“Do as I say but not as I do”: Influence of political leaders’ populist communication styles on public adherence in a crisis using the global case of COVID-19 movement restrictions

This paper explores the influence of political leaders' populist communication styles on public adherence to government policies regarding COVID-19 containment. We adopt a mixed-methods approach that combines: theory building with a nested multicase study design for Study 1 and an empirical study in a natural setting for Study 2. Based on the results from Studies 1 and 2, we develop two propositions that we further explain theoretically: (P1) countries with political leaders associated with engaging or intimate populist communication styles (i.e., the UK, Canada, Australia, Singapore, and Ireland) exhibit better public adherence to their governments' COVID-19 movement restrictions than do countries with political leaders associated with communication styles that combine the champion of the people and engaging styles (i.e., the US); (P2) the country whose political leader is associated with a combination of engaging and intimate populist communication styles (i.e., Singapore) exhibits better public adherence to the government's COVID-19 movement restrictions than do countries whose political leaders adopted solely engaging or solely intimate styles, namely, the UK, Canada, Australia, and Ireland. This paper contributes to the research on political leadership in crises and populist political communication.


Introduction
Political leadership is essential to managing complex and high-stakes global problems requiring solutions that may be hindered by public resistance to collective actions (Bene et al., 2022). For example, during the COVID-19 outbreak, political leaders, together with public health experts worldwide, attempted to provide valuable advice to the public about necessary collective efforts, such as continuous adherence to new government policies and restrictions on movement between regions and cities, to effectively manage this health crisis (Van Bavel et al., 2020). The authorities played a central role in persuading the public to adhere to COVID-19 prevention measures, although these were deemed costly in terms of the emotional toll of quarantining and social distancing (Grossman et al., 2020, pp. 24144-24153).
Paradoxically, the greatest barrier to the reinforcement of these preventative measures typically originates from political motives. For example, the Trump administration in the White House consistently countered the rhetoric of public health experts and downplayed the health risks of COVID-19. On March 4, 2020, then-President Trump asserted that COVID-19 was like the flu; two days later, he falsely claimed that the situation in Italy was improving and that the US was handling the crisis much better than other developed countries. As late as March 15, 2020, with reported cases rising rapidly, Trump maintained that the crisis within the US was under control (Grossman et al., 2020, pp. 24144-24153). Related to this, the current paper examines the influence of politicians' populist communication styles on public adherence to government policies for COVID-19 containment.
Social media plays an important role in crisis management, especially in terms of delivering information, communicating with the public, and (re)establishing public confidence in government-led crisis management efforts (Rao et al., 2020;Xie et al., 2021;Zhang et al., 2022, pp. 1-24). This is especially important because the general public tends to struggle to find helpful information to follow during a crisis, which was especially the case during the COVID-19 pandemic (e.g., Huang et al., 2022). Social media platforms are accessible to most of the public in a given country; thus, they generate massive data and can have a meaningful influence on public engagement (Bene et al., 2022;P erez-Escoda et al., 2020) and partially help overcome disparities in technology adoption by socially vulnerable populations (e.g., racial and ethic minorities, people who are socially isolated or with disabilities, older adults, and children), as was a key factor during COVID-19 (cf. Dubois et al., 2022). Due to the "distributed, non-hierarchical, and democratic" (Bruns & Burgess, 2011, p.93) nature of Twitter as a medium, political leaders can use tweets as "digital soundbites" to reach wider audiences. Hence, social media platforms are considered the "embodiment of populism" (Bracciale & Martella, 2017, p. 1312. A key issue, however, is just how accurate and helpful such information is and if the communication style used matters during a crisis like  Analysis of the communication styles employed by political leaders is a valuable conceptual tool for exploring the political realm (e.g., Bast et al., 2022;Bracciale et al., 2021;Cassell, 2021;Lee, 2022;Moffitt & Tormey, 2014. Thus, we adopt the populist political communication style framework of Bracciale and Martella (Bracciale & Martella, 2017) as the theoretical lens to analyze political leaders' Twitter posts and to explain their populist political communication styles during their management of the COVID-19 crisis. Based on the communication-centered perspective, Bracciale and Martella Barcciale and Martell (2017) identified four populist political communication styles: (1) engaging (positive and nonaggressive; aiming at engaging supporters on political issues); (2) intimate (positive and personalized; focusing on building nonpolitical discussions around shared emotions and aspects of private life); (3) champion of the people (negative and aggressive; aiming to build political narratives around simplification, position taking, and taboo-breaking); and (3) man of the street (negative and aggressive; aiming to legitimize a political leader as a "man in front of the people" through the use of vulgar language and the exploitation of fear and current news).
To date, research on political leadership in crisis and the use of social media for crisis management has focused on (1) conceptualizing, identifying, and examining key factors and their influences on political leaders' ability to successfully manage a crisis and (2) investigating the information-sharing behaviors of governments, political leaders, or the public on social media in single-country settings. Related research on populist political communication (e.g., Bast et al., 2022;Cassell, 2021;Kluknavsk a & Eisele, 2021, pp. 1-18) focused on investigating the influence on (1) public engagement, (2) public receptivity, (3) voters' attitudes toward gender equality in politics, (4) truth contestation, and (5) proliferation of political ideology. So far, only a few comparative studies (e.g., Doogan et al., 2020;Green et al., 2020;Lyu et al., 2020) have attempted to understand political leaders' use of social media for crisis management; however, no study has empirically examined the role of populist political communication styles in crisis management. Addressing this gap is crucial in considering the outsized "social media bully pulpit" that political leaders have, especially in times of crisis, as exemplified by Trump's use of Twitter during the preliminary stages of the COVID-19 pandemic.
Based on the populist political communication style framework (Bracciale & Martella, 2017), we employ a mixed-methods approach in two studies. Study 1 identifies the populist communication styles of six political leaders in six countries using inductive and deductive thematic analyses of their COVID-19-related Twitter posts. We comparatively examine the cases of the United Kingdom (UK) and Boris Johnson, Canada and Justin Trudeau, Australia and Scott Morrison, Singapore and Lee Hsien Loong, Ireland and Miche al Martin, and the US and Donald Trump from January 2020 to July 2020. Study 2 investigates the moderating effect of political leaders' populist communication styles on the relationship between government policies for COVID-19 containment and public movement by conducting a regression analysis of panel data from  Community Mobility Reports (GCCMR) and Oxford's COVID-19 Government Response Tracker (OxCGRT). We further develop propositions linking political leaders' communication styles and public adherence decisions, implying that certain populist communication styles are more suitable for social media crisis communication media than others. We theorize that countries with political leaders associated with the engaging and intimate or a combination of both populist communication styles exhibit better public adherence to the government's COVID-19 movement restrictions than countries whose political leaders are associated with a combination of the champion of the people and engaging styles. We also posit that countries with political leaders associated with a combination of the engaging and intimate styles exhibit better public adherence to the government's COVID-19 movement restrictions than counties whose political leaders adopted solely the engaging or intimate styles.
Our research makes two key contributions to the literature. First, we contribute to the research on political leadership in crises by shedding light on the more effective and suitable populist communication styles for social media crisis communication. We do so by being the first to conduct a comparative study of political leaders from multiple countries to examine the role of their populist communication styles in crisis management, particularly in terms of public adherence to COVID-19 movement restrictions. Our results provide valuable insights for political leaders who are preparing to manage the next crisis. Second, we contribute to the research on populist political communication by undertaking a comparative case study with a two-staged analysis-withincase and cross-case-of political leaders and their tweets from January to July 2020. Doing so can help us understand the dominant and effective populist political communication styles associated with COVID-19 crisis management.

