Leaders can be essential in setting the tone of deliberation in the public sphere (Chambers 2009; Dryzek 2010). But can leaders’ discursive style influence the wider public sphere? This exploratory research studies whether leaders influence the quality of citizen deliberation. Since mass communication usually mediates leader-citizen interactions, there is a need to examine deliberation in systemic terms and consider the relevant technological affordances. To explore this, this study looks at whether a political leader’s use, or lack thereof, of deliberative communication style online, in turn, impacts citizens’ use, or lack thereof, of deliberative dialogue online, and if so, if this is dependent on the online arena. These questions are examined by analyzing the deliberative quality of former New Zealand Prime Minister Jacinda Ardern’s and former opposition leader Judith Collins’s discussions during three periods and in two online arenas. It then analyzes the deliberative quality of online citizen discussions and whether the leaders’ impact on citizen deliberation is dependent on the online arena.

This study presents theoretical arguments for leaders to participate in legitimation processes as part of the response to the problem of scale, introducing a communication model for leaders to support public deliberation. According to the proposed model, the leader’s ability to affect the deliberative quality of citizens’ discussions is mediated by their level of influence within that space. While exploratory, the study suggests that when leaders have greater access to support deliberative dialogue and actively engage as both participants and facilitators, they can foster more deliberative discussions among citizens. This is an important finding with significant implications for political communication literature, helping in building a communication model for leaders that supports deliberation in the public sphere.

To achieve the purpose of this article, leaders’ speeches were analyzed, along with the corresponding sample of citizen online discussions. Former Prime Minister of New Zealand Jacinda Ardern and former opposition leader Judith Collins were chosen because both use Facebook Live to communicate with citizens in real-time, have different communication styles, and are female leaders within the same geographical area. The leaders’ gender was held constant because perceptions of the effectiveness of female and male leaders can vary according to position and institution (Paustian-Underdahl et al. 2014). Additionally, New Zealand and its political leaders were considered as a useful context for this exploratory research for three main reasons. First, New Zealand is a country broadly committed to democratic norms and values, being among the top five strongest democracies in the world (Levine 2004; Willige 2017). Second, former Prime Minister Ardern has been globally acknowledged for her leadership style (Simpson et al. 2022), which was tested when faced with a massive shooting and the COVID-19 pandemic, both events that created discussion and opposing views to her policies (Every-Palmer et al. 2021; McGuire et al. 2020). And third, just like in other parts of the world, New Zealand society is getting increasingly intolerant and polarised (Porat 2023; Jaworska & Vásquez 2022), an issue exacerbated by social media that more experienced politicians argue somehow helps explain Arden’s resignation (Clark 2023). While these features shed some light on the institutional context that supports deliberative dialogue in New Zealand, highlighting their regulative, normative, and cultural dimensions (Scott 2013), further research is encouraged to deepen the role that institutional factors play in citizen deliberation.

Video clips of interviews conducted on news and radio channels, as well as sessions in the House of Commons, and their Facebook Live sessions and tweets between July 2020 and June 2021 were assessed to understand Ardern’s and Collins’ communication styles and use of deliberative talk. Additionally, Ardern’s interviews and Facebook Live sessions while campaigning before taking office in September 2017 were reviewed to assess for differences in her communication style under this context, of which none were found. Thus, it was decided she would serve as an effective comparison. Since Collins’ Facebook Live sessions only took place during the campaign period, two sets of interviews with the leaders from a national show were included.

Three separate contexts were considered: 1) during the campaign period in August 2020, 2) during the non-COVID lockdown period and on a non-COVID subject in November 2020, and 3) during the COVID lockdown and discussing COVID-related subjects in March 2021. The subject of the discussions was held constant where possible: the housing bubble during the non-lockdown COVID period and COVID during the COVID period. During the campaign period, both conversations had the broader COVID subject. Ardern discussed the decision to push back the election date, vaccination, and initiatives to buffer the impacts of lockdowns. Collins discussed the impact of COVID on farming, current regulations, and international workers’ VISAs. Discussions coded from Facebook comments (citizens) and leaders (videos), female political leaders, New Zealand, and interviews chosen close together in date for the comparison period were held constant.

Three interviews from each leader were analyzed, along with the corresponding comments made by citizens on Facebook, for a total of six leader interviews and six citizen discussions. The interviews and Facebook Live sessions that were closest in date were chosen within the date ranges of the three contexts as comparisons. The interviewer was held constant and, when possible, the specific topic of discussion under the broader topic of COVID-19.

