Deny or bolster? A comparative study of crisis communication strategies between Trump and Cuomo in COVID-19

Abstract

This study applied the situational crisis communication theory (SCCT) in political crisis communication amidst the COVID-19 outbreak, a “sticky crisis” that is longitudinal and politicized, thereby involving multiple challenges and complexities. Considering the critical role of Twitter in the information transmissions during the ongoing pandemic, this study considered politicians’ tweets as a proxy to access their crisis communication strategies and conducted a systematic content analysis to critically evaluate COVID-19 crisis communication strategies of two politicians, Trump and Cuomo, according to their perceived day-to-day circumstances during COVID-19. Three strategies categorized by SCCT, deny, diminish, and bolstering, surfaced with significance for both Trump and Cuomo. A new strategy specific to the political context, cohesion, was also identified. In addition, significant differentiation was observed in the strategic narratives between Trump and Cuomo, which reveals the evolving political dynamics in disease representation and crisis messaging. For example, Trump emphasized social exclusion and accusations of Democrats whilst Cuomo stressed care for vulnerable and minority groups and compassion delivery. Moreover, deny strategy, especially accusing other races, significantly boosted audience engagement for Trump. The results are discussed in relation to the idiosyncrasy of the complex COVID-19 pandemic and crisis communication in the political realm. Our findings demonstrate practical implications including online crisis messaging recommendations that foster public trust during politicized and polarized health emergencies and cultivate grounds for information exchange beyond partisan barriers.

1. Introduction

On March 16, 2020, then-President Trump first replaced COVID-19 with ‘Chinese Virus’ in his tweet, “The United States will be powerfully supporting those industries like Airlines and others that are particularly affected by the Chinese Virus. We will be stronger than ever before! (The Hill, 2021)” This racialized terminology that he has used frequently since and alternated it with variants such as kung flu (The Hill, 2021), ignited nationwide discussion and concern. Incidents of anti-Asian violence spiked, along with hashtags and negative tweets from Twitter users displaying hatred toward Asians (Hswen et al., 2021, The Hill, 2021). On the other hand, this attempt of redirecting blame has encountered backlash; public criticism against Trump for his racialization and politicization of COVID-19 has spiraled (USA Today, 2021b).

Meanwhile, whilst being severely attacked by his misconduct, then-governor Cuomo gained media attention with his emphasis on scientific expertise, detailed facts, and instructive messaging via Twitter (Forester & McKibbon, 2020). While anecdotal evidence from the U.S. press highlights a substantial difference in the crisis communication strategies implemented by Trump and Cuomo, few studies evaluated their strategic communications empirically. This study attempts to provide a qualitative and quantitative examination of how Trump and Cuomo honed crisis communication strategies and nursed audience opinions on the COVID-19 crisis using Twitter.

This study is significant on multiple theoretical and practical levels. First, this research fulfills the need to examine crisis communication strategies that occur in Twitter, an online participatory system, from an engagement perspective (DiStaso, Vafeiadis, & Amaral, 2015). Social media have played an ever-increasing role in information circulation and accessibility during global health crises (Guidry, Jin, Orr, Messner, & Meganck, 2017). Social media such as Twitter collapses multi-layers of social contexts into one and proffers a comprehensive, instant, and engaging informational channel (Eriksson and Olsson, 2016, Freberg, 2012, Guidry et al., 2017, Lin et al., 2016, Jiang et al., 2016, Xu and Wu, 2015, Xu, 2020), especially for threatening situations (DiStaso et al., 2015). During the ongoing pandemic, more than half of U.S. news readers accessed COVID-19-related news via this format (Pew, 2021a), leading crisis managers, including politicians, to increasingly view these platforms as an essential vehicle for direct, interactive dialog in strategic crisis communications (Guidry et al., 2017). In this regard, we accessed Trump and Cuomo’s crisis communication strategies and the audience metrics (favorite and retweet number) in response to those messages on Twitter.

Second, the present research makes an offering to the understanding of political crisis communication. Theoretically, there are two broad crisis communication areas: corporate and political crisis communication (Coombs, 2011). While the two fields share similarities, there exist radical differences in the conceptualization and application of crisis communication strategies (Coombs, 2011). By critically assessing and discussing the degree to which the crisis response strategies implemented by Trump and Cuomo coincided or differed, prompted by their perceived crisis responsibility and political motivations, we implicated possible avenues to investigate where the unique aspects of political crisis communication reside. The present analysis is an urgent necessity considering COVID-19 has been increasingly politicized (Hart, Chinn, & Soroka, 2020) and linked to politicians’ political status and policy agenda.

Third, this study applied situational crisis communication theory, also known as SCCT, in the context of COVID-19, a “sticky crisis” in which evolving socio-political factors work in tandem due to the longitudinal and omnipresent nature of the harms, to influence the disease contagion and eventually extend the core crisis into recurring paracrises (Coombs, Holladay, & White, 2020). SCCT contends that to resolve the reputational threat, crisis managers should tailor their strategies to the level of the crisis responsibility perceived by the relevant public(s) and account for past crises history. Further, although SCCT has been adopted in the context of COVID-19 to examine Norwegian (Christensen and Lægreid, 2020) and Chinese (Li, Chandra, & Fan, 2021) governments’ messaging on social media, scant attention has been dedicated to investigating U.S. politicians’ crisis response strategies during the pandemic, reinforcing the novelty and importance of this study. From this perspective, we contribute to the SCCT literature by (1) exploring it in a single yet lengthy crisis with multifaceted challenges, (2) extending it to the digital information ecosystem, and (3) applying it to U.S. political crisis communication.

Practically, this study yields implications for crisis managers in online messaging that helps develop public trust and societal solidarity during politically polarized and longitudinal health emergencies. Different modes of public leadership in crisis communication facilitate different patterns of public reaction and social collaboration (Forester & McKibbon, 2020). As opinion leaders, politicians’ online crisis communication, empowered by their personalized authoritative appeals, carries consequential outcomes on public crisis responses and health behaviors (Guidry et al., 2017). By critically analyzing Trump and Cuomo’s crisis communication strategies, we offer suggestions about harnessing digital technologies for crisis communication that restore political trust as well as encourage compliance with health guidelines against the partisan divide in information dissemination and reception (Guidry et al., 2017).

This study draws from SCCT and relevant research to (1) compare the differences and similarities of Trump and Cuomo’s COVID-19 crisis communication strategies on Twitter, (2) examine the degree to which they conformed to or deviated from the original strategic recommendations of the SCCT, and (3) investigate the relationships between their strategies and the audience engagement metrics (i.e., favorite and retweet number) in response to those messages.

2. Literature review

2.1. Situational crisis communication theory (SCCT)

Crisis communication is an act of gathering, processing, and distributing information in response to crises (Reynolds & Seeger, 2005). Some scholars conceptualized crisis communication as mediated strategic communication and suggested a micro-level analytical approach to capture the complexity of crisis by focusing on four parameters: context, media, genre, and text (Frandsen & Johansen, 2010). In a similar vein, Coombs and Holladay (2001) demonstrated two crisis communication processes: crisis knowledge management and stakeholder reaction management. Crisis managers collect and process information into crisis knowledge and subsequently shape public reactions to the crisis and the organization (Coombs & Holladay, 2012). In other words, crisis communicators employ communication strategies to reconstruct public perceptions by mitigating blame attributions and restoring tarnished organizational reputations (Coombs, 2018, Coombs, 2010).

Lee (2005) identified two developmental crisis communication stages. First, researchers focused on discovering response strategies and typologies like an apologia, employed in a particular crisis during stage one (Allen and Caillouet, 1994, Hearit and Roberson, 2020). For the next stage, crisis communication research began to identify specific features of crises that affect individuals’ crisis attribution and eventually trigger appropriate response strategies (Coombs, 2010). The theory that represents the second stage is SCCT. It is rooted in attribution theory that explains people’s tendency to assess and attribute crisis responsibility to situational factors (Coombs, 2018). According to Coombs (2012), crises are primarily perceptual, and thus, their perceived nature determines effective crisis response strategies. In other words, SCCT proposes response strategies to sustain and restore a positive reputation based on feedback from the crisis situation (Butler, 2021). It assumes that when crisis managers can gauge the level of responsibility individuals attribute to the organization, response strategies can operate more effectively (Butler, 2021, Sisco, 2012). Therefore, to grapple with different situational factors of a crisis, including initial crisis responsibility, crisis history, and prior relational reputation, SCCT introduces guidelines for crisis managers who can match proper strategies to different crisis situations (Richards, Wilson, Boyle, & Mower, 2017).