Theoretical background
To identify the research opportunities that our study addresses, we review the published research on political leadership in crisis and the use of social media for crisis management. To establish our theoretical lens for analyzing political leaders' use of social media for crisis management, we also review the literature on populist communication styles.

Political leadership in crisis
There are three main streams of research on political leadership in crisis (refer to Table A1 in Appendix A for details on the literature review). The first (e.g., Boin & Hart, 2003;Boin et al., 2012;Boin et al., 2013) is conceptual in nature and focuses on developing frameworks for the analysis and assessment of political leaders' crisis-management skills and performances in the pre-and post-crisis periods. The second stream (e.g., Bligh et al., 2004;Davis, 2012;Kapucu, 2009) focuses on empirically investigating four key aspects of political leaders' managing a crisis: (1) their abilities to prioritize needs, select a team of qualified crisis-management experts, and execute a crisis-management plan; (2) their willingness to coordinate international partners to assemble a collective crisis response; (3) their ability to sustain domestic political support during a crisis period; and (4) their use of charismatic language to help others cope with a crisis. The third research stream (e.g., Bol et al., 2020;Dada et al., 2021;Shao & Hao, 2020) focuses on other aspects of political leadership, such as gender, decision to impose lockdowns, proximity of reelections, and public confidence, and their influences on crisis management.

Social media use for crisis management
There are three main streams of research on the use of social media for crisis management (refer to Table A1 in Appendix A for details on the literature review). The first stream (e.g., Ajzenman et al., 2020, pp. 1-47;Chen et al., 2020;Liao et al., 2020) uses content analysis to study governments' use of social media to communicate crisis-related information and its effects on public behaviors. The second stream (e.g., Druckman et al., 2020;Green et al., 2020;Grossman et al., 2020, pp. 24144-24153) conducts a content analysis of the political leaders' official social media accounts to explain their crisis management activities and approaches (i.e., lowballing a crisis, political partisanship, and political parties' consensus for crisis management), as well as the associated influence on public behaviors, such as noncompliance with preventative government measures. For example, President Jair Bolsonaro of Brazil opposed the adoption of social distancing measures, an attitude he clearly conveyed in his social media posts during the pre-election campaign. Such behavior encouraged many of his supporters to disregard preventative measures; in fact, some of the pro-Bolsonaro federal districts did not implement any social distancing measures (Ajzenman et al., 2020, pp. 1-47). The third stream (e.g., Gharavi, 2020, pp. 1-10;Li et al., 2020;Liao et al., 2020) examines the general public's use of social media to share crisis-related information, including tweets and retweets of situation-specific events, to avoid crisis consequences while coping with a crisis, including wearing protective masks and avoiding COVID-19 hotspots. For example, Gharavi (pp. 1-10) investigated the variation in Twitter activities at the state level based on Twitter posts-geolocated in the US-describing the most common symptoms of COVID-19, including cough and fever. A temporal lag (varying from 5 to 19 days) between the increases in the number of symptom-reporting tweets and officially reported positive cases is observed, implying that when pandemic-related behavior patterns appear in the media, they are usually early warnings of local outbreaks.
Thus far, only a few comparative studies (Doogan et al., 2020;Green et al., 2020;Lyu et al., 2020) have attempted to explain political leaders' use of social media for crisis management; however, none have focused on political leaders' populist political communication styles and how they influence public behavior during a crisis. To address this opportunity, we conducted a comparative analysis of the populist communication styles applied by six political leaders from the UK (Boris Johnson), Canada (Justin Trudeau), Australia (Scott Morrison), Singapore (Lee Hsien Loong), Ireland (Miche al Martin), and the US (Donald Trump) during the COVID-19 crisis. We combine Twitter data from these leaders with panel data on the public movement and government restrictions to understand their populist communication styles and theorize about their influence on public adherence behavior. Doing so can provide insights into which populist political communication styles work well for crisis communication and which ones do not.

Populist political communication styles
The political communication style is a crucial constituent of populism (Bracciale et al., 2021) and refers to "a heterogeneous ensemble of ways of speaking, acting, looking, displaying, and handling things, which merge into a symbolic whole that immediately fuses matter and manner, message and package, argument and ritual" (Pels et al., 2003). A political leader's communication style is a useful conceptual tool for investigating various phenomena in the political realm (e.g., Bast et al., 2022;Bracciale et al., 2021;Cassell, 2021). The populist political communication style framework (Bracciale & Martella, 2017) integrates two key elements: (1) content (what is being said), which includes the topic and function aspects, and (2) form (how it is being said), which includes the stagecraft and register aspects. Bracciale and Martella (Bracciale & Martella, 2017) further added three key types of strategies associated with political populism to their framework: accentuating the people's sovereignty, criticizing the elite, and blackballing others (Engesser et al., 2017) (see Table 1). By integrating key concepts from political communication style and populist political ideology, Bracciale and Martella Bracciale and Martella (2017) define four general populist political communication styles based on a communication-centered perspective (see Table 2): (1) engaging, (2) intimate, (3) champion of the people, and (4) man on the street.
(2) Criticizing the elite: expressing anti-elitist and anti-establishment sentiments, thus emphasizing the distance between "us" and "them" in a political, institutional, economic, or general sense. (3) Blackballing others: use of a political narrative that conveys "dangerous others" concepts, thereby establishing a common enemy in groups of populations "above the people." 2022; Cassell, 2021;Kluknavsk a & Eisele, 2021, pp. 1-18) focuses on investigating its influence on (1) public engagement (reactions, comments, and shares), (2) public receptivity (evoking emotions), (3) voters' attitudes toward gender equality in politics (warmth and competence), (4) truth contestation (relativizing the truth and discoursing untruthfulness), and (5) proliferation of political ideology (constructing one's version of reality). In doing so, these studies conducted content analyses of political leaders' Twitter or Facebook posts in single-or cross-country settings, including Austria, Denmark, France, Germany, Hungary, Ireland, Italy, Poland, Romania, Spain, Sweden, and others. To seize this research opportunity, we use the populist political communication style as a theoretical lens to analyze political leaders' Twitter posts in cross-country settings and to explain their populist political communication style during the management of the COVID-19 pandemic.

Methodology
This work adopted a mixed-methods approach that combines theory building for Study 1 and an empirical study for Study 2. Study 1 adopted a theory-building approach based on inductive and deductive analyses with a nested multicase study research design (Eisenhardt, 1989;Eisenhardt & Graebner, 2007;Strauss, 1987) to identify the populist communication styles used by political leaders in their COVID-19-related Twitter posts. Theory building is particularly suitable when limited knowledge about the focal phenomenon exists (Eisenhardt, 1989;Eisenhardt & Graebner, 2007), in this case, the role of populist political communication in crisis management.

Data collection
The data used in Study 1 were collected from Twitter, one of the most popular social media platforms today, with 126 million daily users (Shaban, 2020). In relation to our topic, previous studies have shown that Twitter is frequently used by political leaders as a public communication channel (Bracciale & Martella, 2017;Rufai & Bunce, 2020). Twitter post data were collected using data extraction software that was developed based on Twitter's application programming interface [API (Guides, 2021)]. The Twitter API provides a function called Search User Timeline, which allows a developer to directly query the Twitter server and retrieve tweets posted on the timeline of specific users. The query for the search function was the screen names of the Twitter accounts of interest.
The returned data contained full tweet text and other metadata, such as tweet ID, date and time of the posts, associated links, hashtags, and the number of retweets and likes. Notably, the Twitter API has a quota such that only the last 3,200 tweets posted by each user can be retrieved. The data were collected in late July 2020, and the quota allowance was sufficient to capture all tweets posted by the selected political leaders from January 2020 to the end of July 2020.
To better understand the influence of political leaders on public adherence to public movement restrictions, data were collected from countries that managed the COVID-19 crisis exceptionally well, moderately well, and inadequately based on the total cases and total new cases over time (see Table 3). Based on the screening, Singapore and Lee were selected as a country and political leader that managed the crisis exceptionally well over time; the UK and Johnson, Ireland and Martin, Australia and Morrison, and Canada and Trudeau were selected as countries and political leaders that handled the crisis moderately well over time; and the US and Trump were selected as a country and political leader that failed to effectively manage the crisis.

