US Political Leadership and Crisis Communication During COVID-19

Abstract The research explored the role of political leadership in response to the novel coronavirus (COVID-19). The researchers conducted a political discourse analysis on 239 transcripts from the press briefings of President Trump and seven U.S. governors to determine the extent to which the research subjects used effective crisis leadership and communication. These results suggest that President Trump and Governors DeSantis, DeWine, Ducey, and Ivey are particularly vulnerable to political fallout for their handling of COVID-19 because stakeholders might view them as inattentive to the crisis and ineffective in their policy responses. Governors Cuomo, Newsom, and Whitmer may be in a better position to avoid fallout due to their information-seeking, hands-on approaches, which some will deem as competent and appropriate to the threat (although others may see their efforts as over-controlling). The research demonstrated how discourse analysis could predict political behaviors and blame assignment for crisis responses.


PUBLIC INTEREST STATEMENT
Political leaders have a duty and primary responsibility for using governmental resources and communication channels to protect their constituents. Coordinated leadership at federal, state, and local levels is critical to ensure that resources are distributed, and that information is available for people to make informed decisions. Presidential and gubernatorial leadership plays a critical role in U.S. public policy decisions that have life and death consequences for Americans. Politics can play a role in crisis communication and response. We mapped political leadership style to political behaviors and potential for political fallout. We analyzed press briefings to see how each leader's style tracked to their crisis response.
Crisis leadership is an essential component of a robust governance structure that would enable resilience. This research demonstrated that discourse analysis of political leaders' press-briefings could predict their response patterns and political fallout based on stakeholder perceptions. This finding is important because politicians play a crucial leadership role in disaster response, resource allocation, public perception, and development of resilient systems. federal and state responses exacerbated these divisions. The Trump administration differed with governors over personal protective equipment, reporting, analytics, essential businesses and services, and ultimate re-opening authority. The widening rift begs the question of whether leadership styles and crisis communication encouraged risky behavior that will prolong the pandemic, ultimately compromising U.S. resilience through the politicization of the pandemic response.

Literature review
A considerable amount of literature addresses general leadership. For an exhaustive review, readers can turn to Bass and Stogdill's comprehensive Handbook of Leadership (Bass & Stogdill, 1990) or the Center for Creative Leadership's more contemporary Handbook of Leadership Development (Van Velsor et al., 2010). Political leaders are responsible for protecting their constituents from harm and economic damage (Comfort et al., 2020;Kearns et al., 2019). The phenomenon of political leadership relates to three broad areas of shaping policies, achieving policy goals, and mobilizing support (Torfing et al., 2020. Crisis leadership relates to leadership within the context of a crisis (Boin, 2009;Liu et al., 2020). This review synthesizes the literature on political leadership and crisis leadership.

Political leadership
Four philosophical traditions form the foundation of Western political leadership ideology: Greek and Roman philosophies, Christian theology and philosophy, Social Contract Theory, and Representative Government Theory (Parrish, 2010). These four traditions have commonalities related to the belief that democratic societies must have individuals who rise to political leadership; however, essential differences create an underlying tension that affects current political leadership concepts. The tension relates to concepts of elitism and equality. Political leadership is necessary either because people are not capable of self-rule or necessary because they are capable of self-rule. In the first case, elites must take the reins of political leadership because they have the intelligence, the financial resources, and the character to guide the masses (Keller, 1963). It follows that political leaders must exercise their authority in the interest of what they deem best for the people. In the second case, people cede self-rule to achieve the common good-robust democratic processes are essential to check the excesses and prevent the inevitable abuses of power. The political leader must balance their values and political beliefs with their duty to fulfill the will of the people. Political leaders actions will reflect that, ultimately, people will have an opportunity to reject their sovereignty through the ballot box.
Political leaders have a moral duty to serve their constituents through upholding values, enacting legislation, and ensuring public safety (Couto, 2010;French, 2011;Kearns et al., 2019;Smith et al., 2007). Scholars have advanced different types of political leadership. For example, democratic political leadership honors the pure democratic traditions of collaborative decision-making and policymaking (Couto, 2010;. Vanguard political leadership refers to fringe political groups, usually consisting of intellectuals and activists, that work to overthrow the dominant groups (Marcy, 2020).
Robert Dahl, an influential 20 th -century theorist of democracy, coined the term polyarchy, referring to the democratic governance of large, complex societies (Keohane, 2015). Within a polyarchy, citizens control the political agenda, have recurring opportunities to accept or reject their representatives, require free access to information, and demand political equality. Dahl's early writing depicted political leaders as controlled by the people; however, he evolved to believe that political leaders developed unusual influence compared to other people (Dahl, as cited in Keohane, 2015). An essential distinction of Dahl's concept of exceptionalism was that political leaders' exceptionalism derived from the intense learning they accomplished as they took on increasingly responsible political leadership positions (Keohane, 2015).
In marked contrast, Max Weber argued that an optimal political leader is a virtuous person willing to act corruptly to achieve certain ends and maintain an illusion of control (Weber, as cited in Parrish, 2010). Only such a virtuous person would act with restraint, given the potential to abuse power. Weber observed that successful politicians must build and control political machines capable of inspiring followers and wielding power to its ultimate end of violence.
Joseph Schumpeter identified two primary threats to representative government, incompetence or interests misaligned with the general community welfare. (Schumpeter, as cited in Mackie, 2009). He rejected the idea that a political leader could be elected only to carry out the common good and the will of the people, arguing that there is no such thing as a common good or the people's will since individuals have different needs and desires (Schumpeter, 2008). Therefore, the electorate's primary role must be to elect (or evict) a government, and secondarily that the elected represent its many interests. Schumpeter (2008) described four conditions necessary for the success of the democratic method: (a) competent individuals of "adequate ability and moral character" (p. 290) willing to serve the political machinery, (b) limitations to the real power of government (possibly in the form of checks and balance or electoral limits), (c) a competent bureaucracy with some independence to run the government bureaucracy, and (d) democratic self-control. "Politicians in parliament must resist the temptation to upset or embarrass the government each time they coud do so. No successful policy is possible if they do this" (Schumpeter, 2008, p. 295). Garzia (2011) reviewed the personalization of politics in Western democracies. Politics have moved away from political parties and symbols towards an emphasis on political leaders' personalities. The evolution results from technological innovations in media and changes in political parties' structures due to the declining role of ideology in voter decision-making. Constant attention on political leaders has stripped voters' idealized conceptions in favor of a desire to identify with the leader. To remain relevant, parties cater to voters by emphasizing the candidates' characteristics instead of party principles and ideologies.
From the constellation of general leadership traits (for an exhaustive list, see Bass & Stogdill, 1990), voters judge candidates on a subset of attributes such as competence, leadership, integrity, and empathy (Kinder as cited in Garzia, 2011). Voters consistently rank honesty as the most critical individual trait. Yet, national contests in the U.S., Germany, and Italy demonstrated that candidates deemed dishonest could win national elections (Garzia, 2011).
Decision-making by political leaders is relevant to the current study. Keller and Yang (2008) applied the poliheuristic theory of decision-making to political leaders. Poliheuristic decision-making theories attempt to integrate cognitive and rational approaches. Political leaders exercise two stages of decision-making. In the first noncompensatory stage, leaders apply the cognitive domain by considering their preferred choice's political feasibility as the main factor in decision making. In the second rational stage, they turn to cost-benefit analysis to execute a comparison of options. The research demonstrated that political leaders discount their constituents' perspectives in the face of provocation. This finding suggests that leaders believed that situational variables such as being provoked or involved in a crisis gave them license to follow their instincts rather than the constituents' will. Men were significantly more likely than women to ignore their constituents' preferences.
The preceding may help to explain why scholars have also found that constituents tend to be dissatisfied with their political representation, except in times of collective grief or security, such as after 9/11 (Smith et al., 2007;. The research demonstrated that positive outcomes did not drive constituent satisfaction (Smith et al., 2007). Instead, people were more satisfied with politicians when they deemed that decision-making processes were sufficient, collaborative, and representative of diverse perspectives. Moreover, other research has shown that people appreciate leaders who they believe are sacrificing personally on behalf of the group's needs. People resent leaders who are selfish, arrogant, elitist, or self-aggrandizing. That disdain feeds the personalization of politics discussed above (Garzia, 2011).