SCCT’s suggested strategies include denial, diminish, rebuild, and bolstering (Coombs, 2018, Coombs, 2007). To elaborate on the strategies, deny indicates an organization’s deleted link to the crisis (Coombs, 2007). The subcategories of the deny strategy are attacking the accuser, denying, and scapegoating (Coombs, 2007). The diminish strategy tries to minimize the perceived severity of the crisis and includes excuse and justification (Coombs, 2007). Rebuild strategies serve to restore an organization’s tarnished image by providing both material and immaterial aid for the victims (Coombs, 2007). Finally, bolstering is about drawing attention to positive attributes of an organization to protect its image, praising stakeholders, promoting better relationships with stakeholders while eliciting sympathy from the audience (Coombs, 2007). These strategies serve to reduce the public’s blame attributions, improve perceptions of the organization, and alleviate negative emotions (Richards et al., 2017).

Another significant factor in SCCT is the level of crisis responsibility, demonstrated through different types of crises (Coombs, 2018, Coombs, 2007). According to Coombs (2007), crisis types provide a frame. Crisis communicators highlight particular cues of the event, such as external forces or actors as the locus of control for the event, thereby shaping how the public attributes crisis responsibility to organizations. It aligns with one of Hallahan’s (1999) seven framing models in public relations (PR), namely the framing of responsibility. Crisis managers frame causes and explanations of a crisis event by considering whom the public holds responsible. For instance, a victim crisis is when the public views an organization as a victim and, therefore, attributes less crisis responsibility to the organization (Coombs, 2018, Coombs, 2007). An accidental crisis is an event in which outcomes are considered unintentional or accidental (Coombs, 2018, Coombs, 2007). Conversely, a preventable crisis, as a consequence of human error, is when the event is intentional, which eventually induces the highest amount of crisis responsibility attribution (Coombs, 2007, 2018).

SCCT holds that the stronger the public attributions are toward the organization, the higher the need for accommodative response strategies by the crisis managers (Coombs, 2018). Overall, SCCT highlights the importance of crisis manager objectives in selecting crisis response strategies that serve to (1) change attributions about the crisis and (2) change public perception about the organization (Coombs, 2010). Furthermore, SCCT functions in wide-ranging contexts, including organizational (Clementson and Beatty, 2021, Sisco et al., 2010), health (Kim and Liu, 2012, Labonte, 2021), and corporate (Barbe and Pennington-Gray, 2018, Cheng, 2020; Ham & Kim, 2017; Butler, 2021; Huang, 2021; Tu, Chang, & Chiao, 2021) crises. Although crisis communication theories work in political realms (Benoit, 2006, Jong, 2017, Liu et al., 2018, Wukich, 2016), quantitative approaches that explore politicians’ use of crisis strategies are minimal.

A large portion of SCCT studies was conducted in experimental and quasi-experimental design, stressing the theory’s predictive nature which requires empirical examinations and validation (Claeys, Cauberghe, & Vyncke, 2010; Coombs, 2018; Ham & Kim, 2019; Mason (2019)). Other studies examined the theory by performing quantitative content analysis on news media (An, Gower, & Cho, 2011), Facebook posts (Formentin, Bortree, & Fraustino, 2017), tweets (Wukich, 2016, Barbe and Pennington-Gray, 2018), organizational documents (Kim & Liu, 2012), and press releases (Sisco et al., 2010).

Departing from these approaches, this study employs qualitative and quantitative content analyses to apply SCCT to a single, yet lengthy, crisis event, COVID-19. COVID-19 has idiosyncratic characteristics that distinguish it from other crises.

First, it is a complex, recurring, and longitudinal crisis, and perceptions of it might change as the pandemic continues. Second, COVID-19 is communicated on social media by both crisis managers and stakeholders. In other words, COVID-19 is a “sticky crisis” involving an interplay of multiple challenges, organizations, and stakeholders that can further extend the scope of the SCCT application (Coombs et al., 2020). According to the Crisis Communication Think Tank, sticky crises are “complex and challenging crisis issues,” as their attributes are more unique and challenging than normal crises for crisis managers to handle (Coombs et al., 2020, p. 35). Types of sticky crises include longitudinal crises, scansis, industry-wide crises, and public health crises (Coombs et al., 2020). COVID-19, being a longitudinal and public health crisis is a sticky crisis, as social distancing and lockdowns for a long period induced additional crisis situations for organizations in the process of grappling with government restrictions and stakeholder communication (Coombs et al., 2020).

In addition, sticky crises are enhanced by contextual factors such as misinformation, social issues, race and gender, and emotions (Coombs et al., 2020). With the advent of the digital age, social and digital media channels have been actively integrated into PR and crisis communication practices which eventually elicited various challenges and additional crises (Frandsen & Johansen, 2017, Wright & Hinson, 2017). Likewise, one of the contextual factors, COVID-19-related mis- and disinformation pervaded social media due to the politicization of the crisis and disrupted public communication efforts during the pandemic, which makes COVID-19 a sticky crisis. In a recent work, Coombs et al. (2020) contended that SCCT can be extended and applied to examining sticky crises which attests to the importance of empirically exploring SCCT in the context of the current study.

Since COVID-19 is a type of sticky crisis entwined with an array of natural, accidental, and human factors (ScienceDaily, 2020, USA Today, 2020b), a wide spectrum of stakeholders attributed blame in unique ways, on which basis Trump and Cuomo adopted their distinct crisis response strategies. Based on SCCT’s theoretical propositions and its linkage to political PR during a sticky crisis, RQ1 and RQ2 are proposed as follows:

RQ1:

What were Trump’s and Cuomo’s COVID-19 crisis response strategies from their tweets?

RQ2:

How did COVID-19 crisis strategies co-occur from Trump’s and Cuomo’s tweets?

2.2. Political crisis communication in the context of COVID-19

As political communication is concerned about the amplification of the distinctive element of power decided through communication and persuasion, it is inextricably connected with political PR (Strömbäck & Kiousis, 2019). Political PR is the process by which a political actor attempts to impact and establish and sustain advantageous relationships and reputations with critical publics and stakeholders to buttress political objectives (Strömbäck & Kiousis, 2019). Considering the shared interest in crisis communication between political communication and PR, Coombs (2019) pointed out the necessity of analyzing political crisis communication in the PR framework in terms of (1) crisis rhetoric roots in politics, (2) framing strategies, (3) the echo chamber effect in information dissemination, (4) use of negative affect, and (5) a demand to address scandals.

Under this framework, in political crisis communication, politicians often encounter greater conflicts and confront more complex stakeholders compared to their corporate equivalents (Coombs, 2019, Strömbäck and Kiousis, 2013). Their careers and policies are especially in a precarious condition when societies undergo major and disruptive public emergencies (Garnett & Kouzmin, 2007), resulting in crises often exploited and utilized as a political tool to win passage of certain policies or elections (Coombs, 2019). Due to these conditions, Coombs (2019) contended that political crisis managers are more likely to contest crisis responsibility attribution, engage in framing battles, exogenize causality, and stoke anxiety relative to the private sector. During crises like COVID-19, politicians attempt to escape blame for crisis attribution, repair images, consolidate political capital and defend policies (Garnett and Kouzmin, 2007, Liu, 2008, Jong, 2017). Therefore, crises should yield framing contests between different political actors on interpreting the events, causes, responsibilities, and government responses, corresponding with their political goals and visions and revealed in their crisis communication content (Garnett & Kouzmin, 2007).

With a focus on the media content, a strand of empirical studies examined how politicians and government agencies responded to crises to enhance their political goals from the PR perspective (Benoit, 2006, Chen et al., 2020, Garnett and Kouzmin, 2007, Jong, 2017, Li et al., 2021, Liu, 2008, Liu et al., 2018;). For example, George W. Bush faced a constant barrage of attacks and criticism amid his 2004 re-election campaign. In this regard, Benoit (2006) applied image repair theory to evaluate Bush’s discourse on Meet the Press television show and identified attempted strategies of denial and defeasibility. Later in 2005, to suppress criticism over the federal failures during Hurricane Katrina, Bush adopted the deflection/blaming approach, transferring responsibility to the inadequate responses from local officials that might have aggravated the crisis (Garnett & Kouzmin, 2007). At the same time, Liu (2008) used image repair discourse strategies to assess Republican presidential candidate Allen George’s crisis communication strategies in response to reputational controversies. George’s media releases used reducing offensiveness, evading responsibility, mortification, and denial strategies. To recap, politicians’ messaging throughout the crises uses strategic discourse to shape the crisis responsibility and public attribution, pointing to the applicability of crisis communication theories to explain the framing contests in political crisis communication.

Referring back to the previous section on SCCT, a few studies have extended SCCT to the domain of political crisis communication on social media. For example, Liu et al. (2018) conducted a semantic network analysis of government tweets dealing with Hurricane Harvey. They observed that multiple strategies emerged under the framework of SCCT, with instructing information strategy predominantly employed before and during the disaster and adjusting information about corrective action and bolstering strategies more utilized in the post-crisis phase, which points to the longitudinal characteristic of strategies in accordance with the crisis evolvement. Similarly, a recent study by Li et al. (2021) drew upon SCCT and conducted a topical modeling analysis of Weibo posts about COVID-19 crisis messaging from the Chinese government. Results displayed that while the Wuhan local government agencies primarily adopted instructing and adjusting information, the central government endorsed using advocacy and bolstering. These findings reflected that different levels of government are engaged in distinct strategic crisis communication patterns to coordinate and control work at all levels of government, demonstrating the situational nature of political agents’ strategies in the power dynamics to fulfill political agendas.