Data analysis
First, the data on political leaders were manually cleaned to include only Twitter posts that were directly and indirectly related to COVID-19. After filtering, the final dataset included 437 posts for Johnson, 112 for Martin, 203 for Morrison, 167 for Lee, 1,240 for Trudeau, and 1,310 for Trump. Thus, a total of 3,469 tweets were analyzed. Then, the Twitter posts were categorized into five phases of COVID-19 management for each country based on the number of total cases and new cases (refer to Table B1 in Appendix B). The latter framework, developed by the Norwegian Institute of Public Health (Norwegian Institute of Public Health, 2020), allowed us to conduct a more comprehensive analysis of political leaders' influence on public adherence to COVID-19 restrictions. In Phase 1 (Sporadic Cases), most COVID-19 cases were imported into the country in question, and there was a little local transmission of the disease. In Phase 2 (Clusters), most cases consisted of imported cases and local clusters in the country. In Phase 3 (Main Wave), most cases resulted from widespread local transmission in the country. In Phase 4 (Full Epidemic), the cases reached their peak and exceeded the maximum capacity of the country's healthcare system. In Phase 5 (Late Epidemic), the country's cases went beyond their peak and began to decline (Norwegian Institute of Public Health, 2020).
After preparing the data, we conducted the analysis in two stages (Miles & Huberman, 1994). In the first stage, we performed a within-case analysis, which focused on the salient characteristics of the six individual cases and examined the content and sentiment of Twitter posts for each political leader throughout the five phases. The template analysis technique (King et al., 1998) was adopted to inductively analyze emerging themes from Twitter posts in two rounds. We also allowed codes to emerge naturally from the data via grounded theory coding, and these were constantly modified throughout the analysis based on their usefulness and suitability, resulting in the modification of several themes. A final coding template consisting of multiple main themes and subthemes was created for each political leader. The data were also coded for sentiments, including neutral, positive (very positive and moderately positive), negative (very negative and moderately negative), and mixed using NVivo 11's autocode function. In addition, a summary of the political environment for each country was included to demonstrate the context in which the political leaders operated (refer to Appendix D for detailed within-case analyses for each of the six cases).
In the second stage, a cross-case analysis was conducted to identify Using an aggressive tone to engage in simplification, position taking, and taboo breaking to portray oneself as a defender of the people against "the establishment." Man on the street Negative and aggressive Using vulgar language, negative emotions, and instrumental actualization to legitimize oneself in the eyes of the people similarities and differences across the six cases by highlighting emerging patterns via two rounds of analysis. Similar to the within-case analysis, we adopted the template analysis technique (King et al., 1998) for a deductive coding of the Twitter posts based on the five dimensions of the populist political communication style: topic, function, stagecraft, register, and strategy (Bracciale & Martella, 2017). Specifically, the data were coded for each dimension, and values were assigned for the two key factors of the populist political communication style framework (Bracciale & Martella, 2017): (1) communication tone (positive vs. negative) and (2) communication focus (personal vs. political). After assigning values ranging from À2 to 2 for these two factors, the main constructs were plotted on a graph (the x-and y-axis represent communication tone and focus, respectively) to illustrate the interdependencies between the different populist political communication styles. We also conducted an additional comparative analysis of the content and form of the Twitter posts, enabling us to identify their populist political communication styles. Importantly, we acknowledge that certain political leaders may not have had a populist orientation before this period. However, we argue that during the COVID-19 crisis and in certain political circumstances, their communication style took on populist rhetoric. Full details of the cross-case analysis are included in the Results section for Study 1.
To ensure the systematic documentation and rigor of the research process in Study 1, we followed formal procedures for conducting qualitative research (Eisenhardt, 1989;Eisenhardt & Graebner, 2007;Strauss, 1987). This enabled us to establish internal validity, construct validity, reliability, and external validity for this study (refer to Table C3 in Appendix C for details on the rigor criteria of Study 1).