Crisis leadership
A crisis is a "breakdown of familiar symbolic frameworks legitimating the pre-existing sociopolitical order" t' Hart as cited in (Boin et al., 2010). In a crisis, leaders must recognize the emerging threat, mitigate the problem, and deal with the consequences (Boin et al., 2010). Leaders must make important decisions that will necessarily affect lives, and they must communicate their decisions to a frightened and vulnerable public (Boin, 2009;Comfort et al., 2020;Körösényi et al., 2016). In a crisis, leaders must cope with an enhanced threat, uncertainty concerning actions that will mitigate the threat, and the urgency to act (Lawton, 2013). A crisis raises the stakes of political leadership because the leader must develop an effective response to the crisis to retain constituents' support (Boin, 2009;Boin et al., 2010;Comfort et al., 2020;Körösényi et al., 2016). Crisis leadership is "the process of leading group members through a sudden and largely unanticipated, intensely negative, and emotionally draining circumstance" DuBrin, as cited in (Liu et al., 2020, p. 128)

Crises frameworks
Much of the literature on crisis leadership frameworks pays particular attention to the organizational perspective (Hermann, 1963;Pearson & Clair, 1998;Quarantelli, 1988). Some researchers developed frameworks concerning the impact of international crises on political leadership (Brecher, 1979;McClelland, 1977). A relatively small body of literature is concerned with crisis leadership from the political leadership perspective, especially from the executive rather than group perspective (Boin et al., 2010;Brändström et al., 2004;Rosenthal et al., 1991).
The following discussion relates to general crisis response frameworks proposed in the literature. The most elementary crisis model, the three-stage model, breaks a crisis into three stages: (a) precrisis, (b) crisis, and (c) post-crisis (Sellnow & Seeger, 2013). In the pre-crisis phase, a rising threat is often minimized or misunderstood. The danger becomes increasingly likely during the incubation or gestation period. The crisis stage begins with a trigger, sometimes sudden and dramatic, that eventually causes people to recognize that a crisis has occurred. People often become emotionally reactive and confused about what to do. Sometimes, the structure necessary to deal with the problem collapses because it is unfit for the task. Self-organizing, emergent forces eventually deal with the situation. The post-crisis phase begins when the crisis abates and a sense of order returns. The post-crisis stage "is also a time of intense investigation and analysis that includes efforts to create plausible explanations of what went wrong; why, how, who is to blame; and what should be done to prevent future crises" (Sellnow & Seeger, 2013, p. 32).
Steven Fink developed a four-stage crisis model during his work leading the crisis team that responded to the accident at the Three Mile Island nuclear plant in Pennsylvania, US (Fink, 1986;Sellnow & Seeger, 2013). Fink (1986) likened a crisis to stages of chronic disease: (a) prodromal, (b) acute, (c) chronic, and (d) resolution. The prodromal phase of a disease refers to when symptoms first occur before the outbreak of disease. During the prodromal stage, if people act quickly, there may be time to stave off the crisis's impending severity. During the acute stage, the problem has fully erupted. Planning or actions taken during the prodromal stage may control critical elements of the crisis such as flow, speed, direction, and duration. Responders use strategic communication to control public actions. The chronic stage represents recovery, analysis, and healing. The chronic stage can take years for people to get over the effects of the crisis. The final stage, recovery, represents a return to normal or, at least, a new normal. Fink's four-stage model has been applied as an analytical tool (Sellnow & Seeger, 2013). Fink's (1986) four-stage cycle was preceded by a decade by Barry Turner's six-stage sequence of failure in foresight (Turner, 1976). The six-stage crisis model included (a) a notionally normal starting point, (b) an incubation period, (c) a precipitating event, (d) onset, (e) rescue and salvage, and (f) full cultural readjustment. Because Turner was a sociologist, his model emphasized cultural dimensions, social processes, and the organization as a complex system. Turner also focused heavily on the need to potentially create or recreate communication channels during a disaster because existing channels are often inadequate or nonexistent. Reynolds and Seeger (2005) described a Crisis and Emergency Risk Communication (CERC) model that Reynolds and her colleagues had developed for the CDC after the anthrax scare and 9/11 terrorist attacks in 2001. The model included a five-stage crisis communication response model: (a) pre-crisis stage consisting of risk messages, warnings, and preparations; (b) initial event, consisting of uncertainty reduction, self-efficacy, and reassurance; (c) maintenance, consisting of ongoing uncertainty reduction, self-efficacy, and reassurance; (d) resolution, consisting of resolution updates, discussion about the cause, and new risks or new understandings about risk; and (e) evaluation, which includes discussions of the adequacy of the response, consensus about lessons learned, and new understandings of risks (Reynolds & Seeger, 2005;Sellnow & Seeger, 2013).

