You’ve got mail: how the Trump administration used legislative communication to frame his last year in office

ABSTRACT

Legislative communication frames how constituents perceive politicians’ successes. However, most research on legislative communication focuses on Congressional or Senatorial email correspondence, without considering the importance of presidential emails or the way politicians frame their failures. Existing work on legislative communication also tends to analyze the documents in isolation, leaving open the opportunity to analyze the networked effect of information flows. To fill this gap, we analyze a year of 1600 Daily content – The Official White House email style newsletter created by the Obama Administration and subsequently adopted after the Trump administration took office. In doing so, we identify the central frames the Trump White House relied on leading up to the 2020 election and the media sources used to legitimize these claims. Drawing on frequency counts, structural topic modeling, and qualitative content analysis, our data reveal the important role electoral communication plays in framing current events and the extent to which email is an essential node in the right-wing media ecosystem.

KEYWORDS:

• right-wing media ecosystem (RWME)
• frame analysis
• misinformation
• disinformation
• legislative communication

Introduction

During his last year in office, Trump faced a series of political events that could have tarnished his support (Galston, 2020). In October 2019, Trump was accused of betraying his oath of office and initiating foreign interference in a U.S. election (Bade et al., 2019). In December 2019, the Judiciary Committee formally approved the articles of impeachment on charges of abuse of power and obstruction of Congress. Shortly thereafter, the COVID-19 pandemic began to spread around the globe with the first cases of the virus confirmed in the United States in late January 2020. During a national health crisis, George Floyd was murdered by police officers, setting off a series of civil rights protests across the United States (Burch et al., 2021). The economic fallout from COVID-19 was equally dire. Unemployment rose higher in the months following COVID than it did during two years of the Great Recession, and upwards of 60% of closed restaurants will never reopen (Kochhar, 2020; Sundaram, 2020).

While news coverage of the impeachment, COVID-19, and the Black Lives Matter protests were plentiful, little work has attempted to understand how the White House tried to frame these same events as they unfolded. Foundational research on legislative communication (e.g., websites and emails) explains the integral role email correspondence plays in influencing voter impressions and electoral outcomes (Cormack, 2017, 2016; Graf, 2015; Grimmer, 2016; Grimmer & Stewart, 2013; Grimmer et al., 2014; Rubin & Lukoianova, 2015). This research has demonstrated that legislators directly communicate with their constituents to try and shape their perception of political events. However, most of these studies focus exclusively on computationally-driven analysis of correspondence from Senatorial and Congressional representatives to identify the key terms and ideas that legislators focus on. This rich work has yet to capture how the president uses the same medium to try and frame current events, does not look closely at the sources elected officials rely on to make their claims, nor address how topics being ignored are analytically valuable.

A growing body of work has also focused on the intersection of misinformation and democracy. What these authors reveal is that misinformation is highly effective because it depends on a network infrastructure to amplify and circulate bogus claims (Benkler et al. 2018; Chen et al., 2021; Freelon & Lokot, 2020; Freelon & Wells, 2020; Nisbet et al., 2021; Ognyanova et al., 2020). One prominent example is the right-wing media ecosystem (RWME) that was discovered by researchers at Harvard’s Berkman Klein Center for Internet and Society (Benkler et al., 2018). Using nearly four million messages and news media in digital and broadcast form, they identified political asymmetry in the United States information ecosystem. Their findings reveal that most media outlets comprise an interconnected web of content that adheres to secular and objective standards of journalistic integrity. The RWME, on the other hand, forms a separate sphere that fuels itself by relying on other sources from within the network that reject standard journalistic ethics, norms, and practices (Benkler et al., 2018). As a function of these self-reinforcing mechanisms, the RWME operates in a digital vacuum of insular, extreme, and partisan coverage, isolating conservative consumers ideologically, socially, and culturally. While misinformation research does an incredible job of mapping the political information landscape, most of the research focuses on social media and does not take into consideration the role legislative communication/email plays in sustaining this alternative influence network.

As a way of bridging the gap between legislative communication studies and misinformation research, we combine frequency counts, structural topic modeling, and qualitative content analysis to better understand how the Trump administration used email to frame his last year in office. Frames are important for analysis because they define both the problem and the solutions in society (Goffman, 1959). Media play an important role in normalizing frames because they highlight certain information as more relevant while representing other aspects of an event as less important (Benford & Snow, 2000; Feagin, 2013; Flores-Yeffal et al., 2019; Rohlinger, 2002; Rohlinger & Quadagno, 2009; Vreese, 2005). Studying media frames is particularly important because they demonstrate persistent patterns of cognition and shed light on the conceptual tools people draw on when processing new experiences (Gitlin, 1980).