Political leaders who use an engaging style
Boris Johnson: Johnson's posting frequency and percentage of COVID-19-related tweets increased progressively and remained stable, from 9 (14%) posts in Phase 1 to 98 (98%) in Phase 3 and 290 (96%) posts in Phase 5. The topic of Johnson's posts during this period mostly involved issues related to the COVID-19 crisis. For example, he regularly notified the public of government advice and repeatedly reinforced the importance of certain preventative measures, including regularly washing hands, restricting public movements, self-isolation, working from home, social distancing, increasing mental health awareness, COVID-19 testing, and so on. 1 He also frequently reminded the public of the risks of COVID-19 to ensure that everyone understood the facts about its transmission. 2 The function of his tweets was predominantly linked to calling for collective action to contain its spread and endorsing certain groups for their efforts in handling the crisis. Using general and hopeful statements, he repeatedly mentioned the outcomes of UK citizens' collective efforts to minimize the spread of the virus and praised the efforts of healthcare workers in fighting the pandemic multiple times. 3 He also frequently updated the public on easing COVID-19 restrictions, praised them for adhering to the current regulations, and informed them about the available financial support for businesses and individuals 4 (see Fig. 1).
The stagecraft of Johnson's tweets during the COVID-19 crisis was associated with informality and some degree of emotionalization. For example, he attempted to convey hopefulness about the distant future by sharing insights and using informal language. 5 The register of his tweets was mostly neutral and conversational. 6 Specifically, most of his tweets had neutral (57%) and positive (18%) sentiments (see Table 4). The strategy, however, was not evident in his tweets on the crisis.
In summary, most of the populist political communication style dimensions associated with Johnson's tweets during the crisis were characterized by neutral and positive tones, the aim of which was to encourage the public to actively participate in reducing the spread of COVID-19 by adhering to government advice and taking protective measures. Thus, his populist political communication style during this period was primarily engaging, with some intimate elements (see Fig. 2).
Justin Trudeau: Trudeau's posting activity increased steadily, with 12 (3.4%) posts in Phase 1, 271 (46%) in Phase 3, and 590 (38%) in Phase 5. Relative to Johnson, Trudeau posted more COVID-19-related tweets; however, his posting percentage was not as high as Johnson's throughout the phases. The topic of his posts during this period centered on the actions taken by the Canadian government to tackle the crisis. For example, he frequently mentioned the government's financial package that was established to help businesses (e.g., Canada Emergency Wage Subsidy) and citizens (e.g., Canada Emergency Benefit) and to support the Canadian government's efforts to work with other international governments to slow the spread of the virus, secure supply chains, and minimize economic damage. 7 The function of his tweets focused on promoting the actions taken by his government to contain the spread of the virus, reopen the economy, protect Canadian borders, and bring Canadian residents back home. 8 Like Johnson, Trudeau repeatedly called on the public to stick to preventative measures, such as social distancing, good hygiene, selfisolation, wearing masks, mental health support, COVID-19 testing, and so on 9 (see Fig. 3).
The stagecraft of Trudeau's tweets was associated with informality.
1 "Stay home if you have a cough or fever."; "You should stay at home of seven days."; "Use soap and water for 20 s or use hand sanitizer."; "Stay 2 m apart"; "If you are struggling, or are worried about someone, reach out for help." 2 "Coronavirus: anyone can get it, anyone can spread it."; "If you don't follow these instructions, you are putting people's lives at risk."; "You might catch more than the sun at the beach today."; "#StayAlert, control the virus, save lives." 3 "We are past the peak."; "Together we have made it through the peak."; "Thank you for all you are doing to fight the virus."; "Thank you for all that you have been doing so far."; "Thank you to the outstanding NHS staff." 4 "In Leicester, we instituted a local lockdown in order to bear down on stubborn rates of infection. We will begin to relax the restrictions there next week. We can do so because the data is improving"; "The Coronavirus Business Interruption Loan Scheme is now available for small or medium-sized businesses facing cash flow issues. Get loans of up to £5m with 12 months interest free." 5 "We will get through it together.""Together we will beat this.""You are the best of Britain.""We can run the tide against coronavirus in the next 12 weeks.""We have overcome so many challenges in the past." 6 "Read more here on how best to protect yourself and others whist staying at home.""Stay alert by washing your hands regularly, using soap and water.""If you are contacted by NHS Test and Tracefollow the instructions.""We have published a roadmap for how and when the UK government will adjust our responses to the COVID-19 crisis." 7 "The new Canada Emergency Wage Subsidy would cover 75% of employee wages up to $847 per employee, per week."; "The new Canada Emergency Response Benefit will help everyone who has lost their income because of COVID-19."; "South Korea was one of the first countries hit with an outbreak of COVID-19, and last night I spoke with the President about the latest developments." 8 "I want you to know that we're working around the clock, doing everything we can to support our economy and keep you safe."; "Canadians can count on their leaders working together as we get through this."; "We are pulling out all the stops to support you, your family, and your business through this pandemic."; "This funding will help get infrastructure projects underway, give business the confidence to reopen, and make sure you-all hardworking Canadians-can get back on the job."; "@AirTransat will operate flights from Guatemala City and San Salvador on Friday. We will continue to work with the airlines to help Canadian tourists come home." 9 "Social distancing is what we all need to be doing right now. #Flatten TheCurve"; "You can help prevent the spread of COVID-19 by washing your hands, not touching your face."; "Most importantly, if you're feeling sick: stay home."; "In order to open our economy our economy, we need to be able to quickly detect and control any future outbreaks through testing."; "If you're a kid adjusting to a new kind of routine right now, or a senior feeling particularly isolated during this certain period. It is okay to be worried. But we've got your back." For example, he repeatedly used direct, simple, and informal language to explain his government's efforts in dealing with the crisis. 10 The register of his Twitter posts was predominantly neutral and participatory. 11 Like Johnson's tweets, most of Trudeau's were associated with neutral (45%) and positive (24%) sentiments (see Table 5). However, there was no evident strategy associated with his tweets. Most of the populist political communication style dimensions related to Trudeau's tweets were characterized by neutral and positive tones. These messages aimed to encourage the Canadian public to cooperate with the government to heighten awareness of the latest COVID-19 issues and called for a collective commitment to fight the virus. Thus, his populist political communication style was predominantly engaging (Fig. 4).
Scott Morrison: Unlike Johnson's and Trudeau's, Morrison's posting frequency fluctuated, starting with 41 (67%) posts in Phase 1, 23 (100%) in Phase 2, and 92 (90%) in Phase 5. In contrast, the percentage of his COVID-19-related posts increased steadily throughout the phases. As with Trudeau's tweets, the topic of Morrison's tweets focused on government actions in tackling the crisis. For example, Morrison frequently emphasized the Australian government's financial assistance for citizens (e.g., JobSeeker) and small businesses (e.g., JobKeeper), its investments in industrial job creation, and its international diplomacy efforts to establish stronger economic ties with partners. 12 The function of his posts centered on making a public call to action to follow preventative measures, endorsing government workers' efforts during the crisis, and promoting the Australian government's efforts to deal with the crisis. In particular, he constantly reminded the public about existing COVID-19 preventative measures, such as self-isolation, social distancing, mental health support, and washing hands, and prompted the public to download and use the COVIDSafe app to improve the government's contact tracing capabilities. 13 He also frequently endorsed the efforts of police officers and frontline workers in helping manage the COVID-19 crisis and promoted his government's efforts to  10 "The communities that need this funding have now been identified, and money is getting to them they can keep people safe. For people from Victoria to Hamilton, and Regina to St. John's know that help has arrived.";"Businesses can help to ensure a safe re-opening by following the 5 workplace safety steps"; "What is social distancing? Avoid crowded places and non-essential gatherings, avoid greetings such as handshakes, avoid contact with people at higher risk, keep a distance of at least 2 m from others." 11 "We are also funding four Canadian companies working on what may be breakthrough solution for COVID-19 rapid testing"; "Applications for the Canadian Seafood Stabilization Fund are now open!"; "Do wash your hands often."; "Share this with your friends: keep your environment clean." 12 "Over 838,000 Australian businesses have now registered their interest in our JobKeeper wage subsidy of $1500 a fortnight for each employee."; "This $215m investment to support Australian manufacturing businesses deliver 200 job-creating project projects across the country will be a key part of the recovery."; "Delighted to join President Jokowi in welcoming the entry into force of the Indonesia-Australia Comprehensive Economic Partnership Agreement. The agreement unlocks the vast potential of bilateral economic partnership, driving growth and creating jobs in both countries." 13 "Anyone travelling to Australia from overseas will now be required to selfisolate for 14 days."; "Social distancingstaying 1.5m away from each other is our best weapon against this virus."; "Update on the mental health impacts of coronavirus."; "Join me in helping to stop the spread of COVID-19. Download COVIDSafe. An app from the Australian Government." constantly protect the borders and bring Australian citizens back home safe 14 (see Fig. 5).
The stagecraft of Morrison's tweets involved informality and emotionalization. He repeatedly made reassuring statements about the government's efforts to deal with the crisis using simple language and some emotional statements. 15 The dominant register of his tweets was neutral and participatory, 16 and most of his tweets had neutral (46%) and positive (23%) sentiments (see Table 6). Finally, no strategy was evident in his posts.
Lee Hsien Loong: Relative to the other political leaders, Lee posted fewer COVID-19-related tweets, although his posting frequency increased steadily, with 31 (94%) posts in Phase 1, 32 (84%) in Phase 2, and 70 (57%) in Phase 5. As with Morrison's tweets, the percentage of Lee's COVID-19-related tweets fluctuated throughout the phases. In line with other political leaders, the topic of Lee's tweets focused on informing the public about key developments in the country's crisis management. For example, he regularly informed the public about the number of COVID-19 cases and deaths, current measures in place to reduce transmission, the risks associated with transmission, collection spots for government-supplied protective masks, ways to apply for government financial support, and so on. 17 The function of his tweets centered on updating the public on COVID-19 developments, calling for collective responsibility to prevent the spread of the virus, endorsing the efforts of healthcare workers, and promoting the Singaporean government's efforts to coordinate with international partners and actions to prevent the spread of COVID-19. For example, he regularly reminded the public "to do our part" and "remain united" to highlight the collective efforts necessary to prevent transmission, expressed gratitude for the continuous efforts of healthcare workers in treating patients, and promoted his government's efforts to coordinate with international partners to deal with the COVID-19 crisis 18 (see Fig. 7).
The stagecraft of Lee's posts was associated with emotionalization, intimization, and informality. For example, he repeatedly expressed his hope regarding Singapore's short-and long-term prospects through broad emotional and personally insightful statements. 19 The register of his 14 "Thank you also to all the other incredible people on the frontline working to support our communities -including teachers, teacher aids, cleaners, delivery drivers, bus and train drivers, supermarket workers and so many others. You are the best of Australia."; "Thank you to all our Australian Defense Force personnel who are out assisting in our fight against #COVID19."; "Tomorrow we're bringing home around 170 of the Australians who are on the Diamond Princess in Japan and they'll also be quarantined for 14 days." 15 "Australia is not immune from #coronavirus but are well prepared."; "The protection and safety of Australians is our highest priority."; "Australians can be confident the Government is doing everything to ensure their safety and keep our economy strong."; "Our government is going to continue to do everything it can to ensure Australia bounces back stronger on the other side."; "A heartfelt thank you from all Australians for everything you are doing, and will do. We are doing everything we can to #flattenthecurve to ease the burden on you." 16 "It includes a $750 one-off payment to pensioners, low-income families, veterans and other income support recipients, and eligible concession card holders."; "Job Keeper will be extended by six months to the end of March next year and the temporary supplement for those on income support will be extended until the end of the year."; "A welcome move forward on the path to a COVIDSafe reopening across Australia." 17 "Today, there is a large jump in the number of new cases."; "Sadly, two COVID-19 patients in ICU succumbed to the disease this morning."; "Many of the recent COVID-19 cases are work permit holders living in dormitories."; "We have formed the Teck Ghee CARE Task Force."; "The government has implemented support schemes, packages; will make every effort to ensure no one gets left behind."; "Visited the mask collection point at Tech Ghee Care RN Centre yesterday. Masks will be progressively rolled out to 1.37 million Singapore households at the RCs." 18 "Let's stay united"; "Let's all do our part"; "This massive exercise was made possible by the collective effort of the SAF and many Singaporeans."; "My deepest respect to all healthcare workers."; "Show your support for healthcare workers and others working hard to keep Singapore running."; "New Zealand PM @jacindaardern & I spoke tdy. We updated each other on how we were managing COVID-19 & discussed how we can support each other during this time, including keeping supply chains intact between our countries." 19 "Together keep Singapore strong."; "Reassured that it is still business as usual in Singapore."; "Glad that the coronavirus situation has not dampened the festive spirit."; "Our community spirit remains strong."; "Our people's spirit of fortitude in the face of the COVID-19 crises is captured movingly in this series of @govsingapore videos." tweets was neutral and conversational. 20 The majority of his tweets were associated with neutral (42%) and positive (29%) sentiments (see Table 7).
The strategy emerging from his posts accentuated the sovereignty of the people. He frequently encouraged the public to work together to (re) build the city using emotional and personal statements, thus emphasizing his closeness to the people of Singapore. 21 Most of the populist political communication style dimensions linked to Lee's tweets were characterized by neutral and positive tones. His messages aimed to inform the public about issues related to COVID-19 and to encourage Singaporeans to be alert about the latest developments. Thus, his populist political communication style was a combination of engaging and intimate (see Fig. 8).