At the onset of the pandemic, Trump and his administration stood on the front lines trying to contain the virus. They were accused of mismanagement, incompetence, and fueling the escalation of the pandemic (AP News, 2020). Trump attempted to deny responsibility by fabricating accusations against other stakeholders, including the Democratic party (USA Today, 2020a), “fake news media” (The Hill, 2020), and China (Reuters, 2020a). Trump also firmly contended the controllability of COVID-19 and likened it to the seasonal flu (Forbes, 2020c), which might ill-inform his Twitter followers on health behaviors. In his policy position on minority issues, he repeatedly linked COVID-19 with the place of origin via labels such as the “Chinese Virus,” “China Plague,” and “kung flu,” stoking xenophobia and imposing prejudices against the Asian American community (CNBC, 2020a, Wall Street Journal, 2020).

Meanwhile, New York city quickly became an early epicenter of the outbreak in spring 2020 (CDC, 2020), which placed enormous responsibility on New York Governor Andrew Cuomo. As New York Governor, Cuomo is a Democratic politician who appeared to employ different discursive approaches. For example, Cuomo tended to build personal links, practice compassion, stress facts, and establish public trust (PR Daily, 2020). Forbes, 2020a, Forbes, 2020b noted that Cuomo was distinguished by his steadiness, empathy, and instrumental delivery of functional communication to soothe the panic emotionally. Furthermore, Cuomo tended to emphasize care for vulnerable and minority groups (Reuters, 2020b). He also denounced Trump’s “Chinese virus” narrative and described Trump’s responses as dividing America and diagnosed federal incompetence (CNBC, 2020b).

Although empirical studies have been sparsely conducted on this topic, two scholarly works confirmed the differences between their crisis communication strategies. Through a qualitative comparison of the parallel briefings by Trump’s White House Task Force and Cuomo, Forester and McKibbon (2020) found that Trump was engaged in dismissing fear, distancing from experts, and presenting a semblance of control. Cuomo, conversely, stressed the collective vulnerability of every New Yorker, concerted public action, facts delivery, and compliance with protection guidelines from the experts. They concluded that their discourses demonstrated differentiated directions of conditioning public sentiment and reactions (Forester & McKibbon, 2020).

Similarly, Watkins and Clevenger (2021) conducted a discourse analysis from the perspective of crisis leadership. They argued that Trump downplayed the severity of COVID-19 and side-stepped responsibility, while Cuomo showed attentiveness and called for unity by drawing on significant themes such as promoting the importance of facts and rationality. Watkins and Clevenger (2021) further concluded that Trump’s approach might make stakeholders view him as inattentive to the crisis and ineffective in his crisis responses. Contrarily, Cuomo was seen as competent and appropriate to manage crises due to his active information-seeking and hands-on policies (Watkins and Clevenger (2021)).

However, these two studies did not use a quantitative content analysis method to systematically evaluate the two opposing strategies, which asserts the importance of this study. Considering the important role of political crisis communication strategies to serve political PR purposes and the differentiation between Trump and Cuomo evidenced by the popular press and scholarly research, we expect significant differences existed in the discursive strategies in communicating COVID-19-related issues between Trump and Cuomo, which leads to RQ3:

RQ3:

What were the differences and similarities between Trump and Cuomo in their COVID-19 crisis communication strategies from their tweets?

2.3. Health and political crisis communication on social media

An increasing number of scholars have advocated examining the increasing utilization of social media in crisis communication and suggested applying the engagement perspective to evaluate online audience responses (DiStaso et al., 2015, Guidry et al., 2017, Jiang et al., 2016, Triantafillidou and Yannas, 2020). Social media surpassed traditional media in meeting affected audiences’ informational and affective demands (Eriksson and Olsson, 2016, Freberg, 2012, Jiang et al., 2016, Xu and Wu, 2015, Xu, 2020). For crisis managers, utilizing social media has become a common method to directly reach the readers in an interactive and dialog manner (Guidry et al., 2017).

In the context of political crisis communication, as social media have eradicated barriers in citizen communication and enabled the gauging of citizen feedback through two-way conversations (Graham, Avery, & Park, 2015), they have also enjoyed increasing adoption among political actors. In a survey of more than 300 U.S. local government officials, Graham et al. (2015) found that social media use was linked with higher levels of local government officials’ perceived controllability over crises as well as the perceived strength in their responses. Twitter is especially regarded as a key battlefield of political PR because it facilitates asymmetrical networks that permit a user’s profile and tweets to be read and shared by anyone (Guidry et al., 2017, Hong, 2013, Lee and Xu, 2018). As an open and networked public sphere (Ausserhofer & Maireder, 2013), Twitter has also been noted to elicit greater message credibility compared to communication via traditional media (Eriksson & Olsson, 2016). Nevertheless, quantitative examinations regarding how effectively government and political actors leverage social media for crisis management have been minimal (Liu et al., 2018), necessitating the research need for a systematic analysis of Trump and Cuomo’s crisis communication strategies and their elicited audience responses.

Social media have facilitated assessment tools and a critical barometer for inquiries on the consequences of crisis communication, such as measures of stakeholder experiences and engagement (Jiang et al., 2016). Common approaches to conceptualizing engagement via social media include how the stakeholders interact with the organizations, provide real-time feedback, share content, and expand the outreach (Jiang et al., 2016). While engagement is a multi-dimensional process and some scholars understood engagement via social media from a cognitive or affective approach, most studies have agreed that engagement implies the active role of stakeholders and hence defined social media engagement in crisis communication based on such behavioral manifestations as retweeting or liking to express support or criticism on specific post pages (Chen et al., 2020, Guidry et al., 2017, Jiang et al., 2016, Lee and Xu, 2018). Therefore, we define Twitter engagement with Trump and Cuomo’s crisis communication strategies as the frequency of retweeting and liking elicited by their COVID-19 related tweets.

Studies investigating how crisis messaging strategies link to the elicited social media engagement from the public are still in their infancy and demonstrated a complicated landscape. Most studies in this realm consider social media engagement as a yardstick to infer the outcome or effectiveness of strategic crisis communications. For example, Guidry et al. (2017) content-analyzed Twitter posts of a prominent international health organization, Doctors without Borders, during Ebola. They found that the posts containing risk perception variables such as danger and the identifiable victim(s) significantly stimulated comment and like frequencies. The researchers conjectured that Twitter posts covering the negative aspects of the disease helped followers acknowledge the full scope of the crisis without disregarding public concern. They concluded that understanding public fear was a crucial and effective method to establish public trust and engage the audience. In the context of COVID-19, Chen et al. (2020) explored the relationship between the Chinese government’s social media content features and citizen engagement, reporting that information regarding the latest crisis and government action updates positively predicted users’ likes, reposts, and comments with the government accounts because this content met the public demand of reducing risks and uncertainties.

Some scholars further examined the implementation of SCCT in social media by reflecting the strategy effectiveness via social media engagement. Through an online simulated hospital health crisis experiment, DiStaso et al. (2015) indicated that information-based messaging generated more information sharing intent on Facebook while sympathy messages lowered post-crisis organization evaluation, which contrasted with Coombs and Holladay's (2008) findings. They explained this result by contemplating how information-based messaging increased higher evaluation of message credibility among the respondents and thus the dissemination intention.

In sum, research in PR and SCCT in the context of social media displayed a mixed nature, which calls for more exploration and development. The evidence further illuminates a significant variation in strategies and elicited responses contingent upon the unique characteristics of each crisis and entangled with factors such as involved organizations, crisis nature, audience constitution, media affordances, and so forth (DiStaso et al., 2015, Chen et al., 2020). For example, in a segmentation model proposed by Chon (2019), political disposition (e.g., ideology and voting history), situational factors (e.g., problem recognition), and government-public relations (e.g., trust) have been found to be critical factors in predicting audience responses. Considering the uniqueness of the current study context which examines a Republican and Democratic politician crisis messaging during an exceptional pandemic on Twitter, we raise RQ4 to explore the communication dynamics of this situational crisis communication:

RQ4:

What were the relationships between strategies and Twitter engagement for Trump and Cuomo on Twitter?

