country dilutes the effects, most probably because of the reduced power.
The full model controlling for covariates.
Figure 7 indicates that the best predictor of individual compliance with lockdown measures is authoritarian attitudes in times of a crisis (β = 0.32; p < .00). The higher the respondents’ level of strong government preferences in response to a pandemic, the more likely they are to express their willingness to comply with the lockdown measures. Threat perceptions constitute the second strongest and most consistent predictor of public compliance (β = 0.21; p < .00), followed by trust in experts (β = 0.15; p < .00). Interpersonal trust, trust in politicians, and gender are also positively associated with public compliance. More trustful citizens report complying with the measures more. Males seem to be significantly less willing to adhere to the measures. Additional analyses (Online Appendix G) show that our treatments do not affect compliance differently depending on people’s partisan affiliations in both U.S. and U.K. samples. When political partisanship and ideology are included in the models in the U.S. sample, more conservative respondents report being less willing to comply with the lockdown measures. In the U.K. sample, individuals who do not belong to any political party (N = 192) are more willing to abide by the lockdown measures as opposed to the reference group (those self-identifying with the Labour party).
Finally, for robustness, we conducted a complier average causal effect (CACE) analysis using a 2SLS instrumental approach, where the assignment to the treatment is regarded as an instrument for receiving the treatment and as such is considered to be an endogenous variable in the model. Compliance was measured with the factual manipulation check question in the survey. After being exposed to the stimuli and responding to the questions measuring the variables of interest, respondents were asked: “Do you remember? …. The government (governor, in the US Sample) has extended the lockdown… (1) following a recommendation by a Select Committee (or Congressional Committee, for the US), (2) following a recommendation by a Citizens’ Assembly, (3) following a public consultation, (4) there was no information about it on the news item, (5) I don’t remember.” We constructed three variables by recoding the correct answers as “treatment taken,” and incorrect ones as “not taken” for each treatment condition. Some 88% of U.S. and 91% of U.K. respondents in the minipublic condition got the manipulation check right.16 The findings of the CACE analysis show that the null effect (based on the bivariate relationship) does not remain robust. At p = .089 statistical significance level, a deliberative minipublic exerts a negative effect on people’s willingness to comply with the lockdown measures, which is consistent with the findings of our full model and the model accounting for an interaction term between trust and treatments (see Online Appendix I for more information).
Discussion and limitations
In sum, our cross-national survey experiment yields mixed results for the effect of exposure to a minipublic on people’s compliance with the lockdown measures during a pandemic. In some of our models, it has a null (e.g., bivariate relationship), and in others (e.g., when controlling for important covariates, and moderators; CACE analysis), it has a negative effect on public compliance. In other words, the observed effect is contingent upon model specifications. The effect of reading about a minipublic on public compliance does not differ significantly from that of reading about other decision-making arrangements, such as politician deliberation and public consultation.
The observed negative effect of minipublic in some of our models seems to be driven by respondents with low trust in the capacity of minipublics. This finding has some implications for the literature linking the upscaling effects of deliberative minipublics on nonparticipating citizens’ attitudes, and behavior to people’s trust in such institutions. This study shows that deliberative institutions may act as a double-edged sword in times of a global health crisis. When citizens’ trust in them is high, their recommendations can increase public compliance with difficult decisions. Yet, when their trust is low, it may make citizens question the value of such institutions for the political choices that they consider to require a fast and centralized response from the government.