Political leaders who use an intimate style
Miche al Martin: Martin increased his posting frequency from 13 (73%) posts in Phase 1 to 21 (100%) in Phase 3 to 62 (65%) in Phase 5; however, the percentage of such tweets fluctuated throughout the phases. The topic of Martin's Twitter posts focused on explaining the importance of the public's role in reducing the spread of COVID-19 and prompting them to adhere to current preventative measures, such as wearing protective masks in public, social distancing, washing hands, restricting public gatherings, and so on. He constantly reminded the public to follow the COVID-19 guidelines issued by the government. 22 The function of his tweets centered on updating the public about the availability of government-provided financial packages for small businesses (e.g., JobStimulus) and students (e.g., refunds for accommodation), calling on the public to practice kindness during this period, endorsing the healthcare workers' efforts in helping tackle the crisis, promoting the government's efforts to manage the crisis, and criticizing public misbehavior using irony and sarcasm. Like Johnson, Martin made calls for public kindness using positive and intimate statements, endorsed the efforts of healthcare workers, and promoted government actions to deal with the COVID-19 crisis 23 (see Fig. 9).
The stagecraft of Martin's posts was linked to intimization,  outcome we hope to see."; "Please #stayathomeforSG"; "Temasek Foundation is giving out up to 500ml of water-based hand sanitizer free for all Singapore households"; "We are holding an AMA [Ask Me Anything] tonight at 8 p.m. on our Facebook Page, so do log on if you have questions"; "Do continue to take the necessary precautions even while you are having fun outside." 21 "Tackling COVID-19 needs everyone on the same page. My team will continue working with closely with Singaporeans fight this dangerous threat."; "I look forward to working with our colleagues across the aisle to build a better and stronger Singapore."; "We cannot drop everything and wait for it to pass, we must keep building Singapore."; "Do explore these available job vacancies with an open mind." 22 "You may think you're fit and healthy and unstoppable, but you're jeopardizing people who aren't. Please think of them and follow the advice. We can beat this together."; "If you are to protect our loved ones and the most vulnerable from COVID-19 we must follow the guidelines and work for each other."; "Wearing a face mask out in public has proven to help stop the spread of COVID-19." 23 "We all need to look out for each other, especially those most vulnerable in our communities."; "Compassion is a medicine. COVID-19 is taking so much from us, but it is also giving us something special. The opportunity to come together as one humanity to work together, to grow together."; "We experience these difficult times through #COVID-19, which require from the virtues of unity, courage, and resolve."; "This is one small thing we can all do. Take a moment this evening to talk to your family, our neighbors."   In summary, most of the populist political communication style dimensions associated with Morrison's tweets were characterized by neutral and positive tones. These messages aimed to engage the public in the current COVID-19 issues, promote government efforts to deal with the pandemic, and prompt collective action regarding ongoing preventative measures. Hence, his populist political communication style on Twitter during this period was generally engaging, with some intimate elements (see Fig. 6).  emotionalization, and informality. For example, he repeatedly made informal and emotional statements about addressing issues associated with the COVID-19 crisis, as well as general, positive statements about the future using simple language. 24 The register was mostly neutral and conversational, and even humorous and ironic at times. 25 Most of his tweets had neutral (51%) and positive (22%) sentiments (see Table 8).
However, there was no apparent strategy emerging from his tweets during this period.
In summary, most of the populist political communication style dimensions associated with Martin's tweets were characterized by neutral and positive tones. These tones of communication were used to encourage candid discussions about ways of dealing with the crisis by sharing emotions and personal insights with the public. Thus, his populist political communication style was intimate, with a few engaging elements (see Fig. 10).

Political leaders who use a champion of the people style
Donald Trump: Trump's posting frequency steadily increased throughout the phases, starting with 31 posts (4%) in Phase 1, decreasing to 25 (12%) in Phase 2, and ending with 364 posts (23%) in Phase 5. His posting percentage fluctuated from phase to phase: 4% in Phase 1, 43% in Phase 3a, and 23% in Phase 5.
The topic of his tweets concentrated on highlighting his own efforts, those of his administration, the efforts of the US federal government to manage the COVID-19 pandemic, as well as updating the public on the economic recovery plan related to the crisis. For example, he frequently explicitly outlined the outcomes associated with (1) his own efforts, 26 (2) those of his administration, 27 and (3) those of the government. 28 He also frequently updated the public about his economic recovery plans and actions for dealing with the crisis. 29 The function of his tweets was associated mostly with the self-promotion of efforts to reopen the economy, the endorsement of his administration and government efforts to deal with the crisis, the endorsement of healthcare workers' efforts in fighting COVID-19, and his angst related to the coverage of the COVID-19 crisis by the mainstream media (a.k.a. "Fake News"), which he felt was making the US and its economy falsely look bad. 30 He also repeatedly opposed and criticized the Democrats' performance and policies linked to the COVID-19 crisis 31 (see Fig. 11).
The stagecraft of Trump's tweets during the crisis was associated with simplification, vulgarism, taboo breaking, and instrumental actualization. For example, he regularly made insensitive, offensive, and inflammatory statements regarding the specific actions and decisions of Democrats. 32 The register of his tweets was generally aggressive and provocative, and it was even ironic at times. 33 His Twitter posts showed mixed sentiments: 539 (41%), 309 (24%), and 279 posts (21%) with neutral, positive, and negative sentiments, respectively (see Table 9). The strategy associated with his tweets was a combination of blackballing others, name calling, and criticizing "the elite." For example, he constantly labeled Joe Biden and the Democrats as (1) the "dangerous other" who would be catastrophic for the future of the country and (2) "the establishment." 34 In summary, most of the populist political communication style dimensions associated with Trump's tweets were linked to a mixed tone, which simplified complex issues related to the management of the crisis, Table 7 Sentiment analysis for all phases in Singapore (Lee Hsien Loong).