Crisis response
Much of the research into crisis response has focused on what individuals and groups should do during a crisis or have modeled characteristics of crisis stages. Comfort et al. (2020) described four components of an adaptive decision-making process that leaders consider during a crisis: (a) cognition, (b) communication, (c) coordination, and (d) control. Cognition relates to the leader's ability to recognize the emerging risk and act on information to help the community. Communication is a process of creating shared meaning that results in the effective dissemination of information about the risk and the response. Coordination is the degree of interdependence and alignment between relevant actors. Control relates to the effective response to the threat while maintaining societal systems (e.g., social and economic activities). Boin (2009) described five executive tasks for political leaders during a crisis that track with the adaptive decision-making process suggested by Comfort et al. (2020): (1) Preparing in the face of indifference.
(2) Making sense of emerging and evolving threats.
Preparation and sensemaking occur along with cognition. Managing response networks relates to coordination. Offering credible answers is part of communication. Learning under pressure can be seen as control because maintaining societal systems during a crisis would require learning.
Some scholars have observed that most governmental crisis response features centralized decision-making in three areas: (a) power is concentrated in a few executives, (b) decisional power becomes concentrated in governmental agencies, and (c) there is a desire for strong leadership and crisis governance ('t Hart et al., 1993). However, 't Hart et al. (1993) argued that crisis conditions such as high threat, time exigency, and uncertainty made bureaucratic treatment of a crisis infeasible.
Scholars also refer to structuralist or constructivist responses to exogenous crises such as hurricanes, wildfire, flooding, and terrorism (Körösényi et al., 2016). The structuralist approach is reactive because crisis brings about a change in the existing structure, necessitating a leader's reaction. Generally, leaders react by attempting to reduce uncertainty through the use of conventional crisis management tools. The political leaders view the crisis as an anomaly rather than a systemic problem. The anemic reaction to Hurricane Katrina from multiple government levels exemplifies the reactive nature of structuralist political leadership (Boin, 2009).
Constructivists suggest that leaders leverage the crisis by constructing a narrative to shape the desired response. In these cases, the political leaders see the crisis as a system failure. Thus, the crisis becomes an opportunity to bring about systemic change. President Bush's reaction to the terrorist attacks of 9/11 in the U.S. is an example of constructivist political leadership. He used the crisis to create the Department of Homeland Security, enact the Patriot Act, start the Global War on Terror, and invade Iraq and Afghanistan (Kapucu, 2009;Körösényi et al., 2016).
Political leaders sometimes invent crises from endogenous circumstances to fit a narrative of their choosing (Körösényi et al., 2016;Schnurr et al., 2015). Their motives in using these so-called voluntaristic approaches are to bring about transformational or revolutionary change. These political leaders intend to "question conventions and conventional authorities … to change the balance of power, undermine their rivals' structural position, rebuild institutions or exchange them for new ones" (Körösényi et al., 2016, p. 99). President Trump's assertion that the 2020 U.S. Presidential Election was stolen is representative of the voluntaristic political leadership approach (Hart, 2020;Jacobson, 2020). Trump started a false narrative that the presidential election would only be fair if he won before the first votes were cast (Jacobson, 2020). He continued to perpetuate the story before, during, and after the election. The continuous drip of false and misleading information galvanized tens of millions of people to believe and support the false claims because those supporters believed that the institutions he attacked were corrupt. The deception was successful, but Trump's strategy to overturn the election failed to bring about the revolutionary change he sought. Hermann (1979) presented a comprehensive list of stress indicators in policymakers during foreign policy crises. Not all policymakers experienced stress during a crisis, and some stresses extended beyond the executive to other individuals, groups, and organizations. Indicators of a stress reaction included verbal and nonverbal indicators (e.g., flustered or rapid speech, eyeblinks, irritability, vigilance, changes in voice quality). General coping behaviors included contending (fight), avoidance (flight), and inaction.