By quantitatively and qualitatively analyzing emails the White House sent during Trump’s last year in office, we can comprehensively understand how the administration used the medium to frame Trump’s actions as good for the public and undermined the integrity of the 2020 Presidential election. In doing so, we wish to demonstrate the power email communication has to frame catastrophic events as ‘normal’ and unearth the ways in which politicians act as interlocutors within the RWME, amplifying, legitimizing, and, in some cases, serving as the origin of misinformation to their constituents.

Theoretical background

Analyzing the importance of legislative correspondence

In today's media landscape, people are constantly inundated with news. As a result, there is an increased perception that we are well informed on major social and political information without deliberately seeking information (Gil de Zúñiga et al., 2020). News that ‘finds us’ is also dependent on content heavily curated by politicians, for example email alerts or social media updates. The idea that politicians actively try to shape the public’s perception of them was first developed by Nimmo and Savage (1976) who relied on sociological theory to explain how attributes associated with an event, object or person are socially constructed. Elected officials must devote substantial time and money to craft and maintain their public image (Balutis, 1977; Mayer, 2004). Given the ever-evolving media landscape, politicians, including the president, increasingly rely on direct communication with their constituents to influence voter perceptions (Graf, 2015; Grimmer et al., 2014). Not only are legislative communications (e.g., congressional newsletters, constituent emails, candidate websites, and press releases) central to their representational duties, they also play an important role in their re-election efforts.

A growing body of work in political science examines the important role legislative communication plays in shaping public perceptions of elected officials (Clarke et al., 2020; Cormack, 2013, 2016, 2017; Grimmer, 2010, 2013, 2016; Quinn et al., 2010). Primarily relying on computational methods, these studies reveal that members’ communication styles emphasize certain topics depending on their constituency's perception of those topics (Grimmer, 2013, 2016). For example, a district that veers more conservative might send emails about spending cuts, whereas more progressive constituents might learn about a legislator's efforts to improve public school programs in their emails. Legislators consistently try to influence electoral perceptions by trying to appear more ideologically similar to whichever subset of voters they consider more important to their re-election (Cormack, 2013). By quoting their own sources, choosing what to focus on, and engaging directly with the public, legislative communication subverts the authority of the fourth estate, undermining the press’ ability to shape the discourse around a political event (Gamson & Modigliani, 1989; Graf, 2015).

This researchopens the opportunity to analyze executive-level communication and understand how politicians might use legislative correspondence to frame topics their supporters might ideologically oppose. Except for a research paper presented at the Association for Education in Journalism and Mass Communication (Graf, 2015), no academic researcher has published on executive/presidential email correspondence. Since the most electorally vulnerable are more likely to engage in constituency communication (Peskowitz, 2018), analyzing the email correspondence of the Trump administration in the year leading up to the 2020 Presidential election is particularly important given that Trump’s overall approval rating was lower than the last three presidents to win re-election in the United States: Barack Obama, George W. Bush, and Bill Clinton (The Gallup Organization, 2021).

Political misinformation and framing

Political misinformation research could be broadly categorized as studying either production/distribution or audience reception. Research focused on production (including both international and domestic propaganda), demonstrates how misinformation campaigns leverage existing media networks to extend the reach of their messaging and gain audience trust. Take the alternative influence network, a web of YouTube influencers who cross-promote their content and guests across channels and platforms (Lewis, 2018). This research mirrors parallel research on the RWME which demonstrates the important role networks play in disseminating and legitimizing conspiracy theories (Benkler et al., 2018). Recent work by Krafft and Donovan (2020) provides a detailed example of ‘trading up the chain’ (Marwick & Lewis, 2017), where the spread of misinformation traverses between social media and legacy networks to gain more credibility. False claims that a drug called hydroxychloroquine could treat COVID-19 circulated widely because those who spread these claims utilized social media’s network structure (Donovan, 2020).