3. Method

3.1. Data collection

We accessed Trump and Cuomo’s tweets from January 24 to July 8, 2020. Trump’s tweets were collected via Trump Tweet Archive (https://www.thetrumparchive.com/), a website that was launched in 2016 that captures and records all available tweets of Trump through Twitter API. Cuomo’s tweets were collected via Export Comments (https://exportcomments.com/), a website to access and record Twitter posts based on the API address. We selected this period because it covered the critical stage of the initial COVID-19 outbreak in the U.S. First detected at Wuhan, China, in early December of 2019, COVID-19 has drawn worldwide attention, with Trump and Cuomo both first tweeting about it in late January 2019. On January 30th, 2020, World Health Organization (WHO) declared the outbreak of COVID-19 as a Public Health Emergency of International Concern (WHO, 2020). Later on March 11th, 2020, WHO assessed COVID-19 as a global pandemic (WHO (2020)). Beginning from March, Trump invoked the Defense Production Act multiple times to ensure critical resource supplies. The U.S. infection also increased drastically during this period, reaching a peak of daily cases/deaths in July (CDC, 2022). These events attested to the emerging and growing severity of the pandemic situation in the U.S. and might have exerted an effect on Trump and Cuomo’s crisis communication strategies as well as public reactions, pointing to the research significance of this period.

First, all tweets for Trump and Cuomo were collected during the outlined time frame. Second, keywords such as “coronavirus”, “virus”, “pandemic”, “crisis”, “COVID 19”, and “COVID-19” were utilized to target COVID-19-related tweets. A manual review process was implemented to ensure the tweets were related to COVID-19. Third, three exclusion criteria were conducted: 1. Unoriginal tweets, 2. Duplicated tweets, and 3. Tweets that did not show any crisis communication strategies. In the end, a total of 140 tweets were collected for Trump and 341 for Cuomo.

3.2. Coding procedure

The study was both deductive and inductive in ways that we adopted and tested SCCT’s four response strategies under the COVID-19 context and that a new significant category emerged from the media texts. The unit of analysis was the individual tweet. To elaborate, we conducted the first round of pilot coding and generated categories that were repeatedly observed from the tweets by employing inductive category formation (Auter, Douai, Makady, & West, 2016) and grounded theory (Glaser, 1965). The new categories were continuously added through the open coding process until all possible strategies were covered. The variables were coded dichotomously as the coding categories were not mutually exclusive. Two researchers pre-coded 10% of the sample size (n = 48), proportionate to Trump and Cuomo’s tweet size via random sampling and calculated the Krippendorff’s alpha for the intercoder reliability. After two rounds of revisions and reconciliations, our Krippendorff’s alpha ranged from.73–1 with an average of.94 (See Table 1). After achieving intercoder reliability, the two coders proceeded to code the tweets independently.

Table 1.

Summary of Crisis Communication Strategy Variables.

Code#	Variable	Variable Definition
DENY1(α = 1)	MEDIA	1 =accusing media organizations of fake news, opposing reportage and stoking panic
DENY2(α = 0.90)	DEMOCRATS/REPUBLICANS	1 =criticizing Democrats in Trump’s case and Republicans in Cuomo’s case
DENY3(α = 0.73)	FEDERAL	1 =criticizing Trump, his administration and the federal government
DENY4(α = 0.88)	EXCLUSION	1 =projecting racial discriminations and xenophobia
DIMINISH(α = 1)	DIMINISH	1 =diminishing the fatalness of the virus and the likelihood of contagion; describing the U.S. situation in a non-severe manner.
BOLSTERING1(α = 0.91)	GOVERNMENT	1 =mentioning Trump/Cuomo’s achievements as leaders; mentioning the achievements of the government in Trump’s case or the local government in Cuomo’s case.
BOLSTER2(α = 1)	ACTIONS	1 =ingratiating the proactive measures in handling the virus
BOLSTER3(α = 0.91)	EMOTIONALSUPPORT	1 =conveying empathy; building intimacy; delivering encouragement, assurance and confidence; using emotional appeal
BOLSTER4(α = 0.90)	TRANSPARENCY	1 =praising and promoting transparency in information
COHESION1(α = 1)	PARTISANSHIP	1 =creating party cooperation to tackle the crisis
COHESION2(α = 1)	INSTITUTIONALCOOPERATION	1 =promoting institutional cooperation, including cooperation with the federal government and different state governments
COHESION3(α = 1)	RALLY-ROUND-THE-FLAG	1 =eliciting patriotism, societal cohesion and loyalty to the country/states to tackle the crisis
COHESION4(α = 1)	INCLUSION	1 =criticizing those who discriminated minorities; promoting integration and cooperation among different ethnic and social groups; and enhancing benefits for vulnerable groups.

We compared the individual categories with the four general strategies from Coombs’ SCCT (2007) and categorized them under the four strategies if applicable. As a result, we came up with 4 deny, 1 diminish, 4 bolstering categories. There is only 1 (sub) category of diminish strategy because diminishing the fatalness of the pandemic was the only inductively observed strategy that conformed to the original definition of diminish strategy (Coombs, 2007). The rebuild strategy is excluded because none of their tweets mentioned apology or compensation, which are its subcategories. Meanwhile, another prominent pattern of new categories emerged, which did not fall under the four strategies according to SCCT. These individual categories were named and grouped under a new general category, cohesion, which indicated an intention to promote cooperation, cohesion, and inclusion among various social members, institutions, and groups.

3.3. Measurement

3.3.1. Manifest content categories

Four categories grasped content embedded within the tweets, including account, date, retweet count, and favorite count. These four categories were documented by data retrieving resources. The retweet and favorite counts for Trump were accessed from https://www.thetrumparchive.com/ which collected the retweet and favorite number of every Trump tweet by the time of collection (July 8, 2020). The retweet and favorite number for Cuomo were accessed from https://exportcomments.com/ which collected the retweet and favorite number of the target url (https://twitter.com/andrewcuomo) by the time of collection (July 8, 2020). Account indicated either Trump or Cuomo. The date ranged from January 24 to July 8, 2020, representing the time frame of interest. Considering the difference in the size of Trump and Cuomo’s followers, there might be an asymmetry in the relationship between their tweets and Twitter metrics. Therefore, we calculated an engagement ratio for both retweet and favorite count, which were the means of interactions per tweet per 1000 followers (Lou et al., 2019, TrackMaven, 2016).

3.3.2. Latent content categories

As indicated in the coding procedure, we identified four categories of crisis communication strategies: deny, diminish, bolstering, and cohesion. Deny comprised four individual categories, covering the attempts of shifting responsibility to other stakeholders such as the media, the opposing party, the administration, and other races. Diminish referred to downplaying the severity of the virus or the pandemic situation. Bolstering consisted of expressing appreciation for government, action praise, emotional support, and information transparency. Lastly, cohesion was constituted by the aspects such as cooperation of the two-party system, social institution cohesion, rally-round-the-flag stimulation, and social inclusion. The individual categories were coded in a dichotomous way. For the summaries of variables, please see Table 2. For a detailed codebook containing instructions and examples, please see Appendix A.

Table 2.

Distribution of Trump and Cuomo’s Tweeting Strategies.

	Trump (N = 140)	Cuomo (N = 343)
Category	Percentage (%)	Percentage (%)
DENY* **	51.4	6.5
DENY1 * **	23.6	1.2
DENY2 * **	20	0.3
DENY3	1.4	5.0
DENY4 * **	31.4	0
DIMINISH* *	20.7	9.7
DIMINISH* *	20.7	9.7
BOLSTERING* **	75	94.7
BOLSTERING1 * **	58.6	39.6
BOLSTERING2 * **	32.9	56.9
BOLSTERING3 * *	35.7	51.6
BOLSTERING4 * **	14.3	56.3
COHESION* *	41.4	54.8
COHESION1	4.3	1.8
COHESION2	7.9	9.7
COHESION3	32.9	30.5
COHESION4 * **	5.7	20.8

Note. * p < .05, * *p < .01, * **p < .001.

4. Results

Two waves of analysis were conducted to structure the data. The first-wave analysis collapsed each subcategory of strategies as nominal variables so that, for example, 1 for deny of each tweet indicated that at least one of the four deny narratives was present. The second wave examined the presence of each individual theme in terms of their proportion to the tweet population.

4.1. Distribution of strategies

For RQ1 which examines the crisis communication strategies used by Trump and Cuomo, Table 2 portrayed the overall distribution of Trump and Cuomo’s crisis communication strategy from January 24 to July 8, 2020. As illustrated, 51.4% of Trump’s tweets involved deny, among which blaming other races and countries (31.4%), attacking fake news media (23.6%), and accusing Democrats (20%) are the most prominent subcategories. Trump also applied the diminish strategy (20.7%) to downplay the severity of the virus and the situation. Bolstering strategy was also frequently utilized by Trump, which surfaced in 75.0% of the tweets. Most of the bolstering attempts engaged in praising his leadership and the administration (58.6%), distributing emotional support (35.7%), and action endorsement (32.9%). Furthermore, Trump resorted to cohesion enhancement, through which he primarily stimulated patriotism and nationalism (32.9%).