Codes/Phases
Phase 1  Phase 2  Phase 3  Phase 4  Phase 5   Neutral  16  8  11  2  33  Negative  3  9  8  0  1 0  Positive  11  10  8  2  18  Mixed  1  5  3  0  9  Total references:  31  32  30  4  70 24 "This virus is a very serious issue for the whole country and everyone has their part to play in delaying its spread and helping to protect our communities"; "It is not good enough that students still don't have clarity on refunds for accommodation that they're not using while the colleges remain closed. Needs to be sorted ASAP"; "We can beat #COVID-19 together"; "Together we will do it"" 25 "So many people are making such huge sacrifices. When your kids ask you, what did you in 2020 to fight the virus, you really want the answer to be 'I was an idiot'?"; "A question for those who think that coughing and spitting at people during #COVID-19 is funny. What is wrong with you?"; "A message to those in our pubs and restaurants who are ignoring #COVID-19 recommendations on social distancing." 26 "President @realDonaldTrump DELIVERED: 523 million gloves, 69 million surgical masks, 55 million N95 respirators, 10.5 million surgical gowns, 5.9 million face shields, 10,998 ventilators, and 8,450 federal medical station beds"; "President @realdonaldtrump just signed Phase 3.5 of coronavirus relief legislation into law."; "We are United in our effort to defeat the Invisible China Virus, and many people say that it is Patriotic to wear a face mask when you can't socially distance. There is nobody more Patriotic than me, your favorite President!" 27 "CDC and my Administration are doing a GREAT job of handling Coronavirus, including the very early closing of our borders to certain areas of the world"; "We have perfectly coordinated and fine-tuned plan at the White House for our attack on Coronavirus." 28 "The bill passed in the Senate today will provide an additional $310 billion for the Paycheck Protection Program"; "$49.9 million in Federal Funding is on its way to @DallasLoveField Airport to aid economic recovery"; "$817 million will aid in the economic recovery." 29 "I'm proud to propose a $6.8M award to @WisconsinDOT from @USDOT to complete the last phase of the Merrimac Bridge rehabilitationñ more efficient and lower cost rail car travel over Lake Wisconsin!"; "Big $$$$ headed to the Pittsburgh area! I'm sending $141.7M in CARES Act funding to @PGHtransit from @USDOT to aid in the recovery." 30 "Wow! May retail sales show biggest one-month increase of ALL TIME, up 17.7%. Far bigger than projected."; "@realDonaldTrump authorized @USDA to ensure our patriotic farmers, ranchers, and producers are supported. We are moving quickly to open applications to get more payments out of the door and into the pockets of the farmers."; "Low Ratings Fake News MSDNC (Comcast) & @CNN are doing everything possible to make the Caronavirus look as bad as possible, including panicking markets, if possible. Likewise, their incompetent Do Nothing Democrat comrades are all talk, no action. USA in great shape! @CDCgov." 31 "Dems are ignoring bipartisan calls for more funding from PPP. We need action not politics"; "Republicans were ready to expand funding for PPP last week, but Democrats blocked funding"; "Speaker Pelosi and Senator Scumer [sic] are holding [the funding] hostage, trying to get other priorities."; "You will never hear this on the Fake News concerning the China Virus, but by comparison to most other countries, who are suffering greatly, we are doing very well -and we have done things that few other countries could have done!" 32 "Crazy Nancy Pelosi said I made a mistake when I banned people from infected China from entering the US in January. Tens of thousands of lives were saved as she danced in the Streets of Chinatown (SF) in late January. Biden agreed with her, but soon admitted that I was right.""I NEVER said people that are feeling sick should go to work. This is just more Fake News and disinformation put out by the Democrats, in particular MSDNC. Comcast covers the CoronaVirus situation horribly,""Joe Biden Plagiarizes President @realDonald-Trump is #Coronavirus Plan!" 33 "The Democrats would not have BANNED travel from heavily infected China, especially so early, therefore, far more people would have died. Corrupt Joe Biden now admits this!"; "Why isn't the Fake News reporting that Deaths are way down? It is only because they are, indeed, FAKE NEWS!"; "Oh no, the Dems are worried again. The only one that can kill this comeback is Sleepy Joe Biden!" 34 "Would destroy our American cities, and worse, if Sleepy Joe Biden, the puppet of the Left, ever won. Markets would crash and cities would bum. Our Country would suffer like never before."; "Joe Biden is on a mission to eliminate as many American jobs as possible. No wonder union workers in Pennsylvania tell the Washington Examiner, [Joe Biden] scares me now."; "With Biden drug prices will quadruple! After many years, the game is finally up for Big Pharma, and they're happy about it!" downplayed the health risks of COVID-19, and attempted to portray the Democrats as the "dangerous other" and "the establishment." Therefore, his dominant populist political communication style was a combination of champion of the people and engaging (see Fig. 12).

Summary of political leaders' populist communication style
A summary of the mapping of the populist political communication styles for the selected political leaders is presented in Fig. 13 (Table E1 in Appendix E provides representative tweets for each political leader).

Study 2 methodology
Building on Study 1, Study 2 adopted an empirical approach using dynamic panel data to investigate the relationship between leaders' populist political communication styles and public adherence to government policies for COVID-19 containment. During the crisis, each country had distinct sets of restrictions and political leaders who employed specific populist political communication styles, as shown in Study 1. Furthermore, the ongoing nature of the COVID-19 crisis enabled us to capture the influence of populist political communication on public adherence in a natural setting. Thus, we built a model to explore the correlation between public movements and government restrictions during the crisis. Then, we tested the model based on the available data regarding public mobility and government policy responses to the COVID-19 outbreak in the UK, Canada, Australia, Singapore, Ireland, and the US. The objective of the model was to test the moderation effect of political leaders' communication styles on the relationship between government restrictions and public movements (see Fig. 14), which also informed the proposition development.