Transboundary crises
According to Boin (2009), in a transboundary crisis, "the functioning of multiple, life-sustaining systems, functions, or infrastructures is acutely threatened, and the causes of failure or courses of redress remain unclear" (p. 368). Transboundary crises may cross geopolitical lines in addition to functional and temporal boundaries. Thus, it is difficult to discern why, where, or when a transboundary crisis started and who should be responsible for ending the crisis. Therefore, transboundary crises present a challenge for political leaders because they might not know the what issues to address, how to address them, or whether they should be primarily responsible. Furthermore, there might not be systems in place to respond to or learn from a transboundary crisis, there will likely have been no training or communication networks created to address the specific challenges, leaders may have no understanding of the complexity of the crisis, and leaders may not have the capacity to organize an effective response to cover all of the affected boundaries. Finally, political leaders must maintain transparency in providing accurate and accessible informaition to generate public and political support for their actions.
The world is becoming more susceptible to transboundary crises due to globalization, technological advances, modernization, climate change, terrorism, and economic inequality (Boin, 2009;Hannah et al., 2009;Lawton, 2013). Hannah et al. (2009) advanced a leadership framework for extreme contexts, recognizing that there might be no greater need for leadership than in extreme contexts. Hannah et al. (2009) differentiate crisis as a more general term than extreme on four accounts. First, the threat in an extreme event or context must reach the level of intolerable magnitude, which by its nature might bring about different leader and follower responses. For example, in an extreme context, followers might be willing to risk their lives in a way not thinkable during a crisis. Second, definitions of crisis include urgency of action, whereas extreme situations may be ongoing (e.g., wartime operations). Third, crises may be lower probability occurrences than extreme events. Consider the example of paramedic first responders who deal daily with extreme events. Lastly, extreme events may not be ambiguous and uncertain, as in the case of crises.
Nevertheless, some crises, such as pandemics and natural disasters, do reach the level of extreme events and extreme contexts. Hannah et al. (2009) The proposed framework suggests that leaders need to be aware of defining factors such as location in time, the magnitude and probability of consequences, physical or psycho-social proximity, and the form of the threat. According to the framework, the factors are attenuated by psychological, social, and organizational resources and attenuated by time and complexity. These considerations appear relevant to the political leadership of transboundary crises.
We now turn to the theoretical framework for the current study.

Theoretical framework
Destabilizing events increasingly bring about intense politicization of the events and the responses (Boin et al., 2010). Leaders become concerned about whom stakeholders will blame and hold accountable for the events. Furthermore, elected leaders might assess how their decisions and actions will affect their electability. The closer to an election, the more likely political calculus will influence the leaders' decision-making. Boin et al. (2010) argued that politicians worry about the political fallout for crisis outcomes during a crisis.
Whether elected leaders survive politically may depend on (a) the extent of blame they receive from the press, politicians, and formal investigations; and (b) how well they manage fallout both during the crisis and while responding to the scrutiny. Therefore, it becomes of primary importance to understand how leaders will act in these circumstances. As previously discussed, political leaders have a primary responsibility to serve their constituents and keep them from harm. If political leaders elevate their political fortunes over their constituents' safety, they fail their primary democratic leadership responsibility.  Table 1).
The Boin et al. (2010) framework has two axes, (a) the need for control and personal involvement on the horizontal axis and (b) leader sensitivity to context on the vertical axis. Political leadership styles predict the leader's vulnerability to political fallout and how constituents will judge their competence (Boin et al., 2010). The combination of political and crisis leadership in a single framework is unique within the literature. Furthermore, the framework provides a lens to examine a political leader's crisis behavior. Thus, the framework facilitates key elements of our research agenda, as explicated in the conceptual framework section.

Need for control
When leaders have a high need for power, they are likely to be hands-on, attempt to exert more control over a situation, and be more invested in the outcome. During a crisis, high control leaders demand to be at the center of decisions and processes. By contrast, low control leaders are handsoff, preferring to delegate all but the most critical decisions to trusted subordinates.

Sensitivity to context
Leaders may have a high or low sensitivity to context. Leaders with high sensitivity for context need extensive information before they make decisions. Thus, they are likely to seek input from a diverse set of advisors; they are more likely to understand nuanced policy contexts and perspectives. Leaders with low sensitivity to context do not need much information, may not seed advice, and often are seen as decisive decision-makers. High Seen as more "hands-on" and engaged in handling crisis (little delegation to staff) Seen as more "competent" due to limited visible bureau-political conflict Seen as more "responsive" to the situation due to willingness to consider alternative viewpoints, broad search for feedback, and emphasis on expertise over loyalty in staff Seen as "slow or tentative" in response to situation due to high need and broad search for information when making decisions Mixed vulnerability when crisis has a rapid onset ("fast-burning" or "long-shadow" crises). Less vulnerable due to high personal engagement. More vulnerable due to slow decision process and high need for information Low vulnerability to slow-building crises (like "cathartic" or "slow-burning"), especially if situation complex and characterized by substantial ambiguity Seen as more' detached and uninvolved' in handling crisis (substantial delegation to staff) Seen as less "competent" due to visible bureau-political conflict and slow decision process Seen as more "responsive" to the situation due to willingness to consider alternative viewpoints, broad search for feedback, and emphasis on expertise over loyalty in staff Seen as "slow or tentative" in response to situation due to high need and broad search for information when making decisions High vulnerability to rapid-onset crises ("fast-burning" or "long-shadow" crises) due to limited personal engagement, delegation, and slow decision process resulting from high information needs Low vulnerability to slow-building crises (like "cathartic" or "slow-burning"), especially if situation complex and characterized by substantial ambiguity Low Seen as more "hands-on" and engaged in handling crisis (moderate delegation to staff) Seen as moderately "competent" due to rapid, decisive decision style, but visible bureau-political conflict Seen as more "unresponsive" to the situation due to unwillingness to consider alternative viewpoints, limited search for feedback, and emphasis on loyalty over expertise in staff Seen as "more decisive" in response to situation due to low need for information when making decisions Mixed vulnerability when crisis has a rapid onset ("fast-burning" or "long-shadow" crises). Less vulnerable due to high personal engagement. More vulnerable due to general inattentiveness to policy environment increases likelihood they will be caught unawares or unprepared High vulnerability to slow-building crises (like "cathartic" or "slow-burning"), especially if complex and characterized by substantial ambiguity due to insensitivity to context Seen as "detached and uninvolved" in handling crisis (substantial delegation to staff) Seen as "incompetent" due to highly visible bureaucratic infighting and conflict over policy, plus limited personal engagement which slows down decision making. Seen as more "unresponsive" to the situation due to unwillingness to consider alternative viewpoints, limited search for feedback, and emphasis on loyalty over expertise in staff Seen as "generally decisive" in response to situation due to low need for information when making decisions, but limited personal engagement results in "reactive" style High vulnerability to crisis with rapid onset ("fast-burning" or "long-shadow" crises) due to low need for personal engagement. If situation has low ambiguity, may respond well. However, if ambiguous, low sensitivity to context often leads to inappropriate policy responses Highly vulnerability to slowly developing crises (like "cathartic" or 'slow-burning) due to general inattentiveness to policy environment increases likelihood they will be caught unawares or unprepared