In contrast, Cuomo only used the deny strategy in 6.5% of his tweets, which were focused on criticism towards Trump and Trump’s administration (5.0%). Cuomo presented 9.7% of his tweets in employing the diminish strategy. Overall, Cuomo appealed primarily to the bolstering strategy (94.7%), revolving around emphasizing compensation measures (56.9%), providing information transparency (56.3%), offering emotional support (51.6%) as well as emphasizing government achievements (39.6%). More than half of Cuomo’s tweets also concentrated on political and social cohesion, which were aimed at strengthening state/national identification and integrating minority and vulnerable groups.

Furthermore, we conducted cross tabulations to test the interrelationships among the strategies to develop a comprehensive understanding, which is posed in RQ2. We conducted cross tabulations using Fisher’s exact test on two levels: (1) interrelationships among the collapsed strategies, and (2) interrelationships among the individual strategic subcategories. As shown in Table 3, for the collapsed categories, both Trump (χ² [1] = 18.45, p < .001, Phi = −0.36) and Cuomo (χ² [1] = 45.45, p < .001, Phi = −0.37) tended to disassociate the deny strategy from the bolstering strategy. Also, Trump used the bolstering strategy accompanied by the cohesion strategy (χ² [1] =6.63, p < .05, Phi = 0.22) while Cuomo’s tweets showed a negative association between these two strategies (χ² [1] = 6.23, p < .05, Phi = −0.14).

Table 3.

Chi-square Test (Phi) Among Collapsed Categories for Trump (n = 140) and Cuomo (n = 341).

	Trump				Cuomo
Variables	1	2	3	4	4	3	2	1
1. DENY		.04	-0.36 * **	-0.20 *	.00	-0.37 * **	-0.05
2. DIMINISH			.09	.11	-0.12 *	.03
3. BOLSTERING				.22 *	-0.14 *
4. COHESION

Note. * p < .05, * *p < .01, * **p < .001

For individual categories (see Appendix B), Trump and Cuomo interestingly shared some combinations of sub-strategies in their communicative practices with the public. For example, both Trump (χ² [1] = 22.75, p < .001, Phi = 0.40) and Cuomo (χ² [1] = 51.83, p < .001, Phi = 0.39) significantly integrated government praise and action praise. Moreover, they both appealed to the parallels of government praise and information transparency, government praise and institutional cooperation, and emotional support and patriotism.

Nevertheless, the results also surfaced significant differences in their combination of strategies in terms of individual category combinations. For example, Trump significantly allied action praise and information transparency (χ² [1] = 18.78, p < .001, Phi = 0.37). He also synthesized blaming other races and countries with action praise (χ² [1] = 4.62, p < .05, Phi = 0.18). Trump also echoed the narrative diminishing the pandemic circumstances with patriotism appeals (χ² [1] = 5.90, p < .05, Phi = 0.21). In comparison, Cuomo often expressed criticism towards Trump and his administration while calling on institutional collaboration (χ² [1] = 20.31, p < .001, Phi = 0.24). He also significantly integrated the subcategories of government praise and action praise (χ² [1] = 51.83, p < .001, Phi = 0.39), information transparency and downplaying the pandemic conditions (χ² [1] = 5.62, p < .05, Phi = 0.13), partisan cooperation and institutional cooperation (χ² [1] = 11.36, p < .05, Phi = 0.18), and action praise and social inclusion (χ² [1] = 4.20, p < .05, Phi = 0.11).

4.2. Comparing Trump and Cuomo’s strategies

RQ3 delves into the differences and similarities between Trump and Cuomo’s communication strategies when dealing with COVID-19. Cross tabulations were performed with chi-square and Fisher’s exact test was used to assess the significance of statistical differences.

As shown in Table 2, Trump used the deny strategy eight times as that of Cuomo, χ²(1) = 127.69, p < .001. More specifically, Trump used three out of the four deny subcategories, media attack, democrats attack, and exclusion attack, significantly more than Cuomo at p < .001 level. It was worthy of attention that Trump also adopted the diminish strategy significantly more than Cuomo, χ²(1) = 10.77, p < .01. For Cuomo, our analysis showed that he significantly used bolstering to a greater extent than did Trump, χ²(1) = 39.37, p < .001. Furthermore, each individual bolstering strategy was adopted significantly more by Cuomo than by Trump. With respect to the cohesion strategy, Cuomo endorsed this strategy significantly more in his tweets relative to Trump, χ²(1) = 7.14, p < .01. The individual category that distinguished this difference was their utilization of social inclusion, χ² (1) = 16.50, p < .001. In other words, Cuomo put a premium on integrating different racial, economic, and age groups compared with Trump.

4.3. Strategies and engagement ratio

RQ4 attempts to capture the effects of the strategies on engagement ratio. Again, two waves of point-biserial correlations were conducted between overall and individual strategies and engagement ratio for Trump and Cuomo’s tweet population respectively. For the initial analysis, four collapsed categories were submitted to correlational analysis with engagement ratio. To determine the impact of the individual strategies, the second wave of analysis then tested the correlations between individual categories and engagement ratio.

As shown in Table 4, for Trump’s tweets, deny was significantly correlated with retweet ratio (r = 0.17, p < .05), meaning the collapsed deny strategy elicited more retweets among Trump’s audience readers. However, it did not show a significant correlation with favorite ratio (p > .05). Specifically, accusations of other races and countries significantly caused more retweets (r = 0.23, p < .01) as well as favorites (r = 0.18, p < .05). Meanwhile, for Cuomo’s tweets, neither the collapsed deny category nor the individual categories were conducive to more retweets (p > .05).

Table 4.

Correlational Analysis between Strategies, Retweet Ratio and Favorite Ratio for Trump (n = 140) and Cuomo (n = 341).

	Retweet Ratio		Favorite Ratio
	Trump	Cuomo	Trump	Cuomo
Collapsed Category
DENY	.17 *	-0.02	.07	-0.02
DIMINISH	.12	-0.02	.12	-0.04
BOLSTER	-0.20 *	-0.14 * *	-0.11	-0.10
COHESION	-0.08	.00	-0.03	-0.03
Secondary Category
DENY1	.00	-0.05	-0.06	-0.04
DENY2	-0.01	-0.01	-0.09	-0.01
DENY3	-0.10	.01	-0.09	.00
DENY4	.23 * *	–	.18 *	–
DIMINISH	.12	-0.02	.12	-0.04
BOLSTER1	-0.24 * *	-0.13 *	-0.18 *	-0.10
BOLSTER2	-0.13	-0.12 *	-0.13	-0.07
BOLSTER3	.06	.05	.16	.04
BOLSTER4	.00	-0.12 *	-0.03	-0.09
COHESION1	-0.07	.09	-0.06	.06
COHESION2	-0.15	.03	-0.15	.05
COHESION3	.02	.07	.06	.03
COHESION4	.00	.01	.01	-0.01

On the other hand, bolstering significantly reduced the retweet ratio for both Trump (r = −0.20, p < .05) and Cuomo (r = −0.14, p < .01). To detail, according to Trump’s audience, leadership and administration bolstering significantly decreased retweet ratio (r = −0.24, p < .01). This individual frame also significantly decreased the favorite ratio for Trump (r = −0.18, p < .05). According to Cuomo’s audience, retweeting was significantly negatively correlated with government praise (r = −0.13, p < .05), action praise (r = −0.12, p < .05) and information transparency (r = −0.12, p < .05).

5. Discussions

Generally, the strategies identified in our study attest to the uniqueness of politicians’ crisis communication strategies in an unprecedented health emergency. On the one hand, from a communicator-centered perspective, their strategies can be generalized by SCCT to a certain degree. On the other hand, new categories have emerged compared to the traditional research context of crisis communication, and Trump and Cuomo’s COVID-19 crisis communication techniques demonstrated significant differences, digressed from SCCT recommendations.

Politicians and corporate PR practitioners approach crisis communications differently (Coombs, 2019). Politicians restore their tarnished reputation and transform the crisis into a political event for the purpose of maintaining or reinforcing their political standing in the two-party system (Coombs, 2019). It is especially true for Trump, who was occupied with his re-election campaigns while confronted with challenges such as the impeachment, tax return controversy, and conflict with Black Lives Matter protesters. Therefore, politicians’ COVID-19 crisis communication is further complicated by situational conditions where multiple disease, political, and societal realities factor into crisis responses and messaging. Considering our unique research context, we hope to offer in-depth discussion about our findings not only from SCCT but also through additional relevant theoretical lenses.

5.1. Discrepancy with SCCT guidelines and complexities of COVID-19

According to the SCCT, COVID-19 adheres to the definition of victim cluster as a disease outbreak, i.e., a form of natural disaster with the lowest crisis attributions (Coombs, 2007). The crisis response strategy guidelines in SCCT suggest using informing and adjusting information and diminishing to grapple with a victim crisis. Informing refers to relaying information that instructs the public with accurate facts. Adjusting information denotes information on corrective actions for public assurance (Coombs, 2007).