Video posts were transcribed, and the number of comments and views were recorded. Each video post’s comments were then copied sequentially into a spreadsheet. The first 100 sequential comments were coded from each television interview comment section. Since comments often followed a discussion thread and responses to leaders’ video comments, and because this study aims to explore deliberation within the context of the conversation, this study did not make a random selection of comments as similar studies have done (for example, Esau et al. (2021)). The first 200 sequential comments out of 395 were copied for analysis from the comment section of Collins’s Facebook Live video. Because Ardern’s video had over 10 thousand comments and extracting them in a sequential order was not possible, 418 of the first comments marked as most engaged (number of replies) in sequential order were copied into a document. Ultimately, a final sample of 1,018 user comments and 82 leader speech instances were analyzed across three different contexts.

Comments from the leaders and citizens were analyzed for their deliberative discourse quality using qualitative content analysis based on a combination of a modified version of Steenbergen et al.’s (2003) discourse quality index (DQI) and elements from Stromer-Galley’s (2007) deliberative coding scheme. The modified version of the DQI was developed to better align with the current research in deliberation: the addition of narrative in the form of storytelling and personal impact to the justification of claims (Boswell 2013; Young 2002). Additionally, similar to the modification made by Elstub et al. (2021), all speech instances were coded, not just those making demands as the original DQI requires (Steenbergen et al. 2003) to capture the overall level of respect within the discussion. DQI incorporates a participation score to capture the ability of participants to participate freely in the discussion. A score of 1 is provided for a speech instance where normal participation was possible and 0 when it was interrupted by a speaker (Steenbergen et al. 2003). Because participants’ speech instances are not interrupted in this arena (text-based), the participation code was removed from the DQI score (Elstub et al. 2021). The scores from each category of the DQI were added to create an index on a scale from 0–1.

The modified DQI, despite the addition of narrative to content justification, lacked important measures to understand the context of the conversation that is also important for deliberation, such as group heterogeneity (Sunstein 2000), the degree to which the discussion remained on topic (Stromer-Galley 2007), and engagement or reciprocity (Habermas 1984, 1985). As such, additional coding measures were added based on Stromer-Galley’s (2007) coding scheme, including the number of participants who agreed and disagreed with the leader’s position and the number of comments on topic. Furthermore, to supplement the measures of engagement and reciprocity, the number of comments that took the form of metatalk, the number of social talk instances participants engaged in, and the number of replies to other citizens’ comments were coded. Table 1 summarises the aspects coded in both the leaders’ and citizens’ comments and the measures used to capture these aspects. Coded speech instances and comments from citizens and leaders were analyzed using the individual measures and the modified DQI. The study had one coder; thus, intercoder reliability was not obtained. To account for this, the coder completed several independent revisions for each discussion to ensure consistency.

All the discussions contained a heterogeneous mix of citizens with views for and against the opinion of leaders (Table 2), discussions were on topic for much of each discussion, and citizens displayed reciprocity by responding to one another and the leader.

Though the number of participants who replied to others decreased in both leaders’ Facebook Live discussions with citizens to 32.14% (Collins) and 35.31% for Ardern (Table 3), more citizens posted social greetings to Ardern.

Little social talk was present within the news mass media conversation threads (1%–3%). Similarly, leaders used minimal social talk during these. The leaders’ use of social talk increased in their Facebook Live talks, as did citizens, with Collins’ speech instances, comprising 17.14% social talk and the corresponding citizen discussion 20.5%. Ardern used a lot of social talk (80%), and citizens responded using a higher amount as well (51.2%) (Table 3).

Only Ardern used metatalk during her COVID period interview, clarifying the interpretation from the interviewer of comments she had made. Citizens’ use of metatalk varied (5%–50%), with the highest found in response to Ardern’s interview and the lowest in response to Collins’ interview from the Non-Lockdown COVID period (Table 3).

The highest overall deliberative quality (DQI = 0.59) was associated with the leaders’ Facebook mass media arena when Ardern also scored high on deliberative quality (DQI = 0.66). Comparatively, when the leader scored low on deliberative quality within their Facebook mass media arena (DQI = 0.34), so did citizens (DQI = 0.29). Similar results were not observed in the news mass media arena, where DQI scores were variable and did not always align with the leader using more or less deliberative talk (Table 4).