In addition to a focus on networks, a great deal of misinformation research is also centered around Twitter. In their study of the Russian Internet Research Agency, a private company sponsored by the Russian government operating fake Twitter accounts to interfere with American political activities, Freelon and Lokot (2020) found that foreign agents leveraged existing political identities to maximize their impact on the platform. These findings also replicate Yin et al.’s (2018) study which revealed that identity claims applied to authentic account holders who could better relate to messaging with which they identified. Chen et al.’s (2021) analysis of tweets generated between March and August 2020, focused more on the origin of misinformation, finding that while the entire political spectrum has tweeted problematic content, a dense concentration of extremely conservative accounts tended to produce the most tweets containing misinformation.

On the reception side, researchers find that misinformation exposure has a negative impact on whether audiences trust their political and/or media systems (Ognyanova et al., 2020). Repeated exposure to misinformation may have an indirect and long-term impact on the democratic system as people’s presumed influence of misinformation could negatively impact trust in democracy more broadly (Nisbet et al., 2021). Research on the socio-political aspect of why and how misinformation is believed demonstrates the institutional and epistemological connections between the scope and spread of misinformation via social media (Anderson, 2021; Benkler et al., 2018; Lee et al., 2021; Marwick, 2018; Tripodi, 2018; Tripodi, 2021). Together, these studies reveal the network background and systematic nature of misinformation ecosystems. However, these studies tend to gloss over the way email is also a node within misinformation networks, and how it can serve an important role in framing events and reaching constituents.

Since the 1980s, sociologists have sought to explain how the media frame political events. Media frames are important because they organize the world for journalists and the public alike. Media frames contend and define social reality for social groups, institutions, and ideologies (Gitlin, 1980; Goffman, 1959). By creating their own catchphrases, paraphrasing or quoting their sources, and choosing what to focus on, journalists shape the discourse around a political event (Gamson & Modigliani, 1989). Frames also play an important role in mobilizing activist groups who construct meaning for political bystanders (Rohlinger & Quadagno, 2009). Previous research has focused on how constituents frame political events to try and exert influence over a politician’s actions (Rohlinger et al., 2015), but few have examined how politicians who hold the executive office use email as a framing mechanism.

Political theorists have argued that legislative communication effectively undermines the power of the fourth estate by directly engaging with the public and subverting the authority of the press (Graf, 2015). However, it may be that email correspondence does rely on media authority. This study fills that gap, drawing together existing research on political communication and mis/disinformation, while utilizing a sociological understanding of frame analysis. Specifically, we sought to understand how email correspondence from the White House relies on existing media networks to frame current events. Through frequency counts we were able to determine the most common sources relied on by the President in emails to newsletter subscribers during his last year in office. We then combined topic modeling and qualitative content analysis to understand how the administration framed the ongoing events to his supporters and the role the RWME played in legitimizing those claims.

Data & methods

The Official White House Newsletter is a form of e-mail correspondence started by the Obama Administration on the day of his inauguration (January 20, 2009). After Trump was elected, his administration continued the tradition, but renamed the newsletter The 1600 Daily in March 2017 and shortly thereafter made the content accessible via subscription only, removing its archive from whitehouse.gov (Bergman, 2018). The subscription consisted of two kinds of emails – a daily update and a weekly recap. Monday through Friday, subscribers would receive at least one email with the subject referring to a current event. For example, ‘NEW: Economic comeback under President Trump breaks 70-year record,’ or ‘President Trump announces massive Coronavirus testing expansion.’ Every Friday, the email subject line was titled ‘Five Stories President Trump Doesn’t Want You to Miss’ and included headlines, summaries, and hyperlinks to five news stories within the email.

The data for this paper consists of emails from The 1600 Daily (n = 460) sent between October 19, 2019 and January 20, 2021, when Biden’s administration took over the account. Our focus on email correspondence from the White House during Trump’s last year in office was twofold. First, we wanted to determine which news sources the White House relied on to support their arguments. Second, we wanted to understand what topics the administration focused on in the year leading up to the election and how they framed those topics.

Method 1: frequency count of cited media sources

To determine which news sources the administration referenced, we performed frequency counts using the nltk python package (Bird et al., 2009) and performed a basic named entity recognition, which is ‘the task of identifying entities mentioned in a text and labeling them’ (Wilcock, 2009, p. 20) using the python package spaCy (Honnibal & Montani, 2017). We first extracted all the proper nouns that are labeled as ‘ORG’ which consists of the names of companies and institutions. The frequency count was based on all the names labeled as ‘ORG’.