More than 50% of Cuomo’s tweets were appealing transparency by providing unambiguous COVID-19 information and news corresponding to informing and adjusting information. The result is consistent with the SCCT’s recommendation of informing and adjusting information when the organization is perceived as the victim and, therefore, attributed with minimal responsibility (Coombs, 2007). However, our findings demonstrated that Cuomo and Trump use the bolstering strategy most frequently. Diminish was the least frequently observed in Trump’s tweets and second least frequently for Cuomo’s tweets, even though Trump used diminish more than Cuomo.

The specific discrepancy between our findings and the SCCT guidelines may be due to the possibility that Cuomo and Trump do not perceive COVID-19 as a simple natural disaster. Coombs et al. (2020) categorized COVID-19 as a “sticky crisis,” in which external factors consistently pose challenges and are subject to transforming into a paracrisis one after another, expanding from the core crisis. In the case of COVID-19, its longitudinal and politicized nature constituted and enhanced its stickiness, as it affected long-term political agenda and involved contextual factors such as disinformation and xenophobia. Eventually, the complexity of COVID-19, which induced wide-ranging socio-political problems along with the disease, may have led Trump and Cuomo to take unconventional crisis management approaches (Flinders, 2020). The SCCT recommends using bolstering in preventable crises with the highest attribution (Coombs, 2007). Trump and Cuomo may have adopted the bolstering strategy in response to public reaction viewing COVID-19 as a preventable crisis (USA Today, 2021a) and attributing blame to the state- and local-level government for mismanaging the crisis, as the COVID-19 infection and death rates grew exponentially over this period (CDC, 2022). Overall, the study demonstrated that Trump and Cuomo’s crisis communication strategy deviated from the original guidelines proposed by Coombs (2012). Their political motivations and the unique nature of COVID-19 made it difficult to categorize communication strategies under a single, mutually exclusive crisis type.

5.2. Differences in crisis response strategies and party dynamics

Our results demonstrated the differentiated strategies between Trump and Cuomo’s COVID-19 crisis messaging: while Trump used significantly more deny and diminish, such as social exclusion (deny 4) and liberal media bias (deny 1), Cuomo adopted significantly more bolstering and cohesion such as social inclusion (cohesion 4) and compassion delivery (bolster 3). Consultations with the literature that examines U.S. politicians’ individual-level political PR content have implied that this difference may be linked, at least to a certain extent, to politicians’ partisan leadership qualities influenced by the polarized political environment (Buccoliero et al., 2020, Lee and Xu, 2018, Meeks, 2020).

The U.S. political landscape has been witnessed to undergo the process of political polarization (Abramowitz and Webster, 2018, Nithyanand et al., 2017). Competition between the parties evolved (Robinson & Mullinix, 2016) such that Democratic and Republican politicians grew more distant from each other and more ideologically coherent in terms of values (Fiorina and Abrams, 2008, Robinson and Mullinix, 2016). Recent decades have witnessed Democrat and Republican politicians increasingly occupying different or even extreme positions to obtain policy differentiation and issue ownership to maintain and increase voter share (Wagner, 2012). Democrats tend to emphasize social welfare, responsible capitalism, and care for minorities (Shapiro & Mahajan, 1986). Compassion issues are at the core of the Democratic party, and Democrats appear to be better able to manage these issues (Hale & Grabe, 2018). Conversely, the Republican party and politicians appear competent in managing law and order, immigration, and national security. In the digital PR age, scholars have confirmed that politicians’ PR content often reflects their affiliated parties’ owned, divided issues and political values (Lee and Xu, 2018, Meeks, 2020). For example, a content analysis of Trump and Clinton’s tweets in a three-month election campaign phase conducted by Lee and Xu (2018) found that their overall tweets of the two candidates precisely reflected the Republican- and Democratic-preferred issues.

Our study corroborated this partisan distinction as well. For example, policies toward minorities have often been a contested area in which the Democrat party highlights inclusion while the Republican party utilizes exclusion of minority groups (Kuo, Malhotra, and Mo (2017)). In our research context, Trump adopted more denial and diminishment approaches such as stoking xenophobia and sanctioning external stakeholders (deny 4). Considering the upcoming 2020 election, this could be a particularly salient political strategy in his COVID-19 messaging toolbox that helped affirm and reaffirm his stance on Republican-owned issues to his supporters. In comparison, Cuomo was featured by more bolstering and cohesion attempts such as advocating care for vulnerable and marginalized groups and promotion of diversity programs (cohesion 4), a long-standing terrain for the Democratic party (Hale & Grabe, 2018).

5.3. Negativity and twitter engagement

Another prominent finding of this study is that denial and negativity emerged as a prevalent and powerful strategy, especially for Trump. It digressed from the original SCCT recommendations (Coombs, 2007) and the guidance of crisis communication scholars, who emphasized that denial could provoke negative effects such as public mistrust toward institutions (Lee, 2005). While Trump emphasized media attacks (deny 1), Democrat accusations (deny 2), and social exclusion (deny 4) in his denial attempts, Cuomo stressed criticizing Republicans (deny 2) and Trump and his administration (deny 3). Moreover, the strategic and systemic instrumentalization of negativity is indeed a way of eliciting Twitter engagement for Trump, as demonstrated by the positive correlation between the deny frequency and retweet ratio. Notably, social exclusion, namely projecting racism and xenophobia, one of the most controversial, politicized topics for which Trump has been harshly criticized and vilified, significantly boosted the retweet and favorite ratio.

Such results echoed previous political PR and communication research revealing that negativity has become a persuasion tool that draws public attention and audience engagement (Boulianne and Larsson, 2021, Lee and Xu, 2018, Lilleker and Jackson, 2019, Meeks, 2020, Sahly et al., 2019, Sweetser, 2019). For example, controversial tweeting has always been argued to work to Trump’s advantage in drawing public attention (Lee & Xu, 2018). Also, by examining Trump’s tweets and user reactions in the 2016 U.S. presidential election, Lee and Xu (2018) found that Trump’s attack strategy was significantly effective in sparking user engagement such as retweets and favorites on Twitter. Similarly, in the study by Meeks (2020), the tweets of Trump that employed attack and bias frames of the media, a conventional pillar of Republican ideology, prompted more retweets and favourites than other frames.

According to the literature, the consistent relationship between negativity and the increased social media engagement was mostly explained by (1) negative partisanship on the mass level, and (2) negativity bias on the micro level (Meeks, 2020). First, negative partisanship induces negative sharing of attack messaging (Meeks, 2020, Lee and Xu, 2018). Elite polarization parallels the rise of negative partisanship on the mass level (Hetherington, 2001). According to Hetherington (2001), party competition and differentiation have strengthened the two ideological positions for ordinary U.S. citizens so much that the party membership has become a core facet of the U.S. identity. The two parties increasingly view each other as fundamentally different and dislike or even loathe their opponents (Iyengar, Sood, & Lelkes, 2012), captured by the concept of negative partisanship (Meeks, 2020). Party identity has thus become a function of opposing the out-group and subsequently yielded outcomes on the political representation process (Abramowitz and Webster, 2016, Meeks, 2020). Consequently, political leaders tend to adopt attack and negative frames to capitalize on partisan dislike of the opposing party to fortify intra-party association and prompt negative sharing in their partisan bases (Lee and Xu, 2018, Meeks, 2020), which resonates with the crisis rhetoric to define enemies and evade responsibilities in political PR (Coombs, 2019).

Second, the negativity bias elevates viral sharing in the digital space (Hansen et al., 2011, Meeks, 2020). Negativity bias dictates that negative content is likely to have a greater and perhaps more enduring influence on audiences because individuals tend to devote more attentiveness and cognitive resources to processing those information (Soroka, 2012, Soroka and McAdams, 2015). Research has shown that negative content is more contagious (Rozin & Royzman, 2001) and more likely to get virally spread and enthusiastically engaged on Twitter (Lee and Xu, 2018, Meeks, 2020), especially in political conflicts (Lee &Xu, 2018). This negativity bias might be a particularly significant factor in inflaming negative sharing on Twitter considering the rise of negative partisanship (Lee and Xu, 2018, Meeks, 2020) and the heated political tension in the election year (Pew, 2021b).

6. Conclusion

By analyzing Trump and Cuomo’s COVID-19 tweets, we showed how the politicians emphasized various crisis communication strategies, including denying, diminishing, and bolstering under the SCCT categories. Furthermore, a new strategy, cohesion, emerged concerning the specific context of politicians communicating a global epidemic. To tackle the pandemic, both politicians needed to integrate societal and institutional resources and strengthen public solidarity.

6.1. Theoretical implications

This study’s findings theoretically advanced SCCT by applying it in the contexts of (1) the pandemic, (2) the digital media environment, and (3) political communication.