Data collection
The data used in Study 2 were collected from the GCCMR and the OxCGRT. The data that captured the public movements and activities  (i.e., GCCMR) were sourced from the COVID-19 Community Mobility Reports developed by Google Inc. (Google. (2021)). These reports were regularly compiled to show how public mobility changed in response to government restriction policies during the COVID-19 crisis. The reports capture movement trends over time across various categories of places, such as retail and recreation, groceries and pharmacies, parks, transit stations, workplaces, and residential areas. The data collection was done daily to capture changes in mobility from a baseline measure based on the period preceding the COVID-19 crisis. Study 2 extracted mobility data from the reports on the UK, Canada, Ireland, Australia, Singapore, and the US to test the model.
The data for measuring government policy responses to the COVID-19 outbreak (i.e., OxCGRT) were sourced from the Government Response Tracker developed by the University of Oxford (OxCGRT, 2021). The platform systematically collects information on several common policy responses made by governments from 180 countries. The tracker's indicators include, among others, containment, economic response, and health systems. Among these, containment indicators are the most relevant to public mobility, because they track government policies related to school closures, workplace closures, public event cancellations, restrictions on gathering size, public transport closures, stay-at-home requirements, and restrictions on domestic and international travel. Therefore, Study 2 focused on extracting data on containment indicators from the OxCGRT.
The public mobility data collected from the GCCMR were aggregated by averaging the data for all five phases for each country. This process was carried out for every location in each country (e.g., retail, recreation, groceries, and pharmacies) and then plotted using line charts for visual inspection and analysis. Next, the containment indicators from the OxCGRT were also extracted and aggregated by phases for each country. Given that several indicators were available (e.g., school and workplace closures), the values of all daily indicators were summed to create an overall measure of the COVID-19 containment level.
To ensure consistency with Study 1, data from the GCCMR and the OxCGRT were collected from February 15 to July 31, 2020. Given that government restrictions began in February 2020 in most of the selected countries, the data collection for Study 2 began on February 15, 2020. The resulting period included data covering 168 days for each of the six selected countries (the UK, Canada, Australia, Ireland, Singapore, and the US). Each record included a country ID, the number of daily public movements, the level of government restriction, and the number of new daily COVID-19 cases. The final dataset consisted of 1,260 records for each of the six selected countries over a period of 210 days.

Operationalization and model construction
In Study 2, public movement was operationalized as the total number of movements recorded by the GCCMR (i.e., movements in retail stores and recreation areas, groceries and pharmacies, parks, transit stations, workplaces, and residential areas) each day for every country. Government restriction policy was operationalized as the total number of restriction levels recorded by the OxCGRT each day for every country. Government restriction policy was organized into seven categories: school closures, workplace closures, public event cancellations, restrictions on gatherings, public transport closures, stay-at-home requirements, and restrictions on internal movements. New cases were  Standard errors in parentheses; significance: ***p < 0.01, **p < 0.05, *p < 0.1. operationalized as the total number of newly confirmed COVID-19 cases each day for every country. Populist political communication style was coded as various dummy variables (engaging, intimate, and champion of the people) based on political leaders in different countries, as discussed in Study 1 (see Table 10). Population was operationalized as the total population of a country in 2020. Table 10 presents the descriptive statistics and bivariate correlations among the variables included in Study 2.

Data analysis
In Study 2, we paired days to form a lagged model, a dynamic panel data structure wherein a series of pairs of two adjacent days were compared. Here, Movement it , LeaderCommunication it , Governmen-tRestrictionPolicy it , and NewCases it respectively represent the daily public movements, government restriction policy in a country, a political leader's populist communication style in a country, and the newly confirmed COVID-19 cases per day in a country. Hence, the general research model is proposed in the form of two equations: (1) where subscripts i and t denote the specific country and time, respectively. The equations represent autoregressive-distributed lag models from panels with a few cross-section units (i ¼ 4 political communication styles), each observed for a relatively large number of periods (t ¼ 168 days). This situation is typical for the use of dynamic panel data. The approach focused on methods that can be used in the absence of any strictly exogenous explanatory variables or instruments and extend easily to models with endogenous variables, thus leading to a widely used class of generalized estimating equations (GEEs) for the analyzed panel data (Hansen, 2007) (refer to Table F1 in Appendix F for details of the descriptive statistics and correlations).
Two separate analyses were conducted in Study 2. The first examined the influence of government restrictions on public movements and new daily COVID-19 cases (see Table 11). The second analysis examined the moderating effect of populist political communication style on the influence of restrictions on public movements and new daily COVID-19 cases (see Table 12). To address potential endogeneity issues, we conducted a difference-in-difference (DID) robustness test for Study 2 (Hill et al., 2021). From October 2021, the governments of all six selected countries dropped their COVID-19 restrictions policies. This change in the COVID-19 restriction policies allowed for further examination of the different effects of government restriction policies on public Movements. Two factors were considered to determine the length of the examination timeframe for DID testing. First, the timeframe could not be too short; otherwise, it would not truly reflect behavioral changes in the public. Second, the timeframe should match the period when the governments in the six selected countries initiated their COVID-19 restriction policies, that is, February 15 to July 31, 2020; otherwise, the timeframe would accumulate too many unrelated activities. Therefore, we performed a robustness check by collecting data from the six selected countries from February 15 to July 31, 2022.
A new variable was created for two types of COVID-19 restriction policies, where "0" represents the situation in which the restriction policies were applied in the period of 2020, and "1" represents the situation in which the restriction policies were not applied in the period 2022. The populations of each selected country were added as a control variable in the model. The DID results indicated that without governments' COVID-19 restriction policies, public movements were more active (β ¼ À9.104, p < 0 0.000, Model 7), consistent with the findings from the data analysis in Study 2 (see Table 13).

Results
To slow the rapid spread of COVID-19, the analyzed governments implemented a wide range of containment measures, including closures of schools, workplaces, and public transport; restrictions on private gatherings, public events, international travel, and internal movements between cities/regions; and stay-at-home requirements. The governments showed considerable national-level variation in terms of which measures were adopted and how quickly they were implemented. Study 2 examined the effects of six different government restriction policies on public movement in Model 1. Further, the overall aggregate government restriction policies on public movement were investigated in Model 2 and those on new cases were examined in Model 3.
In Model 1, the results showed that government-initiated restriction policies on school, workplace, and public transport closures; gathering restrictions; and stay-at-home requirements all had significant influences on public movement, indicating that these policies work effectively on regulating the latter. However, policies on cancellations of public events did not show significant influence on public movements. Interestingly, the policy regarding restrictions on internal movements had a positive influence on public movements, indicating that the former had an opposite effect on the latter. This means that the signal restriction policy on internal movement cannot restrict public movements; thus, to become effective, such a policy should be used together with other restriction policies. Standard errors in parentheses; significance: ***p < 0.01, **p < 0.05, *p < 0.1.
The results also showed that the aggregate government restriction policies had a significant effect on public movement (β ¼ À8.316, p < 0.000, Model 2) and that public adherence to movement restrictions resulted in a decrease in new cases (β ¼ 22.289, p < 0.000, Model 6). In particular, the respective government policy indexes on COVID-19 containment increased gradually throughout the phases, starting in Phases 1 and 2, reaching the highest point in Phases 4 and 3b, and stabilizing in Phases 5 and 4b. In addition, public movements in most of the categories, including retail and recreation, transit stations, workplaces, and groceries and pharmacies, decreased gradually throughout the phases. The decrease in public movements started in Phase 2, reached the lowest point in Phases 4 and 3b, and started increasing again in Phases 5 and 4b (refer to Figures F1 and F2 in Appendix F for the visualization of the COVID-19 restrictions and public movements across the phases).
Next, to further understand the relationship between public movements and government policies for COVID-19 containment, the different communication styles were compared in Model 4 in terms of the relationship between government restrictions and public movements. First, the populist political communication style was coded as a set of dummy variables. For the first dummy variable (i.e., engaging), data were coded "1" if the populist political communication style was engaging and "0" if it was any other style. Similarly, the second and third dummy variables were coded as "intimate" and a "combination of engaging and intimate," respectively. With the dummy variable coding, the reference group was the group coded 0 on all dummy variables. In this analysis, the reference group was a combination of "champion of the people" and "engaging." The unstandardized regression coefficients for the populist political communication style dummy variables represented the difference effects between the mean value of the aggregate government restriction policies of each political communication style and that of the reference group (i.e., champion of the people and engaging). In addition, the influence of the interaction of government restriction policy and populist political communication style on public movements was examined in Model 5.
The results showed that the combination of engaging and intimate populist political communication styles (β ¼ À8.805, p < 0 0.000, Model 5) had the highest correlation with public adherence to COVID-19 movement restrictions, followed by solely intimate (β ¼ À4.750, p < 0 0.000, Model 5) and solely engaging (β ¼ À1.722, p < 0 0.000, Model 5), relative to the combination of the champion of the people and engaging styles. In the UK, Canada, Australia, Ireland, and Singapore, where the political leaders relied on engaging, intimate, or a combination of engaging and intimate styles, the correlation was stronger than in the US, whose political leader adopted a combination of champion of the people and engaging styles. Similarly, in Singapore, whose political leader used a combination of engaging and intimate styles, the relationship between public movements and government policy was stronger than those reported in the UK, Canada, Australia, and Ireland, where the political leaders adopted solely engaging or solely intimate populist communication styles. Therefore, we derive the following: Proposition 1. "Engaging," "intimate," or a combination of "engaging and intimate" populist political communication styles have a stronger positive influence on public adherence to COVID-19 movement restrictions than a combination of "champion of the people and engaging" styles.
Proposition 2. A combination of "engaging and intimate" populist political communication styles has a stronger positive influence on public adherence to COVID-19 movement restrictions than solely "engaging" or solely "intimate" styles.