Regarding features of the DQI, reason-giving was captured by the level of justification or completeness of the justifications made by demands or appeals and the content of the justifications, whether these appeals or demands were made using narrow group interests, the greater good, the difference principle (Steenbergen et al. 2003) or by a statement of personal impact (Dryzek 2000; Young 2002). In the news mass media arenas, leaders made most appeals using sophisticated justification. While scores varied, citizens also used justifications to support their statements in all discussions, and most did so using neutral statements. The only discussion with a high number of citizens making statements containing no justifications (43.27%) was the citizen discussion in Collins’ Facebook Live discussion. Similarly, in this discussion, both the leader (45.45%) and the citizens (26.92%) made more explicit statements concerning narrow group interests rather than neutral statements, referencing the common good, or statements of personal impact. In contrast, Ardern made no statements concerning narrow group interests, and citizens, in response, made a minor percentage of appeals concerning group interests (2.01%) (Table 5).

Respect was captured according to the DQI. Results were as expected in the leaders’ Facebook Live arena: the leader displaying high levels of respect showed citizen discussion displaying the same (Table 5). In the news mass media arenas, leaders were respectful toward other groups. Collins, however, degraded 25% of counterarguments that she disagreed with in the non-lockdown COVID period. Citizens’ level of respect varied regardless of the leaders’ use of respectful messaging. In the leaders’ Facebook Live arena, while Ardern’s communication remained consistent with the other arenas, Collins displayed significant differences in the level of respect she conveyed in all categories. Notably, 53.57% of Collins’ discussions showed no respect toward other groups, 73.91% showed no respect toward the demands of others, and 100% of her comments ignored the counterargument of others. The citizens’ discussion in response to Ardern’s Facebook Live arena, where she displayed zero instances of disrespect, had the highest levels of respect of any of the talks coded. Citizens’ respect was more explicit, and the majority responded to counterarguments of those they disagreed with neutrally or in a way that explicitly valued the opposing opinion.

The percentage of comments made aiming for consensus by displaying constructive politics, where speakers made mediated proposals across all discussions, was variable among leaders ranging from 0%–100%. Citizens making mediated proposals was low (3.33%–8.43%), with the highest levels observed in discussions where leaders made them.

The purpose of this study was to explore if leaders influence the quality of citizen deliberation by examining whether their use, or lack thereof, deliberative communication style online, in turn, impacts citizens’ use, or lack thereof, deliberative dialogue online and if so, if this was dependent on the online arena.

The citizens’ discussions in both arenas met the criteria of deliberation. Citizens’ discussions contained opposing views, had relatively high levels of reciprocity and engagement on topic, and used justifications to support statements. This means that citizens heard diverse perspectives, responded to each other on-topic, and did so with justifications to support their statements, rather than simply talking past each other, all criteria for deliberation and deep consideration of issues where opinion change can take place (Stromer-Galley 2007). This finding was unlike Halpern and Gibbs’ (2013), who found that although Facebook provides a deliberative space for political discussion, citizens deliberate superficially.

When the leaders stated their appeals without providing justifications, citizens also made a higher number of comments without providing any justifications. Collins’ Facebook Live was the only talk where a leader made comments without providing any justifications. She also produced more comments that provided an inferior level of justification. By comparison, all comments from Ardern during this period contained qualified or sophisticated levels of justification. Generally, citizens’ comments responding to the leaders’ posted videos on Facebook saw a lower level of justification than the discussions from the TVNZ Breakfast interviews. This appeared to be partly because the threads responding directly to the leaders’ Live videos contained many comments without appeals or claims such as greetings or praise.

Looking at the content of justifications, the leaders mainly made neutral comments, followed by statements of the greater good. Citizens also commented neutrally. That is, they did not reference any specific group interests, nor did they reference the common good or make statements of personal impact/storytelling. This was not the case for citizens responding to Collins’ Facebook talk, which had increased comments referencing narrow group interests, as did Collins. Although this talk was about the impact of the current policies on agriculture, there was little in the way of stating impacts in a way that related to the common good or difference principle. Instead, many comments from Collins used an inflammatory tone with unjustified statements.

All citizens’ discussions justified most of their comments, supporting their opinions and considering, through their responses, one another’s statements. The aim for consensus, where citizens made mediated proposals in an attempt to compromise or reconcile (Steenbergen et al. 2003), was fairly low throughout all discussions. A finding consistent with other online deliberations (Esau et al. 2021; Stromer-Galley 2007).