First, since SCCT has not been extensively utilized to investigate health crises (e.g., Kim & Liu, 2012; Li et al., 2021; Liu et al., 2018), the study proffers novel findings on to what extent SCCT applies to an unprecedented pandemic, constituted by many internal and external factors and challenges. It conforms to the contention made by Coombs et al. (2020) on the potential applicability of SCCT in investigating “sticky crises,” which involves external contextual factors: digital environment, misinformation, and paracrises like political, racial, and social issues. Additionally, addressing the longitudinal nature of the crisis, we examined the politicians’ crisis communication strategies on a long-term basis, which previous SCCT studies had not explored sufficiently, and points to the need to study SCCT adoptions of strategies across phases. In summary, the study expanded the research scope of SCCT by applying it to the context of COVID-19, a novel and idiosyncratic crisis type that deviates from SCCT’s original categories.

Second, delving into the tweets as crisis response messages, the study attested to the theoretical potentials of SCCT in understanding crisis communication in the context of social media, in which crisis managers engage in interactive, active, and real-time communication with the public to cope with the crisis via social media (Roshan, Warren, & Carr, 2016). From this perspective, the present research contributes to extant SCCT research by assessing its applicability to the context of social media.

Finally, the application of SCCT to this study has implications for describing and understanding the uniqueness of political PR research. Our study revealed that in the context of messaging a pandemic that knows no bounds, Trump and Cuomo’s strategies digressed from SCCT recommendations and linked to their policy agendas and political purposes, which highlights the idiosyncrasy and complexity of political crisis communication. We further suggest that Trump and Cuomo’s crisis communication efforts and the corresponding induced engagement could be understood through the lenses of partisan dynamics and negative partisanship that work in tandem with a negativity bias. Such political dynamics might be an aspect that makes the COVID-19 crisis messaging of the U.S. government unique and different from other governments. For example, up-to-date crisis facts and government action are rewarded with the highest public attention and engagement for the Chinese government’s social media accounts (Chen et al., 2020), and collaboration messaging turned out to be the key to the success of the Norwegian government in enhancing national coordination. However, in the context of the U.S. government, then-president Trump’s COVID-19 messaging seems to be more effective in attracting attention when exploiting negativity and resorting to attacks. We argue that the analysis of political PR should take into account a variety of contextual factors including political values and goals, power dynamics, audience constitution, media affordances, and so forth. Overall, this study serves as one of the first attempts at an interdisciplinary effort between PR and political communication to make sense of political leaders’ crisis messaging.

6.2. Practical implications

Several practical implications stem from this study. In general, for PR practitioners, our study provided empirical evidence on implementing PR theories and practices in political and health contexts. It is imperative for PR practitioners to consider the effects of political factors when grappling with a crisis that has evolved into a politicized event.

Social media have facilitated flourishing chances for crisis managers to handle the crisis via meaningful, engaging, and interactive dialog with the public (Guidry et al., 2017), which has shifted communication norms, expectations, and protocols for crisis managers as well (Lin et al., 2016). While the recommendations of social media use in crisis management remain unstructured and scattered (Lin et al., 2016), we attempt to derive practical guidance from our findings.

In the public sphere of Twitter, crisis communication messages from authoritative voices from higher government levels, like Trump and Cuomo, have a high impact on audience compliance with safety guidelines (Freberg, 2012, Lin et al., 2016). Considering their potential to influence the public, during the unfolding of the crisis, authoritative communicators should actively utilize the digital affordances of social media. However, at the same time, our findings underscore the importance of considering the complexity of politicians’ social media usages. It is critical for politicians to prioritize crisis communication goals over their political agenda when messaging on social media of which effects are instant and pervasive. We suggest the key emergency and government agencies receive professional training on crisis communication via social media to not only produce consistent messages across administrative levels, policy areas, and parties but also effectively disseminate essential information beyond political barriers.

In all likelihood, political crisis communication messages may particularly be attended and credited by segmented audience clusters that selectively access information for homophilous, ideology-consistent perspectives and hereby produce an echo chamber effect in which in-group framing is resonated, out-group messaging is rejected, and partisan animosity is amplified (Coombs, 2019, Meeks, 2020, Shah et al., 2007, Sweetser, 2019). Hence, balanced and effective uses of social media will facilitate politicians to enhance public trust toward the government and governance legitimacy, eventually achieving the political goals as well as mitigating the public uncertainties. In this regard, politicians are also encouraged to collaborate with social media firms and representatives to build digital strategy, content, and execution (Kreiss and McGregor (2018)) that helps advance an inclusive information dissemination infrastructure beyond political barriers.

Further, when communicating on social media, politicians should adhere to the recommendations of crisis communication and employ more professional strategies and expertise to breed public trust, remedy government reputation, and effectively contain the pandemic. In particular, showing empathy for the victims is a way of providing the public with psychological support to foster trust (Kim & Liu, 2012). Such a role in crisis communication is crucial as victims often expect an organization to devote attention and concern for them (Jong & Dückers, 2019). It is often advised for leaders to deliver social media messages that correspond to the emotional demands of the victims by using rhetorical strategies as a means of complementing heartless, factual messages (Luoma‐aho, Tirkkonen, and Vos (2013)). In the context of political crisis communication in such a health emergency, the delivery of compassion and empathy serves the communication duty of the government to openly address public concerns for the good of the community (Liu & Horsley, 2007) as well as alleviate anger and anxiety to effectively communicate the crisis messages (Coombs, 2007).

It is also recommended that PR practitioners should disseminate timely, factual, transparent, and accurate updates about the crisis (Lin et al., 2016). Especially for political crisis communication, it is more effective when knowledge, generated by experts and professionals, is delivered clearly, timely, and repeatedly by credible political and administrative executives. However, our findings suggest that diminishment has emerged as a critical strategy for Trump such as likening COVID-19 to flu. Such inaccurate statements might be further distorted and misunderstood, contributing to online mis- and disinformation propagation via digital word-of-mouth.

Especially in health emergencies where lockdown and social distancing policies are implemented, people depend on social media as the primary information source, which increases their susceptibility to fake news (Yang & Tian, 2021). Fake news not only misguides individuals to trust unsubstantiated information in making health judgments but also shifts their perceptions and acceptance of true news (Yang & Tian, 2021). For example, in an attempt to diminish the severity of COVID-19, Trump floated a false idea that disinfectant injection helps cure COVID-19, which resulted in the deaths of some of his supporters who believed this misleading statement (Yang & Tian, 2021). Therefore, the delivery of timely, factual, accurate, and transparent information should be assured, which helps the public establish facts, conform to advised health guidelines, reduce uncertainty, and empower self-resilience (Lin et al., 2016, Yang, 2021).

Moreover, in a context of soaring uncertainty and urgency, the success of crisis communication campaigns relies on the voluntary cooperation from each public sector and the entire citizenry (Christensen & Lægreid, 2020). The finding that denial surfaces as a salient tactic for Trump accords with the popular belief that Trump’s COVID-19 messaging led to the threat of racism and social division, which hindered the cooperative intent from certain public sectors (e.g., news media) and groups (e.g., Asian Americans). The detriments of these social dividing messages might be particularly amplified considering our finding that negativity such as social exclusion is more likely to stimulate retweeting, and thus stoke strong emotions among users and be disseminated to a broader spectrum of audiences (Boyd et al., 2010, Lee and Xu, 2018, Meeks, 2020, Tian et al., 2021). Crisis communication is most successful when it is able to unite effective government actions and perceived democratic legitimacy and trust in government (Christensen & Lægreid, 2020). Trust is an intangible asset in government-citizens relations (Chon, 2019), which serves as a crucial component to help people absorb important information and comply with health guidelines and recommended actions (Guidry et al., 2017). Therefore, it is advised that PR practitioners including politicians avoid the usage of exclusive, stereotyped, and prejudiced language in social media campaigning that divides the population, which helps develop democratic legitimacy, cultivate collective solidarity and resistance, and encourage concerted public actions to combat the virus.

6.3. Limitations

This study also has a number of limitations. First, as communicator-oriented research, although we examined the Twitter engagement embedded in the tweets, the results could not infer the valence of audience engagement and cannot distinguish between passive, reactive consumption and proactive, contributory engagement (Men & Tsai, 2013). Also, it is impossible to separate audiences’ reactions to the crisis communication strategies from their responses to COVID-19, their feelings towards the politicians, and their political bent. Future research could conduct content and sentiment analysis of audience responses by adopting computational methods to delve into the valence and substance of audience engagement.

Second, the composition of the audiences in our measured Twitter engagement was unclear. The results should be accepted with caution considering that Trump and Cuomo’s messages might be engaged by specific segments of the users that selectively access their tweets for ideology-aligned information (Coombs, 2019), which is strengthened by Twitter’s recommender algorithms (Sweetser, 2019). Future research might consider incorporating user profiles and network analysis to understand what audience niches are affected by specific types of crisis communication strategies.