The most puzzling finding is the relationship between authoritarian preferences and the minipublic, which goes counter to our H4. Citizens who prefer less authoritarian management of the restrictions are more (not less) reluctant to comply with these restrictions after reading that the government consulted a random group of citizens’ recommendations in issuing the new rules. Those citizens with less authoritarian attitudes are even less likely to comply when they hear it is the recommendation of their fellow citizens. This could be because they may value individual autonomy, and the perceived risk may appear lower if coming via other citizens. This finding also has implications for the argument by Lafont (2015) that deliberative processes that do not meet the “all affected” principle can have negative consequences for democratic participation and deliberation, as they do not extend participation to the broader public. It is possible that those citizens who favor decentralized modes of crisis management prefer their voice to be heard in these processes and do not want to be led by a group of other citizens, in particular, where they do not have the opportunity to participate. Our vignettes describe a minipublic without an opportunity for the wider public to provide input into the process. Some seminal minipublics such as the Irish Citizens’ Assembly enabled the public to be part of these processes, by sending their thoughts either online or offline, which were later considered by the members of the minipublic. While we believe that our study captures the way the more comprehensive public receives information about deliberative processes more truthfully, future research should replicate our study by varying the level of hypothetical participation by nonparticipating citizens and examining the differential effect of a minipublic.
Having said that, the effects are small in size; they could be substantively trivial for real-world policy-making and should be replicated in the future. Therefore, one should interpret them with caution. Relatedly, we acknowledge that our subgroup analyses may be underpowered. Low statistical power has a negative effect on the likelihood that observed and statistically significant results reflect true effects. Therefore, future research with well-powered samples should check the replicability of our subgroup analyses, especially regarding the moderating effect of trust in minipublics and authoritarian attitudes on public compliance with difficult political decisions.
Other limitations and scope conditions are worth mentioning. The first relates to the generalizability of our findings. As theorized elsewhere, we assume that our argument is generalizable to global health crises. Future research should investigate whether similar public reactions are observed during crises of other natures (e.g., economic, climate). Similarly, our conclusions rely on just one cross-national experiment. Future research should investigate whether these findings are generalizable to other countries. Furthermore, replicating this study at a different stage of a crisis could contribute to fine-tuning the theoretical expectations about the role of minipublics in crisis situations.
Second, our design doesn’t allow us to test the theorized causal mechanisms underlying the effect of reading about a minipublic on people’s compliance with a difficult decision in crises (e.g., trust in minipublics and perceived crisis severity). One possible way of strengthening this design is to apply a parallel design (Imai, Tingley, & Yamamoto, 2013), where each respondent is randomly assigned to one of two experiments: In one of them only the treatment variable (e.g., minipublic) is randomized while in the other both, the treatment (minipublic) and causal mechanism (trust and/or crisis severity perceptions) are randomized.
Third, a short news item (like any other survey experiment) cannot fully capture the complexity of a deliberative minipublic. Stronger and more comprehensive stimuli—for example, videos showing the process—may engage citizens’ attention more and convince them more effectively. For example, there is limited information about the composition of the deliberative minipublic and no information about the deliberative quality of discussions in our vignette. An interesting question to study, by drawing on a theoretical argument about the similarity mechanism between the minipublic and nonparticipating citizens, is if the composition of the minipublic matters for public compliance. Another question is whether providing information about the quality of deliberations can lead to different behavioral outcomes. Elstub et al. (2021) find that even in times of pandemic, lay citizens can engage in high-quality deliberation online in synchronous forums (see also Smith, Hughes, Adams, & Obijiaku, 2021). Referring to different aspects of discussions and cuing their quality may condition the effectiveness of the minipublic intervention. Future research should research this question empirically.
Fourth, the news source in the manuscript is presented as realistic and neutral. It does not show whether the source has a specific partisan leaning or not. We acknowledge that real-world politics (including partisan politics) in each country’s context can heavily impact the influence of the minipublics in decision-making. The vignettes in our study did not convey how politics factor into the decision-making processes during COVID-19 to the respondents. Future designs should consider these dimensions. Next, in this paper, we examine the self-reported willingness to comply. Future research should attempt to capture the actual, observed compliance (i.e., behavior).