Method 2: Structural Topic Model (STM) & in depth interpretation of topics

To understand which topics the administration most frequently discussed in his newsletter, we leveraged Structural Topic Modeling (STM) using the stm R package (Roberts et al., 2014, 2019) which supports topic correlation and metadata analysis (Blei & Lafferty, 2007; Lucas et al., 2015). After removing stop words (e.g., ‘the’) and words appearing less than three times, the model was built using a total of 2,931 unique tokens from 446 documents (emails sent between October 18, 2019 – November 3, 2020). To determine the model’s hyperparameter K, which indicates the number of topics, a range of values between 3 and 11 (increases by 2) were explored. The models with different values of K were then compared to each other based on their coherence-exclusivity score. Coherence is a measure of the probability for a set of topic words to co-occur within the same document, while exclusivity is a measure of the probability for a word to fall primarily within the top rankings of a single topic (Roberts et al., 2014, 2019). Based on the joint consideration of these two scores, we selected a final value of 7 for K. Fourteen emails were sent between November 3, 2020 – January 20, 2021 for which we conducted a manual examination.

The STM returned the topic word list for the 7 topics. The appendix provides the prevalence of topics over time, an outline of the topics and word-clouds associated with each topic cluster, and the coherence and exclusivity scores of different topic number values. The topics appear cohesive and exclusive to each other, i.e., no overlapping or repeated words were found across topics. Operationally, STM supports the retrieval of exemplar documents through its embedded function findThoughts. Using this output, we analyzed five exemplar documents from each cluster to determine the top seven topics being discussed (e.g., ‘the confirmation of Judge Barrett to the Supreme Court,’ or ‘COVID-19’).

Method 3: qualitative coding of three topics

After the seven topics were identified, we conducted a thematic qualitative analysis of a subset of emails from each topic. Rather than focus on the frequency of phrases, this mode of analysis sought to identify the underlying meanings, patterns, and processes (Altheide, 2000) present within the White House emails pertaining to these subjects. During this phase, we relied on ‘grounded theory’ to move beyond description and construct thematic conceptions while simultaneously analyzing data (Charmaz, 2006; Charmaz & Thornberg, 2020).

Findings & discussion

Just another node in the network

Using frequency counts, we were immediately able to identify the top sources the White House relied on to back their claims in emails (see Table 1 for a list of the top ten sources). Out of the 559 total sources, 358 (approximately 64%) were associated with the RWME, a network fueled by conspiracy theories and fringe personalities who reject normative journalistic practices (Benkler et al., 2018). Nearly a quarter of all sourced information came from Fox News, which was cited 138 times during the year.

Table 1. The top 10 frequently cited sources.

No.	Entity name	Political leaning*	Frequency
1	Fox News	Right	138
2	The Wall Street Journal	Center	93
3	The New York Post	Right	91
4	The Washington Examiner	Right	78
5	CNN	Slight-left	32
6	USA Today	Center	32
7	The Washington Times	Right	31
8	The Washington Post	Slight-left	23
9	Fox Business	Middle	21
10	The Daily Caller	Right	20

*Political leaning identified in Benkler et al., (2018).

Based on our data, the White House relied primarily on sources within the RWME, but our initial frequency count also noted that CNN and The Washington Post were among the ten most frequently referenced sources in the content of the emails. Given Trump’s tendency to refer to both CNN and The Washington Post as ‘fake news’ during official briefings and at diplomatic gatherings like the G-7 summit (Segarra, 2018), we sought to further investigate this contradiction. A more thorough content analysis revealed that frequency counts alone did not tell the whole story.

When we followed hyperlinks associated with a mention of CNN or The Washington Post, we found that links routinely brought readers to publications within the RWME. For example, an email sent on October 29, 2019 claimed that corporate media like The Washington Post was trying to make the Trump administration’s successes look like political failures. The recap drew on three examples from The Washington Post to support the claims and then encouraged readers to ‘click here’ for more information. However, the hyperlink did not take audiences to The Washington Post to verify the information. Rather, it linked to an article in The Federalist titled ‘The Real Reasons Why Legacy Media Are Freaking Out Over Trump’s Successful Baghdadi Mission.’ For those unfamiliar, The Federalist is a conservative web-based magazine and podcast known for spreading misinformation and engaging in racist metadata practices (Bethea, 2020; Levenson, 2020). An email sent on April 15, 2020, alerted subscribers that ‘CNN anchor John King admits that the WHO dropped the ball in its early response to the Coronavirus, echoing criticisms from President Trump.’ Such a claim insinuates that CNN and Trump were aligned in how the administration handled COVID-19, but the link took readers to an article in The Daily Caller summarizing the CNN exchange, not to CNN itself.