Third, although we contemplated the incongruences between our findings and the original SCCT in terms of the idiosyncrasy of COVID-19 and the partisan political system, causality cannot be assumed based on our data. Also, the two politicians’ deviations from the party norms might have induced such discrepancies. It is recommended for future studies to consider crisis managers’ degree of adherence to party norms when examining political crisis communication strategies.

Besides the above-mentioned limitations, our collection of tweets was restricted to a limited time frame, which did not cover the overall and ongoing outbreak. Our analysis also focused on two politicians, Trump and Cuomo, which might not be representative of other politicians. Lastly, the context of COVID-19 introduced idiosyncrasy into our findings. As a result, our findings bear limited generalizability to research in other contexts. However, our study reveals the contextualized aspects of political crisis communication and serves as a stepping stone for making an interdisciplinary effort to compare politicians’ social media crisis messaging.

Acknowledgement

First and foremost, we would like to expressour gratitude to our advisors, Dr. Lars Willnat and Dr. Hua Jiang, forunconditional support and guidance. We also would like to thank the editors and reviewers for their time and consideration.

Appendix A. Codebook

See Appendix section here.

Code#	Variable	Variable Definition	Exemplars
DENY1	MEDIA	1 =accusing media organizations of fake news.*explicitly criticize “fake news”, “fake news media”, “low media trust”, blame the media for “stoking panic” and “opposing reportage”	“The Fake News Media and their partner the Democrat Party is doing everything within its semi-considerable power (it used to be greater!) to inflame the CoronaVirus situation far beyond what the facts would warrant”
DENY2	DEMOCRATS/REPUBLICANS	1 =criticizing Democrats/Republicans*mention the specific names of the other parties or politician representing other parties*call them “do nothing”, blame the other party for speak ill of himself	“Democrats were busy wasting their times”
DENY3	ADMINISTRATION	1 = criticizing Trump, his administration and the federal government	“Congress needs to fund working Americans.They've funded corporations, airlines and banks.They literally put a tax break for millionaires into their COVID legislation.Don’t make the same mistake twice. Fund working Americans.”
DENY4	EXCLUSION	1 =projecting racial discriminations and xenophobia*define the virus as “Chinese virus”, preach closing borders	“The United States will be powerfully supporting those industries like Airlines and others that are particularly affected by the Chinese Virus. We will be stronger than ever before!”
DIMINISH	DIMINISH	1 =diminishing the fatalness of the virus and the likelihood of contagion; describing the U.S. situation in a non-severe manner* “few cases”, “slow spread”, “on top of it”	“So last year 37000 Americans died from the common Flu. It averages between 27000 and 70000 per year. Nothing is shut down life & the economy go on. At this moment there are 546 confirmed cases of CoronaVirus with 22 deaths. Think about that!”
BOLSTERING1	GOVERNMENT	1 =mentioning Trump/Cuomo’s achievements as leaders; mentioning the achievements of the government in Trump’s case or the local government in Cuomo’s case*brag success, talk about the fighting measures he adopted and express resolution using “I”, “my”, “President”, “Trump” and “President Trump” *praise contributions of “administration”, “federal/state government”, “governors”, “congress”, “politicians and officials”, “white house”, “police officers”, “NYS”*CDC not included*subjects of “we” or no subjects included* “my administration” included	“Cryin’ Chuck Schumer is complaining for publicity purposes only that I should be asking for more money than $2.5 Billion to prepare for Coronavirus. If I asked for more he would say it is too much. He didn’t like my early travel closings. I was right. He is incompetent!”
BOLSTER2	ACTIONS	1 =proactive measures in handling the virus*any concrete measures, progresses, and updates such as travel closing and tests* “monitor”, “watch” included*briefings/updates excluded* “need to []” excluded	“Coronavirus: In addition to screening travelers “prior to boarding” from certain designated high risk countries or areas within those countries they will also be screened when they arrive in America”
BOLSTER3	EMOTIONALSUPPORT	1 =conveying empathy; building intimacy; delivering encouragement, assurance and confidence; using emotional appeal* sharing personal stories included	“My daughter called me and said, don't tell me to relax — tell me why I should be relaxed.So I want to make sure I tell the people of New York what I told my daughter.”
BOLSTER4	TRANSPARENCY	1 =promoting transparency in information*share policies/measures, status updates, statistical information*promote the values of facts and information transparency*share resources for information	Update: There are still no confirmed cases of the novel #coronavirus in New York State.11 tests have returned negative and 6 are still pending.While the risk to New Yorkers is still low, we urge everyone to remain vigilant and stay informed. https://t.co/aYbVuToz197
COHESION1	PARTISANSHIP	1 =creating party cooperation to tackle the crisis*praise teamwork and call for coordination between the two parties.*when present with blame, code the one with more word count	“Good teamwork between Republicans & Democrats as the House passes the big CoronaVirus Relief Bill. People really pulled together. Nice to see!”
COHESION2	INSTITUTIONALCOOPERATION	1 =promoting institutional cooperation, including cooperation with the federal government, different state governments.	“After days of advocating the federal gov't to expand #Coronavirus testing capacity, we just received word that Northwell Laboratories has been authorized to test.”
COHESION3	RALLY-ROUND-THE-FLAG	1 = eliciting patriotism, societal cohesion and loyalty to the country/states to tackle the crisis*use proactive expressions of “we will …”, “the United States will”, “we can …”, “let’s”, “America will …”, descriptions of “great nation”, “proud nation”, “everyone”, people”, “a battle doesn’t choose sides”*use expressions presenting the New York State as a cohesive community	* “America will get it done!”“We will be stronger than ever before!”“It’s hypocritical to say NY doesn’t deserve federal funding now because it would be ‘unfair’ to GOP states”
COHESION4	INCLUSION	1 =criticizing those who discriminated minorities, promoting integration and cooperation among different social groups, and enhanced benefits for vulnerable groups.	“NEW: New York is partnering with @NorthwellHealth to open 24 temporary COVID testing sites at churches in predominantly low-income communities & communities of color.”

Appendix B. Correlation Tables

Chi-square Test (Phi) Among Secondary Strategies For Trump (n = 140).

Variables	1	2	3	4	5	6	7	8	9	10	11	12
1. DENY1
2. DENY2	.14
3. DENY3	.08	-0.06
4. DENY4	.13	.09	.18
5. DIMINISH	.09	-0.04	.09	.11
6. BOLSTERING1	-0.15	-0.16	-0.02	.01	.00
7. BOLSTERING2	-0.17	-0.16	.04	.18 *	-0.06	.40 * **
8. BOLSTERING3	-0.27 * **	-0.19 *	-0.09	-0.12	-0.01	.08	.02
9. BOLSTERING4	-0.13	-0.10	.12	.08	-0.06	.18 *	.37 * **	.04
10. COHESION1	.05	-0.02	-0.03	-0.14	-0.11	-0.04	.00	-0.08	.12
11. COHESION2	-0.10	-0.15	-0.04	-0.03	-0.08	.19 *	.14	-0.11	.11	.07
12. COHESION3	-0.14	-0.12	-0.08	-0.05	.21 *	.13	.06	.27 * *	.02	.08	-0.09
13. COHESION4	-0.14	.03	-0.03	-0.10	-0.05	-0.11	.09	.20 *	.16	.25 *	-0.07	.28 * *

Note:* p < .05, * * p < .01,* ** p < .001

Chi-square Test (Phi) Among Secondary Strategies For Cuomo (n = 341).

Variables	1	2	3	4	5	6	7	8	9	10	11	12
1. DENY1
2. DENY2	-0.01
3. DENY3	-0.03	-0.01
4. DENY4	–	–	–
5. DIMINISH	.06	-0.02	-0.08	–
6. BOLSTERING1	-0.03	-0.04	-0.05	–	.06
7. BOLSTERING2	-0.02	-0.06	-0.15 * *	–	-0.10	.39 * **
8. BOLSTERING3	-0.04	-0.06	-0.13 *	–	-0.12 *	-0.10	-0.25 * **
9. BOLSTERING4	.10	-0.06	-0.12 *	–	.13 *	.19 * **	.04	-0.05
10. COHESION1	-0.02	-0.01	.07	–	-0.04	-0.06	-0.06	-0.00	-0.12
11. COHESION2	-0.04	-0.02	.24 * **	–	-0.07	.12 *	.07	-0.06	-0.03	.18 *
12. COHESION3	-0.07	-0.04	.11	–	-0.00	-0.12 *	-0.04	.17 * *	-0.28 * **	.06	.02
13. COHESION4	-0.06	-0.03	-0.12 *	–	-0.14 * *	.18 *	.11 *	.11	-0.03	-0.01	-0.17 * **	-0.17 * *

Note: * p < .05, * * p < .01, * ** p < .001