Purpose statement
In today's political climate, political leaders must protect their constituents and, arguably, extend their circles of concern to consider global constituents. Since pandemics are global and existential concerns, by definition, politicians must put public welfare above their personal and political fortunes. As of 12 June 2020, the CDC reported 134,572 deaths and 3,236,130 cases of COVID-19 (Centers for Disease Control and Prevention, 2020a). Forecasts for total deaths have continued to increase (Centers for Disease Control and Prevention, 2020b, 2020c) even while the federal government seeks to open schools and businesses.
Political scholars have theorized responses to pandemics that would serve to minimize death and suffering based on ideals of communication transparency, adherence to scientific evidence, and decisive public policy. The question remains of the extent to which U.S. politicians are following a script that will save Americans or if they are attempting to bolster their political opportunities. The purpose of the current research is to identify crisis and pandemic leadership themes from COVID-19 press briefings of U. S. presidential and gubernatorial leaders, based on the following research questions: (1) What themes did President Trump use in his COVID-19 press briefings?
(2) What themes did the sample of Governors use in their COVID-19 press briefings?
(3) Do RQ1 and RQ2 themes map to the theoretical framework?

Selection of subjects
Federal, state, and local agencies typically mount coordinated responses through sharing information, resources, personal protective equipment, and emergency personnel. Ideally, we would have included governors from every state and territory for this research. However, it was necessary to limit the governors to a smaller number for the sake of time and researcher bandwidth. We chose this sample of governors to ensure broad demographic coverage of political party affiliation, gender, and state size. We made this selection at the start of this research in March of 2020. The selection created some methodological challenges due to the wide disparity in the numbers of press briefings. At the start of COVID-19, President Trump and Governor Cuomo held nearly daily press briefings. Governors Ivey and Ducey have held only a few briefings. The remaining governors have held multiple briefings per week. Table 2 lists the subjects for the current research, along with the political affiliation, gender, population, and numbers of COVID-19 briefing transcripts. We sourced the transcripts for the governors from the transcription website rev.com. We searched the site for each governor's name using a search string (e.g., https://www.rev.com/blog/?s=Newsom). The transcripts for President Trump's briefings come from www. whitehouse.gov at the following link: https://www.whitehouse.gov/search/?s=Coronavirus±task±force ±briefings.

Method
Political scientists generally approach qualitative inquiry using rhetorical, discourse, or narrative analysis (Wesley, 2014). Rhetorical analysis is appropriate for investigating how politicians deliver their messages and the strategies they use to achieve their objectives. Discourse analysis focuses on the ideas behind the message-the range of "values, norms, ideologies, and other contextual factors" (p. 137) transmitted through politicians' language. Finally, narrative analysis investigates "the content, origins, evolution, and impact of the message as a 'story' about political life" (p. 139). The current research was concerned with presidential and gubernatorial leadership in the context of the COVID-19 pandemic. Political discourse analysis was appropriate because we were evaluating crisis leadership through the contextual factors of whether the leader's words and actions were consistent with the expected words and actions based on the framework. Thus, while outcomes are important in terms of preservation of human life and avoidance of suffering, outcomes do not necessarily demonstrate effective crisis leadership.    press briefing transcripts were retrieved from Rev.com and Whitehouse.gov for the research subjects covering 27 February 2020, to 14 May 2020. This process resulted in 239 raw transcripts for the eight subjects. The transcripts were downloaded using an NVivo feature that captures a webpage in a PDF format, then imported and automatically coded to the case for each subject, as appropriate. The cases were cleaned by deselecting text that did not belong to the subject (i.e., was spoken by someone other than the subject or was identifying information about the event). At the end of this step, each of the eight cases contained coding only for that individual.

Research protocol
NVivo's automatic coding identified themes, sentiment, and word frequency for each case, individually, and for all cases, collectively. Theme identification and sentiment analysis were accomplished at the paragraph level. The word frequency query included stem words (e.g., talk, talking, talked) with at least four characters.

Addressing researcher bias
To address the potential for bias in this study, we incorporated manual and automatic coding to ascertain themes. We attempted to leave personal political bias out of this analysis by manually coding items against a theoretical framework grounded in the crisis communication literature. Concerning bias in the automatic coding, we assume that NVivo programmers did not introduce bias into the software program through their coding practices that will affect the research findings. In addition, the main author sought feedback and reviewed his findings with several peers. First with the second author, who is a trained qualitative researcher specializing in the research methodology of phenomenology. The second author reviewed the findings utilizing the bias reduction technique of epoche. Schmitt (1968) described reducing bias through epoche as the state of setting aside or bracketing out "prejudgments, biases, and preconceived ideas" (p. 59). The first author also employed the use of an editor who reviewed the findings of the study for personal or political bias as well as blatant coding errors.

Conceptual framework
The conceptual framework for the current study is as follows: (1) By definition, a pandemic represents an existential crisis for humans. COVID-19 may have come from the transfer of the coronavirus from a bat to a human in Wuhan, China.
(2) Effective mitigation of the global effects of pandemics rests on the crisis leadership and communication used by political leaders to calm the public, define the threats, explain the response, and keep the public updated.
(3) U.S. political leadership is sufficiently fragmented that public health norms have become politicized. The U.S. election cycle further exacerbates issues. The politicization of the pandemic response endangers the public and has cost lives.
(4) Effective leadership is essential; thus, it is crucial to examine how our political leaders are leading the pandemic response.
(5) Political scholars typically use rhetorical, discourse, or narrative analysis (6) Because COVID-19 has occurred during the U.S. presidential election cycle, contextual factors appear to play a large part in how some politicians have shaped their response to the pandemic (e.g., masks are mandatory, masks are not mandatory). Thus, political discourse analysis is appropriate to investigate the phenomenon of presidential and gubernatorial leadership during the COVID-19 global pandemic.
(7) Political Discourse Analysis was used to analyze 239 transcripts from the press briefings of President Trump and seven governors to determine the extent to which the research subjects used effective crisis leadership and communication.
Each leader's style informs their response to COVID-19. Leadership style causes political behavior informed by the dynamics of style and context (Boin et al., 2010;Cairns, 2017), which informs public perception. These items interact to create a real and perceived political vulnerability that influences governance choices and policies (Boin et al., 2010). All of these dynamics interact to affect the resilience of the overall system-the environmental, human, and national security; the style, behavior, and public perception of the leader; and the political fallout and policies that result (Stein, 2014). Figure 1 depicts how the conceptual framework relates concepts of leadership, governance, and resilience.