The emails sent on October 29 and April 15 were not isolated cases. Over half of the thirty-one times CNN was mentioned, the email hyperlinked to a RWME outlet. These sources included The Daily Caller (four stories), The Daily Wire (one story), Fox News (three stories), The Free Beacon (two stories), The National Review (one story), The Federalist (two stories), Townhall (one story), The Washington Examiner (three stories), and The Wall Street Journal (one story). This strategy effectively leverages the original architecture of the world wide web while creating an illusion that dubious websites like The Federalist hold equal journalistic authority as CNN. It also allowed the administration to bury their connections to more far-right publications like The Federalist or The Free Beacon while encouraging subscribers to engage directly with RWME content.

Some mentions of The Washington Post were also hyperlinked to RWME sources (The Federalist, The Daily Wire, and Fox News were the destination sites for three of the links), but the rest of the active links took subscribers to opinion pieces, not news reports. This is significant because a recent study conducted by the Media Insight Project (2018) found that many people have difficulty assessing the validity of news media and that over half of the public surveyed do not know what an ‘op-ed’ is. Moreover, these op-eds were politically lopsided. All the opinion pieces linked in the emails were written by conservative pundits or politicians and many were affiliated with the Trump administration (e.g., Jared Kushner or then National Security Advisor Robert O’Brien).

For example, on May 11, 2020 the email linked to an article written by Mary Vought, the executive director of the Senate Conservatives Fund, who argued that churches should be allowed to conduct services during the pandemic and should be considered ‘essential.’ The email on May 6, 2020 featured an opinion piece co-authored by five Republican Governors on why COVID-19 protocols should be a matter of state’s rights. The administration’s decision to highlight the opinions of persons with no epidemiological experience on how businesses and places of worship should operate in a pandemic, framed COVID-19 as a matter of political differences, not a global public-health emergency. This glib position of the virus created long-lasting political divisions as documented by research conducted at the Pew Research Center who found that Republicans were less likely to wear a mask, follow social distance protocols, or get vaccinated because they are less likely than Democrats to see it as a health threat (Tyson, 2020).

In sum, our frequency counts revealed that the White House emails relied heavily on the RWME, playing an important role in closing what Benkler et al. (2018) refer to as ‘the propaganda feedback loop." Since politicians rely on media systems to get elected, the Trump administration was able to leverage a network that emphasized partisan-confirming news over truth. This closed circuit of identity-confirming ideas creates a perfect environment for manipulating public beliefs or attitudes for political gains (Benkler et al., 2018). Content analysis of hyperlinked content also revealed that the administration would frequently circulate unsubstantiated information via opinion pieces or less reputable news sources while referencing more established sources like CNN or The Washington Post in the body of the email.

Accentuate the positive

While frequency counts revealed the extent to which the administration relied on sources inside the RWME, it told us little regarding the topics the administration focused on. To quickly identify the common threads between a year’s worth of emails, we relied on topic modeling. Topic modeling has the capability to capture and disambiguate different uses of a word based on its context, which is a strong reason why it has been embraced by sociology (DiMaggio et al., 2013). This method was particularly useful for analyzing emails sent by the White House as these texts were frequently bound by multiple and even competing voices, what DiMaggio et al. (2013) referred to as ‘heteroglossia’ (p. 582). Initial models revealed seven core themes focused on by the Trump administration during his last year in office (for a breakdown of the topic clusters and word clouds associated with these clusters see Appendix).

Much like existing findings in legislative communication, the White House tended to focus on topics deemed more important to constituents who were crucial to Trump’s re-election (Cormack, 2013; Grimmer, 2013, 2016). These topics included the confirmation of Judge Amy Coney Barrett, conservative foreign policy, working-class job growth, and activated patriotism.

One of the greatest accomplishments of Trump’s presidency was his ability to nominate and secure conservative leaning judges (McCarthy, 2020). Not only did Trump appoint nearly the same number of federal appellate judges as Obama in half the time, his appointees effectively ‘flipped’ the balance of several appeals courts (Gramlich, 2018). The appointment of Coney Barrett was politically contentious, as her position on the Supreme Court filled the unexpected departure of Justice Ginsburg who passed away just a few months before the end of Trump’s Presidency (Baylon & Stephanis, 2020). However, our analysis revealed that the White House emails framed her confirmation as bipartisan.