Conclusion
In this paper, we examined the effect of citizen-centered institutional arrangements on citizens’ compliance with one of the most difficult modern-day government decisions—COVID-19 lockdown measures. We bridged the theories of deliberative democracy and crisis management scholarship. We investigated whether reading about a minipublic being a part of political decision-making leading to these measures can influence citizens’ behavior in the middle of a global pandemic. We advanced two sets of mutually exclusive hypotheses underlying the relationship. It was theorized that during a major health crisis, convening deliberative minipublics could decrease people’s perceptions about the severity of the crisis and have a negative backlash on their compliance with a difficult political outcome. We used a framework of eight possible variables impacting crisis attitudes and focused on three (high immediacy, proximity, and perceived transience) to conceptualize why some crises, such as COVID-19, could be perceived as especially severe by the public. We tested our expectations using a cross-national survey experiment fielded in the United States and United Kingdom (N = 2088) in the context of COVID-19. We believe that the crises that score high on the mentioned characteristics—high immediacy, proximity, and perceived transience—may be subject to similar effects that we observe in our study. For instance, at the start of a war that is proximate to a specific population and perceived to be capable of being quickly resolved, citizens would prefer acting immediately rather than convening a citizen-centered decision-making mechanism. In sum, our findings may apply to other crises that are immediate so they arise at speed, proximate so we can see them around us and perceived to be resolvable so transient.
Our findings contribute to a burgeoning body of work studying the macro influences of participatory and deliberative democratic processes in the wider public. To the best of our knowledge, it is one of the first endeavors to theorize about and test the limits of citizen participation mechanisms for public decision-making in crisis conditions. We believe it is paramount for the field of democratic innovations to investigate the scope of conditions of the influence of deliberative minipublics on individuals’ policy preferences, attitudes, and behavior. We echo the argument made by Diana Mutz (2008), Spada and Ryan (2017), and others that we need to understand better the conditions under which deliberation can deliver different democratic outcomes. Having a deep understanding of the conditions under which deliberative minipublics are more likely to fail to impact public opinion is crucial for not only the science of deliberation, but also for policy-making. In this paper, we identify one of the limitations of the influence of deliberation. We show that during a severe health crisis, citizens prefer faster, more centralized, and more robust government responses to tackle the crisis and can be less affected by citizen involvement mechanisms.
An important, but the unresolved question is: What purpose could minipublics serve during major health crises or crises in general? Crisis management scholarship could help point us in the right direction. Coombs (2015) identifies three stages to dealing with any crisis: (a) precrisis stage (preparation and preparation), (b) during the crisis (response), and (3) postcrisis stage (learning and revision) (Coombs & Laufer, 2018, 199). Lafont argued that minipublics might potentially be more beneficial for vigilant and anticipatory purposes (Lafont, 2019, 152–153). As such, minipublics could be democratically valuable in pandemic planning or following a crisis to allow citizens to contest policies they find unacceptable. This is also consistent with the argument by Parry, Asenbaum, and Ercan (2020) that building a robust participatory system can anticipate a crisis and connect the voters to representative politicians, well before the onset of crises could help to avert autocratic backslide, radicalization, and build democratic resilience (see also Schoch-Spana et al., 2021). In a similar vein, citizen deliberation could be paramount in particular in postcrisis situations and can serve to “question […] official narratives, enforce […] inclusiveness and authenticity in public discourse and ensure […] that the state of humanitarian exception does not become the rule” (Curato, 2017, 1012; see also Curato, 2019). Future research should test these conjectures. In general, with this paper, we aimed to start a conversation among scholars and practitioners of participatory and deliberative democracy about the role of citizen-centered political decision-making during major crises.
Funding
This research was undertaken as part of the Provenance project, which is funded by the European Union’s Horizon 2020 Research and Innovation Programme (grant 825227) and by the DCU Covid-19 Research and Innovation Hub. Jane also wants to acknowledge the European Union’s Horizon 2020 Research and Innovation Programme under the Grant Agreement 959234.
Conflict of interest: The authors have no conflicts of interest to declare.
Data availability
Replication materials can be found at Harvard Dataverse: https://doi.org/10.7910/DVN/M2DSSL
Ethics approval
The research was approved by the DCU Research Ethics Committee
Author note
Lala Muradova (PhD, 2021) is Assistant Professor in Comparative Politics at Universitat Pompeu Fabra, Barcelona. Lala’s work have been published or forthcoming at European Journal of Political Research, Political Studies, Conflict Management and Peace Science, among other journals and in edited volumes with Oxford and Cambridge University Presses.