Auto-coded themes
NVivo automatically identified 43 common themes across all subjects; the top ten are listed in Table 3. In aggregate, NVivo codes these themes in 14,376 locations within the 239 transcripts. No theme occurred in every document.
Every node also contains subnodes, which provide additional context to the node. For example, the node business contained 162 subnodes, with the most prominent being airline business, business community, business owners, essential business, restaurant business, and small business owners. Automatic subnode categorization does not take subtleties into account. Several business subnodes are related to the same or similar concept of business owner, for example, (e.g., business owner, small business owner, business people, small business men, small business folks). Also,

Leadership
Style a  under business subcodes, several industries are coded (e.g., airline, restaurant, cruise line). Table 4 shows all subnodes with at least 50 references. Table 5 shows the number of automatically coded references for each node, by case. The cases are in the same order as the previous Tables 5 Figure 2-9 show the nodes and references for each subject. These nodes are independent of the nodes generated for subjects collectively.

Manually coded themes
Four transcripts were selected for each subject as follows: (a) the first two chronologically, (b) mid-April, and (c) the last before May 15. President Trump stopped attending coronavirus press briefings on 27 April 2020. Governors Ducey and Ivey had one and three transcripts, respectively.
We read each transcript twice to develop context and understanding of the interactions. We then assigned words, phrases, sentences, or passages to the predetermined nodes from the theoretical framework, starting with the earliest and working through the latest. Table 6 shows the number of manual themes coded to the theoretical framework for each subject.  brief or incomplete information as having a low need for information (n = 37). Governor Newsom,on 19 March 2020, demonstrated a high need for information: Newsom What we're seeing not only on the ground, but we're seeing through the data, is this spread continues at a pace that we had anticipated in a number of our models. Let me be precise. It's a number that we have been using for the purposes of guiding our resource delivery and guiding our decision-making. (Newsom, 2020a) President Trump's answer to a reporter's question on 29 February 2020, illustrated a low need for information. The president provides an equivocal answer prompting an intervention from Vice President Pence, who directed the president's attention to an expert who could answer the question: As you know, the process of approving medicines is complex and time-consuming. What powers Trump I think this process will go very quickly.

Reporter
Reporter What powers will you use to speed it up?
Trump I think Democrat, Republican, liberal, conservative-this process is going to go very quickly. Pence The head of the FDA is on your right. Trump And, you know, that was-excuse me? Pence The head of the FDA is here, if you want to call on him. Trump Yeah. We can do that. You want to do that? (Trump, 2020b) 5.2.1.2. Need for control of crisis response. A high need for control (n = 56) or a low need for control (n = 30) of crisis response was coded based on how much power or command the leader appeared to demonstrate. Most of the subjects demonstrate both aspects at different times. Here, Governor Cuomo demonstrates a high need for control of the pandemic response: Cuomo There will be no more gatherings of 50 plus people, so if you were hoping to plan a graduation party, you can't do it in the state of New York … .Gyms are closed, effective 8:00 PM tonight. I know that's a specific hardship for the people in this room because I can all see your masterful shape, buff even. There are other ways to exercise. Theaters closing at 8:00 tonight until further notice. (Cuomo, 2020) The following passage from Governor DeSantis illustrates a low need for control of the pandemic response: Reporter Why has the number of deaths been removed from the dashboard? I mean, it seems like the dashboard is becoming less transparent. Can you address that? DeSantis I did not know that that was the case. I'll ask them. I mean, I told them to put all the data out. (DeSantis, 2020)

Personal involvement in decision-making.
We coded high involvement in decision-making (n = 60) or low involvement in decision-making (n = 6) based on how assertive the leader appeared to be in making decisions or demonstrating direct or indirect involvement in a decision. Governor Ivey demonstrated high involvement in decision-making in the following passage: Ivey On March 13th, I originally issued guidelines, strongly urged recommendations, you might say that if we all just did our part, then hopefully that would have been enough. At that time, I spoke about my reluctance to issue a stay-at-home order because I've always known that if the government kills a business, Washington can't print enough money to bring it back to life. (Ivey, 2020c) The following passage illustrates President Trump deferring to Attorney General Barr: Trump Well, you'd have to ask Attorney General Barr, but I think he wants to see-like everybody, he wants to see people get back and wants to see people get back to work. He doesn't want people to be held up when there's no reason for doing it. (Trump, 2020b) 5.2.2. Political behavior 5.2.2.1. Amount of delegation. High delegation (n = 29) or low delegation (n = 0) was coded based on whether leaders appeared to share responsibility for critical functions with advisors or whether they seemed to take on those functions themselves. We did not find any examples of politicians who demonstrated low delegation. Governor Ivey demonstrated high delegation in the following passage: Ivey I intend to take all of these suggestions and recommendations, and ask six members of my coronavirus task force to help my administration begin to vet all of these good ideas. These six individuals will serve on my executive committee, and they will work to put together a thoughtful, well-planned timeline for us to open up the economy. (Ivey, 2020b)