Level of respect was the most salient factor of the DQI score, followed by level of justification.

Turning to the research question: whether leaders’ use of deliberative talk in discussion elicited greater deliberation amongst citizens online and, if so, was it dependent on the arena, the results suggest that it depends. The discussion with the highest overall deliberative quality was associated with the leaders’ Facebook mass media arena when the leader also scored high on deliberative quality. When the leader scored low on deliberative quality within their Facebook mass media arena, so did citizens. These findings were not observed in the online News mass media arena where DQI scores were variable and did not always align with the leader using more or less deliberative talk, which could be for a variety of reasons.

In the News mass media arena, the leaders’ use of deliberation is largely limited to a participant in dialogue with the interviewer. Here, the interviewer leads the conversation, and the leader responds to their questions. While the leader can respond in varied ways, they do not have control over the tone of the discussion or whether the interviewer raises issues using frames that highlight polarization or conflict. For example, during the opposition leaders’ interview during the non-lockdown COVID period, the interviewer introduced the topic by stating the level of criticism from both sides of the political spectrum of Ardern on ‘her government’s apparent lack of effective action on the housing crisis as house prices continue to go through the roof’ (Breakfast 2020a: 0:00:08). The topic, before the discussion started, was already framed as conflictual and unfair. Additionally, in the interviews with Collins, the interviewer, Jenny-May Clarkson, did not push back on any of her responses in any of the talks coded. By comparison, Ardern received considerable pushback and critical questioning. While the deliberative style of Ardern’s responses may have been enough to influence citizens to discuss with higher deliberative quality in the other talks, the frustration from both Labour and National supporters (Ardern’s and Collins’ parties, respectively) may have had a greater impact on the resulting quality of the discussion.

Alternatively, in other arenas where the leader controls the content, such as their social media page, they can consistently ensure a deliberative frame is delivered to citizens throughout the entire Live discussion. In this arena, leaders also have more freedom to guide citizens to abide by ‘rules of engagement’ much as a facilitator/moderator would and support understanding of different viewpoints by the frames and rhetoric they offer. In this way, leaders using this arena can model as participants in dialogue and guide it as facilitators of deliberation. Thus, the following framework is proposed where a leader’s ability to influence the deliberative quality of citizen’s discussions is mediated by the leader’s level of influence on the deliberative space of that discussion (Figure 1).

This proposed model aligns with Habermas’s (2006) view of leaders at the center as coauthors and distributors, along with mass media, of public opinion by constructing and articulating prevailing public opinion back to citizens, who in turn respond to the issues articulated by elite discourse and set the frame for the range of what decisions they would accept as legitimate. Together, politicians, in constructing and articulating public opinion to citizens, and citizens in responding to these articulations, create a filtering and self-regulating mechanism for creating public opinion (Habermas 2006). The politician’s use of deliberative communication would positively affect the normative requirements for generating more considered public opinion through framing relevant issues, including required information, and supporting appropriate contributions (Habermas 2006). First, by framing the relevant issues or range of acceptable issues for citizens to consider in decision making on an issue, leaders act to define the relevant issues for consideration. Second, leaders support the inclusion of required information by supporting citizens by creating an environment where other perspectives are more likely to be heard and considered to support access to the information prerequisite for reasoned opinion formation. Third, leaders can support appropriate contributions by supporting increased mutual understanding among citizens, so a plurality of views is included and considered. The proposed communication model for leaders supporting deliberation in the public sphere suggests that the ability of the leader to influence the quality of citizen deliberation using the mechanisms presented by Habermas (2006), is dependent on the leader being able to access these mechanisms in that arena and thus dependent on their influence within the deliberative space. Specifically, environments where leaders can guide discussions toward mutual perspective-taking and understanding as participants, facilitators, and rhetoricians of deliberative communication.