Jane Suiter is Professor in the School of Communications, Dublin City University. She is also the director of DCU’s Institute for Future Media, Democracy and Society and PI on H2020 ICT28 Provenance, a multimillion interdisciplinary project to combat disinformation and has published in a range of journals including Electoral Studies, Science, International Political Science Review and European Journal of Communication.
References
Footnotes
We use deliberative minipublics, citizen deliberation, citizen assemblies, and minipublic deliberation interchangeably.
The most seminal example of minipublics is the Irish Citizens’ Assembly (Farrell, Suiter, & Harris, 2019; Muradova, 2021), whose recommendations prompted a nationwide referendum on the issue of abortion and led to a historic constitutional change in Ireland.
Most recently, scholars have challenged this view, arguing against the so-called “blind deference” of voters to the recommendations of the minipublics (Lafont, 2019, 8; see also Curato, Vrydagh, & Bächtiger, 2020). According to Lafont (2015), public recommendations should only serve as one of the many considerations that need to be taken into account in collective decision-making.
An alternative mechanism is that lay citizens may perceive a degree of consensus among the members of a minipublic on the discussed issue. Perceived consensus “can lead respondents to view the information as beyond specific cleavages (e.g., bipartisan agreement)” (Már & Gastil, 2020, 110) and encourage citizens to “align their prior beliefs with that consensus” (Már & Gastil, 2020, 112; see also Suiter, Muradova, Gastil, & Farrell, 2020). This is consistent with psychological research that shows that the perception of a social consensus can influence people’s attitudes on a variety of issues (van der Linden, 2015; Goldberg, van der Linden, Lesierowitz, & Maibach, 2020).
The discussion of each of these dimensions is beyond the scope of this paper.
Reading about a deliberative minipublic can also activate people who were already unsatisfied with the slow government responses to COVID-19 and as such may exacerbate the public perceptions about the (un)seriousness of the COVID-19. We thank the reviewer for bringing this to our attention.
We conducted a priori power analysis, using Stata 14.2, which yielded the sample size estimate of N = 250 individuals per experimental group for detecting a small effect size of Cohen’s d = 0.20, assuming 5% alpha level, and the statistical power of .90.
See Peer, Brandimarte, Samat, and Acquisti (2017) for comparative advantages of prolific with respect to other survey panels.
An overwhelming majority (89%) of respondents in our pilot study (N = 60) found the news item credible.
Replication materials can be found at Harvard Dataverse (Muradova & Suiter, 2022).
Table C1 (Online Appendix C) presents the list of all of variables of interest, together with their operationalization and descriptive statistics.
We acknowledge that self-reported willingness to comply with COVID-19 measures is likely to suffer from social desirability bias. Recent empirical research yields mixed findings: While Daoust et al., (2020) show substantive social desirability bias among the surveyed Canadians, Larsen, Petersen, and Nyrup (2020) find no such bias among the Danish. We understand that high mean compliance may make the population mean suspect. Yet, due to the experimental nature of data, there is no reason to suspect that responses to specific treatments should be differently affected by such bias.
The item was proceeded by the following general information about citizens’ assemblies: “Citizens’ Assemblies are a diverse group of around 100 randomly chosen lay citizens (similar to a jury) who come together to listen to experts, and each other. They discuss an issue in small groups, with moderation, to ensure all voices are heard and make recommendations to the government. They have been used with success all around the world to tackle complex policy issues such as abortion.”
We use authoritarian attitudes, strong government attitudes, and attitudes toward authority interchangeably in this paper.
The percentage of people who complied with the treatment in the politician deliberation was 75% in the U.S. and 73% in the U.K. samples. The lowest compliance with the treatment was among respondents in the public consultation experimental condition (42% and 44% among the U.S. and U.K. respondents).