Emphasis on loyalty or expertise.
Passages where the leader appeared to highlight the importance of falling in line were coded as emphasizes loyalty (n = 71) or referencing expert opinion as emphasizes expertise (n = 9) in making decisions. The following passage from Governor Ducey demonstrates an emphasis on expert opinion: Ducey What I mentioned here today only touches on the surface is Dr. Anthony Fauci, with the NIH, alluded to over the weekend. The social distancing measures we have implemented are making an impact. We need everyone to keep it up, and I want to thank everyone for their cooperation and responsible behavior. Next, I want to hand it over to Dr. Cara Christ for a public health update, doctor. (Ducey, 2020) The passages that emphasized loyalty were all attributed to Governor DeSantis or President Trump. For example, here, the president emphasizes the loyalty of his supporters when armed individuals protesting the mask mandate took over the Michigan Statehouse: Reporter But would you urge those protesters to listen to local authorities? Trump I think they're listening. I think they listen to me. They seem to be protesters that like me and respect this opinion. And my opinion is the same as just about all of the governors. (Trump, 2020a) 5.2.2.3. Appointment of political friends or experts. In some cases, leaders rely on political appointments to run top jobs, codes as appoints friends (n = 4), while in others, leaders called on top experts to help solve issues, coded as appoints experts (n = 8). There were eight examples of expert appointments and four examples of friend appointments. The first passage demonstrates Governor Ivey's appointment of an infectious disease expert to her response team: Ivey Dr. Harris is a widely respected specialist in infectious diseases, and goodness knows his experience and his expertise are needed now more than ever. (Ivey, 2020a) The following example demonstrates the appointment of friends to high places: Trump Mike is really good at it. They're going to work together. They're going to work very closely together. And they're both in the administration. I see them all the time, so it really works. This isn't a czar. This isn't going out and getting somebody that's never been in the administration. (Trump, 2020b)

Conflict or alignment between advisors.
Passages were coded as demonstrating high conflict between advisors (n = 8) demonstrating low conflict between advisors (n = 2). In the first passage, Governor Ducey takes the stage from his expert so that he can refocus the conversation, demonstrating low conflict and a desire for his advisor to stay on message: Ducey I think what Dr. Chris is referring to is that the numbers are changing as we have more data and information. We can fill in the blanks. So I know that you all want a prediction from my myself or Dr. Chris. What we're working on every day is to reduce the number of Arizonans that can contract COVID-19. (Ducey, 2020) Quadrant 1: President Trump demonstrated high conflict with advisors in seven passages. For example, here he seems to side with some governors over Dr. Fauci, a foremost expert in epidemiology:

Newsom
Trump Well, that-it doesn't really matter what they say there-and we just left him; we just had a meeting-but-because we're going to have much more than double it very soon. Now, there are big believers in testing, and then there are some governors that don't feel as strongly about it at all. You understand that. They feel much differently about it. (Trump, 2020b) 5.2.2.5. Amount of feedback and environmental scanning. Passages were coded based on evidence that the leader seeks feedback and monitors the environment to help deal with COVID-19 (n = 49) or is self-assured and sticks with a singular plan (n = 26). Here, Governor Ivey demonstrated seeking feedback: Ivey Like any good leader should, I have sought counsel for many avenues and received many, many recommendations. In fact, these are just some of the many reports and suggestions that we received in the last few days from the Small Business Commission to each member of our congressional delegation in Washington. (Ivey, 2020c) In this following example, President Trump sticks to his guns related to getting the economy running at full tilt: Trump Well, that's not going to be normal … .Our normal is if you have 100,000 people in an Alabama football game-or 110,000, to be exact-we want 110,000 people there. We want every seat occupied. Normal is not going to be where you have a game with 50,000 people. (Trump, 2020a) 5.2.2.6. Considering and balancing alternatives. For considering and balancing alternatives, passages were coded at considers alternatives if the leader seemed to be sharing a perspective that came from balancing multiple alternatives (n = 80), or at single direction if the leader appeared to be moving in a resolute direction without considering other points of view (n = 17). We coded the following passage from Governor Newsom at balancing alternatives.
Newsom We always plan. We always anticipate. Let me give you some proof points. Today, we just secured a very large hospital in Northern California, Seton. It's now part of our portfolio, and we're going to populate that hospital as an example to meet the moment. Tomorrow, we'll announce a hospital in Southern California. (Newsom, 2020a) In the following passage, Governor Ducey demonstrates his priority of keeping prisoners behind bars and demonstrates an unwillingness to consider the alternative of releasing any prisoners.
Ducey Not only are we focused on protecting public health, we're continuing to focus on protecting public safety, and we're not going to be releasing any prisoners at this time. (Ducey, 2020) 5.2.2.7. Assigning blame. For assigning blame, passages were coded to accepting responsibility (n = 49) or shifting responsibility (n = 28). Governor Newsom demonstrated accepting responsibility 17 times. For example: Newsom The number one area of consternation that just a year and a half or so ago that I highlighted was some of the borrowing from special funds and some of the deferrals into new fiscal years that were done in the past … .I never imagined to be back a year later under these extraordinary circumstances … .That opportunity affords us about 19% in solutions in this budget. (Newsom, 2020b) President Trump shifted responsibility 20 times in the four interviews that we analyzed: Trump They're having-they're suffering. This country wasn't built on that principle; it was built on an exact opposite principle, actually. And I watched, in one particular state, where they were-they want to get back. They want to get back. They were very strict sanctions that were put on people; that was probably the most strict of all. (Trump, 2020a) 5.2.2.8. Political play. Passages were coded under the theme avoiding politics where subjects appeared to intentionally stay away from making a political statement (n = 57) versus playing politics where subjects did choose to make a political statement (n = 43). Governor Newsom seemed to go to a greater extent than most governors to avoid politics, as seen in the following example: Newsom Again, I just cannot express more appreciation for the collaborative spirit that has defined this moment across agencies, across jurisdictions, across this country, into the White House, the Pentagon, health and human service agencies across this country, sharing best practices across state, and of course, the CDC and their leadership. (Newsom, 2020a) President Trump used a sharp political attack in his remarks to answer a reporter's question regarding Speaker Pelosi.
Trump Yeah, sure. Sure. I think Speaker Pelosi is incompetent. She lost the Congress once. I think she's going to lose it again. She lifted my poll numbers up 10 points. I never thought that I would see that so quickly and so easily. I'm leading everybody. We're doing great. (Trump, 2020b)