Other factors can explain the variability of DQI scores observed apart from the leader’s influence over the discussion. For example, comparing group heterogeneity, the News arena likely had more diversity in political opinions than the leader-led Facebook arena, which, despite still comprising a heterogeneous group discussion, saw a greater number of each leader’s followers. Additionally, more citizens commenting in the News mass media arena disagreed with the views of the leader being interviewed. It could be that the leaders’ influence over deliberative quality, if it exists, is bidirectional. As Lisa Disch (2011) explains, leaders are at one influencers of and influenced by citizen opinions and orient forward in speculation. Citizens elect representatives who they feel would advance their desired direction and preferences (Mansbridge 2009), even if those preferences are co-created. In this sense, potential constituents who are more deliberative may seek more deliberative leaders, with both updating their opinions accordingly, just as political leaders may adopt a less deliberative style because that is what their constituents want. Future research can seek to answer this by measuring how a leader’s deliberative communication influences the discussion quality of randomized participants.

Alternatively, the variability could be due to the topic of discussion itself and the degree of controversy (Ziegele et al. 2014). Leader’s scoring lower on the DQI also saw corresponding lower DQI in the citizen discussions in the News mass media arena except for the housing bubble talk. In the case of the housing bubble talk, Ardern obtained the highest score in terms of deliberation. At the same time, the subsequent citizen discussion received the second-lowest score among the analyzed conversations. This might reflect the complex nature of the housing crisis and its impact on the perspectives of New Zealanders (Giovannetti 2022). The housing bubble talk was different in that it was contentious on both sides of the political spectrum, including Ardern’s party (Breakfast 2020a). So, while Collins’s talk on the housing bubble from citizens remained more deliberative, especially concerning the level of respect, she presented suggestions on how to fix the bubble, and her complaints aligned with citizens on both political sides. Ardern, on the other hand, ran on a platform that addressed housing prices (Manon 2020). During this interview, Ardern stood by the current measures while attempting to explain the external forces for the current sharp increase in house prices as being impacted by the COVID-19 pandemic and how this similar issue was occurring in other comparable OECD countries. The other coded talks had less cross-party conflict, with clear supporters on each side. Prior content analysis of news articles contradicts these results, where topics with clear sides increase incivility (Coe et al. 2014). However, the same research by Coe et al. (2014) found that news articles containing quotes from a high-profile source with an identifiable partisan leaning, such as Barack Obama, generated the most incivility in online discussions. It is possible that Ardern’s high-profile identifiable partisan, coupled with her explanation of the deteriorating housing market despite her mandate to address the issue, played a more significant role in the ensuing incivility of the citizen discussion on housing, despite the absence of a clear partisan divide between Labour and National supporters on the housing issue. Furthermore, housing is an issue that affects millions of New Zealanders across the country, who most probably voted for Ardern based on her housing platform, a promise her government had to work on, something that opposing non-governing parties do not have to achieve.

Another possible factor is the role that previously received cues from a political leader may have played in the level of disrespect observed in the citizen discussion. While Ardern’s cues to respect differences toward groups and others’ opinions have remained consistently deliberative before and while in office, Collins’s rhetoric fluctuated depending on the platform she used. Of supporters who would have listened to her on other platforms like Twitter and her personal Facebook page, they would have heard primarily bonding rhetoric from Collins that, in addition to energizing those with similar views and beliefs, showed disrespect toward Labour and ‘the left.’ For example, by attacking Ardern’s character and sowing the seeds of distrust rather than pointing out areas of policy disagreement and why:

This aligns with research showing that higher levels of incivility among politicians are linked to an unwillingness to compromise (Mutz 2007), lowers trust toward politicians, and increases animosity toward the opposing party (Skytte 2021). It should be noted here that opposition leaders have, as Waldron (2012) points out, an institutional obligation to loyal opposition that is important to democratic systems. Although this study assesses communication against the standards of deliberation and points out the seemingly negative effects that come from displays of incivility from leaders, it remains to be addressed how these interactions may negatively or positively impact the wider deliberative system.

Upon closer inspection of the respect scores from the housing bubble discussion, the higher number of citizens showing no respect toward other groups or demands were those who did not agree with Ardern and/or those who agreed with the opposition party. The majority of citizens who agreed with Ardern also responded in a neutral (respectful) manner toward counterarguments from those they did not agree with. Despite these differences, proponents and those in opposition to Ardern had higher levels of disrespect than in other discussions where the leader displayed higher deliberation. Previous research shows that when citizens are exposed to counter attitudinal incivility from political elites, that is, elites with opposing attitudes from their own, in this case, Ardern’s supporters hearing incivility from Collins, it increases condemnation toward the other group by triggering defensive anger (Gervais 2018). In this discussion, both Ardern and Collins’s supporters displayed higher levels of disrespect, and Ardern’s supporters could have been triggered by defensive anger of prior exposure of incivility from Collins.