Contributions to theory and research
This paper makes a twofold contribution to theory and research. First, it builds on research on political leadership in crisis (e.g., Bol et al., 2020;Dada et al., 2021;Shao & Hao, 2020) and is the first to conduct a comparative study of six political leaders from six countries-the UK and Johnson, Ireland and Martin, Australia and Morrison, Singapore and Lee, and the US and Trump. We take a social media analytics approach to examine the influence of their populist communication styles on public actions for preventing a crisis. We develop propositions that explain the moderating role of political leaders' communication styles in public adherence to COVID-19 movement restrictions. The first proposition posits that engaging, intimate, or a combination of engaging and intimate populist communication styles have a stronger influence on public adherence to COVID-19 movement restrictions than a combination of champion of the people and engaging styles. The second proposition states that a combination of engaging and intimate styles has a stronger influence on public adherence to COVID-19 movement restrictions than solely engaging or solely intimate communication styles.
Second, this paper contributes to the research on populist political communication (e.g., Bast et al., 2022;Bracciale et al., 2021;Cassell, 2021) by conducting a case study with a two-stage analysis (i.e., within-case and cross-case) of political leaders and their Twitter posts from January to July 2020. The objective of the within-case study is to understand the different populist political communication styles associated with the management of the COVID-19 crisis. The results suggest that political leaders primarily adopt a combination of populist communication styles: (1) engaging with some intimate elements (e.g., Boris Johnson and the UK), (2) engaging with a few intimate elements (e.g., Scott Morrison and Australia), (3) a combination of engaging and intimate (e.g., Lee Hsien Loong and Singapore), (4) intimate with some engaging elements (e.g., Miche al Martin and Ireland), and (5) a combination of champion of the people and engaging (e.g., Donald Trump and the US) styles. The objective of the cross-case study is to theorize about the moderating roles of different political leaders' communication styles in public adherence to government policies for COVID-19 containment. The results indicate that the populist political is a moderator of the relationship between COVID-19 restriction policy and public movements; hence, the populist communication style can be seen as a useful political tool for crisis management.

Contributions to practice and public policy
Our study offers practical insights into which populist political communication styles are more or less suitable for crisis management, which informs both communication practice and public policy. First, during a crisis, political leaders should adopt engaging or intimate communication styles, or a combination of both, while avoiding negative and aggressive styles such as champion of the people and man of the street, as these styles can lead to public distrust and non-adherence to government policies. That is, to build trust and encourage adherence to government policies during a crisis, political leaders should focus on building non-political discussions around shared emotions and aspects of private life. Negative and aggressive communication styles, which focus on simplification, position taking, taboo breaking, and exploitation of fear and current news, should be avoided by political leaders as they can lead to public distrust and non-adherence to government policies. Second, the combination of populist communication styles used by political leaders is important. For example, Singapore political leader's combination of engaging and intimate styles resulted in better public adherence to government policies during the COVID-19 crisis than other countries where only engaging or intimate styles were adopted.

Limitations and future research directions
This paper has several limitations. First, the generalizability of the results may be limited because only six countries were examined. Future studies should thus examine other countries that exhibit similar and dissimilar political dynamics and compare the outcomes of political leaders' communication styles on public perceptions of and behaviors related to the COVID-19 crisis. Researchers, for example, have looked at the public's reactions to the COVID-19 crisis on Weibo in China (Xie et al., 2021), and similar comparative analysis of the reaction of the public in other countries would likely be illuminating. A global study of the COVID-19 crisis could provide further avenues of investigation associated with regional and cultural differences, which could be reflected in the populist political communication styles of government leaders. Moreover, the context of crisis communication we focus on is involving political communication. There is a parallel stream of research on corporate crisis communication on social media that could also likely benefit from a similar focus of communication style used on social media (e.g., Wang et al., 2021;Zheng et al., 2020).
Second, this paper focused solely on the moderating role of political communication style in public adherence to COVID-19 movement restrictions and did not consider other key moderators. Therefore, future studies should theorize and consider other key moderators to provide deeper insights into political leaders' communication styles and the roles of other moderators in COVID-19 crisis management. For example, understanding the moderating role of a political leader's gender in their COVID-19 response would be a worthwhile dire future direction. This study also did not account for the interaction effects of multiple political leaders' communication styles or the conflicts they may generate. Hence, the outcomes that lead to mixed signals in message recipients could be of interest to future research. Moreover, national-level and individual-level cultural differences have been shown to profoundly differentially affect technology use and communication (Lowry et al., 2011), and such difference were also shown in respect to willingness to use contact-tracing apps during COVID-19 (Altmann et al., 2020). Moreover, indications are that there were profound differences in the public's reaction to the COVID-19 pandemic across countries and cultures (Bajaj et al., 2021;Xie et al., 2021), and there were even differences in reactions on Twitter to the pandemic based on occupation (Zhao et al., 2021). Thus, nationaland individual-level cultural differences should also be investigated for the context of political communication style during crisis management on social media. Another opportunity is to consider the effects of different kinds of crisis communication on vulnerable populations, as such populations have unique needs and challenges during major crises like COVID-19 (cf. Dubois et al., 2022).
Third, this paper used GCCMR and OxCGRT to approximate public movements and government-initiated COVID-19 restrictions, respectively. However, these do not fully capture the environmental complexity associated with a certain government's response and public reactions to COVID-19. Thus, future studies should consider the use of mobile data and official government health site data to provide more accurate insights into public movements and government responses. Furthermore, this study did not consider the moderating role of the number social media posts on influence on public adherence to COVID-19 movement restrictions; thus, future research should further explore whether the number of social media posts directly translates to a more effective influence on public adherence to government policies or restrictions.

Conclusion
This paper investigated the moderating role of leaders' different political communication styles in encouraging public adherence to government policies for COVID-19 containment. It adopted a mixed-methods approach that combined (1) inductive theory building with a nested multicase study design for Study 1 and (2) an empirical study in a natural setting for Study 2. The approach in Study 1 featured a two-stage analysis (i.e., within-case and cross-case) of Twitter posts from six political leaders in six countries. In Study 2, a model was built to explore the correlation between public movement and government restrictions during the COVID-19 crisis from January 2020 to July 2020 using data from the GCCMR and the OxCGRT. Based on the results of both studies, propositions were developed explaining the moderating roles of distinct populist political communication styles on the influence of government restrictions on public movements.