Discussion
Based on the automatic and manual coding, each subject appears to fit into the quadrants of the theoretical framework. The findings were inconclusive concerning the amount of delegation. Each of the governors delegated to experts, and President Trump delegated to experts and political appointees. There was not enough information within the transcripts to make a judgment that the amount of delegation was a determinant.  Figure 10 depicts the placement of the research subjects into the quadrants of the theoretical framework. The mapping demonstrates that political discourse analysis provided insight into the Research Questions. These results suggest that Governors Trump, DeWine, and Ducey are vulnerable to political fallout and blame for their handling of COVID-19 because stakeholders might perceive that their policies were ineffective and that they were detached and uninvolved. Governor DeSantis might also be subject to blame unless people view his assertive actions as well-intended, despite policy missteps. Governor Ivey might be vulnerable if people perceive her as detached and uninvolved or slow to act. Governors Newsom, Cuomo, Whitmer may be in a better position to avoid fallout due to their information-seeking hands-on approaches, which many will deem as competent and appropriate to the threat. However, they still might be deemed as slow and indecisive due to their high need for information.

Significance
This research demonstrated that discourse analysis of the press-briefings of political leaders could predict their response patterns and the political fallout based on stakeholder perceptions. This finding is important because politicians play a crucial leadership role in disaster response, resource allocation, public perception, and the development of resilient systems.

Implications
The existential threat of the global pandemic presents an opportunity to consider presidential and gubernatorial leadership from a systemic perspective. Political leadership is critical to creating effective public policy responses that will save lives and use resources effectively (Masys, 2015;Stein, 2014). Leaders at all levels must work together to form a coordinated, effective crisis response (Fineberg, 2014;Gasper, 2014;Reynolds & Quinn, 2008). The current research demonstrates the importance of leadership style on crisis communication and the implications for political fallout. Within the theoretical framework presented here, each leadership style is at risk of some level of fallout. However, low context or low control approaches predict more potential for blame (Boin et al., 2010).
Based on political discourse analysis, we placed the three Democratic governors into Quadrant 1 because they demonstrated high context and high control leadership styles. We placed the other subjects into Quadrants 2, 3, and 4 with respect to their leadership styles. President Trump and the Republican governors each exhibited either low context or low control or both. President Trump's low context, low control style repeatedly placed him at odds with coronavirus taskforce experts in terms of his stance on wearing masks, social distancing, opening schools, and the speed of opening various sectors of the economy. Previous pandemics demonstrated that opening too soon results in higher transmission and more deaths (Fineberg, 2014;Gostin et al., 2016;Hashim et al., 2012;Liang et al., 2018;Thomas & Young, 2011). The implications of opening too fast are more long-term health issues and more deaths. Consequently, if President Trump's leadership style influenced others to follow his lead, despite their instincts or preferred style, then more people are at risk of contracting the virus.

Limitations
The researchers were unable to use measures of intercoder reliability due to time and resource constraints. Manual coding presents the opportunity to apply bias as words carry subjective meaning based on the lens. We attempted to reduce bias through the careful application of the theoretical framework and approaches mentioned in the method section. One researcher accomplished the coding tasks, and another researcher reviewed the coding for content and bias. We also engaged a communications expert to review the draft and final manuscripts for evidence of bias.
Second, due to time constraints, a subset of the transcripts (28 of 239) was coded manually. Coding more transcripts might have resulted in more robust or different findings. It is unclear if President Trump or the selected governors changed their crisis communication as the pandemic proceeded. Extending the timeframe of the study to include additional phases of the pandemic or the runup to the U.S. election might yield different results, and we suggest further study.
Third, we sourced transcripts of the gubernatorial press briefing from a third-party site. Some briefings may have been missing or incomplete, which might have affected the results, especially for Governors Ducey and Ivey. We believe these limitations have not impacted the findings and that the method systematically addressed these concerns by relying on multiple perspectives and research-based frameworks.

Suggestions for future research
The current study looked at President Trump and a small sample of U.S. governors. Extending this research to encompass the U.S. Congress, U.S. state governments, territories, tribal nations, and local politicians would provide more insight into political crisis leadership. There may be different phases of the crisis where different leadership styles are more effective or more appropriate.
It would be informative to study if presidential leadership style and party affiliation moderate the governors' communication styles. For example, Governor DeSantis appeared with President Trump in one press briefing. He may have changed his message to stay in line with President Trump's message. It is possible that the Republican governors suppressed information in their press conferences because they did not want to contradict the president's statement. It is also possible that the Democratic governors amplified their context and control styles to separate themselves from the president. President Trump harnessed Twitter as his primary communication channel. In the current research, we reviewed press briefings against the theoretical framework. It would be informative to analyze political discourse delivered through social media (e.g., Twitter, Facebook, Instagram, TikTok). Social media would present a much different perspective of political discourse to different audiences.
Finally, we conducted this research during the first four months of a global pandemic. Extending the time horizon to include the entirety of the pandemic could help refine lessons learned. It will be possible to map some leaders' behaviors to outcomes such as flattening the curve, hospitalization rates, peak contraction rates, and mortality rates.

Conclusion
This research demonstrated that discourse analysis of political leaders' press-briefings could predict their response patterns and how blame might be assigned based on stakeholder perceptions. The researchers analyzed COVID-19 press briefings of President Trump and seven U.S. governors using a theoretical framework based on the work of Boin et al. (2010). The framework is intended to demonstrate how leadership style affects crisis behavior and communication, thus predicting how blame might fall either during low moments of the crisis or once the crisis is over and the public begins to evaluate the responses.
Based on a political discourse analysis of 28 press briefings, we placed Governors Cuomo, Newsom, and Whitmer into Quadrant 1 (high context, high control) because they demonstrated high context and high control leadership styles. We placed Governor Ivey into Quadrant 2 (high context, low control), Governor DeSantis into Quadrant 3 (low context, high control), and President Trump and Governors DeWine and Ducey into Quadrant 4 (low context, low control) with respect to their leadership styles.