Regardless of the reasons for the variability in the quality of citizen deliberation observed in the News arena, deliberation, at its core, is about having conversations with those who could be affected by a decision. In the News arena, leaders had no opportunity to engage with citizens or respond to their counterarguments. Notably, previous studies of online deliberation highlight the importance that facilitators (Strandberg, 2015) and passing on rules of engagement (Esau et al. 2021) have on the quality of deliberation. Moreover, the presence of a facilitator was of greater importance than group heterogeneity: even homogeneous groups that tend to polarize did not when guided by a facilitator (Strandberg et al. 2019).

Basing the leader’s level of influence within the backdrop of a participant in dialogue aligns with Cho et al’s. (2009) OSROR model of communication effects that demonstrates that the influence of opinion leaders is generated in the interpersonal political discussion and cognitive reflection of citizens. Through a two-step flow, political conversation amongst citizens is an intermediary between political messages from leaders and public opinion. This study differs from the OSROR model by developing a theory for leader-led political communication that includes the influence of the democratic principles of normative reasoning. This may be supported by two mechanisms outlined in the Elaborative Likelihood Model of Persuasion (Cacioppo & Petty 1984), which shows that increasing the personal relevance of messages to the listener and the strength of arguments can act to increase the use of more elaborative, central route processing required for paying careful attention to arguments that conflict with their pre-existing prejudices and rely less on heuristics, making opinion change more likely (Petty & Brinol 2002). The leader increases the personal relevance of their messages by using deliberative talk within the conversation, legitimizing others’ viewpoints, values, and identities. When combined with the potential for leaders to enhance the likelihood of citizens perceiving stronger arguments during discussions and relying less on heuristics, it can contribute to fostering an environment that values diverse perspectives and legitimizes citizens’ experiences. These may also support social learning processes toward diverse groups as the important precursor to deliberation (Kanra 2020). Along with other modes of persuasion outlined by Kruglanski et al. (2004), such as decreasing the processing difficulty by disseminating information on complex policy issues and decreasing ego-defensiveness by creating cues of safety for sharing different opinions, are other ways the leader may influence greater likelihood for the elaboration required to consider opposing views. Facebook itself may increase the perceived accountability of citizens when sharing their opinions, another route for increasing the likelihood of elaboration (Tetlock 1985; Kruglanski et al. 2004).

There are some limitations to this study. First, only one talk each from the leader-led mass media arena was included in the analysis, so this is an exploratory research. Further, it was not possible to hold both the date and topics constant for the leader-led mass media arena regarding the study design itself, although they both included topics under the umbrella of COVID. The analysis did not include talks from the leader-led mass media arena outside the campaign period, which may have generated different results. Participants also differed in the talks coded, so that could not remain constant either. Third, the study used as independent variable the institutional input (information provided, type of communication) and the communicative throughput (level of deliberation measured as civility, reason-giving, constructiveness) as the dependent variable. It excluded any measure of the productive outcome of deliberation (consensus, changes in trust, opinion change) (Friess & Eilders 2015). Finally, the study did not look at how the institutional context of New Zealand may have shaped the quality of deliberation. Reflection on the institutional factors that may impact the influence of the leader on deliberation should be explored in future work.

This exploratory study suggests that when leaders have greater access to support deliberative dialogue by taking on the role of both participant and facilitator, they can influence citizen deliberation.

This study presents theoretical arguments for leaders to participate in the legitimation process in civil society as part of the response to the problem of scale. It suggests the potential for continuing to explore the leader’s influence over deliberation and developing the mechanism to draw on for supporting more authentic public opinion. A communication model for leaders supporting deliberation in the public sphere is presented, proposing that a leader’s ability to influence the deliberative quality of citizen discussions is mediated by the leader’s influence of the deliberative space. It may be that leaders, as participants and facilitators within citizen deliberation, support creating a safe space for citizens to share and learn from each other by promoting listening and understanding that goes beyond simply messaging how citizens should respond to demonstrating it by providing the road map for how to. This study invites further research to determine whether leaders can, in dialogue with citizens, influence the quality of deliberation.

Ultimately, to keep the deliberative project practical, it must come to terms with leaders as an inevitable influence within the deliberative system, for better or worse. Thus, it should seek to develop the knowledge base to guide leaders’ communication in support of that goal.