In an email sent out on October 26, 2020 with the subject line ‘BREAKING: Senate Confirms Amy Coney Barrett’ the White House wrote that ‘After a unanimous vote by the Judiciary Committee last week’ the Senate confirmed Judge Barrett. The email also referenced her 2017 confirmation to the Circuit Court as one of ‘bipartisan support.’ Yet Coney Barrett’s 2017 and 2020 votes did not really garner support across the aisle. The vote on October 30, 2017 was split 54 Yea – 42 Nay, with only three Democratic Senators voting ‘Yea’ (Roll Call Vote 115th Congress – 1st Session, n.d.). The October 22, 2020 vote was only ‘unanimous’ because all the Democrats on the Judiciary Committee boycotted the event, meaning Barrett’s nomination did not receive a single Democrat vote (Senate Judiciary Committee, October 22, 2020).

Emails around foreign policy positioned Trump as an ‘America-first’ President. These emails also emphasized the masculine nature of the Trump administration to ‘stand up’ to China and Iran. The White House emails repeatedly argued that these actions made the United States ‘stronger’ and that ‘American strength, both military and economic, is the best deterrent’ for preventing terrorism on US soil (January 8, 2020). This link between protectionism and masculinity is perpetuated by conservative-leaning organizations as well, like the National Rifle Association (Carlson, 2015a, 2015b). Such a strategy successfully positions the White House as ‘hard’ or ‘objective,’ while stoking fears that the nation is becoming ‘too soft’ or ‘too feminine’ (Du Mez, 2020).

Not only did these emails activate ‘dog whistles’ around gendered stereotypes, they also conveyed a steady drumbeat of racial grievances wrapped up in color-coded solidarity (Flores-Yeffal et al., 2019; Haney-López, 2014). These codes were clearly demonstrated in emails focused on the growth of working-class jobs. For example, on January 13, 2020 an email highlighted ‘a blue-collar, working-class economic boom that began 3 years ago’ yet featured this statistic alongside figures that touted Trump’s commitment to ‘secure the border’ and eradicate ‘illegal immigration.’ A similar pattern was identified on October 19, 2020. The subject line of this email read ‘UPDATE: New border wall nears 200 miles’ but the body of the email began with statistics around the poverty rate in the United States, noting that it had hit a record low as a result of Trump’s ‘pro-growth, pro-worker policies.’ Juxtaposing immigration reform with job growth fuels the idea that ‘illegal immigrants’ ‘steal’ jobs from the working-class. Instead of framing immigration as our nation’s responsibility to accept and protect immigrants, refugees, and asylees; the White House emails appeal to the ‘deep stories’ of conservatism that frames immigrants as ‘line cutters’ (Hochschild, 2016; Polletta & Callahan, 2017).

Finally, the White House used frames to connect to Trump’s base by focusing on U.S.-specific events that emphasized patriotic themes, Christian blessings, and support of the military. These findings echo Rohlinger and Quadagno’s (2009) study of the conservative Christian political movement from 1970 to 1994 which revealed that leaders relied on frames that accentuated moral views and religious beliefs to connect with voters. For example, on November 27, 2019, the Thanksgiving email promised that millions of Americans will say a ‘prayer of thanks’ for ‘all of the men and women of the Armed Forces of the United States deployed around the world.’ This strategy of connecting with supporters via electronic communication might have helped the administration stay connected to and mobilize voters during the election cycle (Rohlinger & Bunnage, 2015).

Given existing scholarship, it is not particularly surprising that the administration focused on the concerns that resonated with Trump supporters. However, three of the top seven topics we identified also focused on subjects that could have negatively impacted Trump’s public perception: his impeachment, Black Lives Matter (BLM) protests surrounding the murder of George Floyd, and the spread of COVID-19 (Galston, 2020). Analyzing the content of emails associated with these topics reveals how the administration relied on the RWME to frame these events in ways that made the impeachment seem phony, allowed Trump supporters to think that COVID-19 was under control, and positioned BLM as a menace.

Resituate the negative

Rather than neglecting the topic of Trump’s presidential misconduct, the White House spent significant time framing Trump’s impeachment as a ‘sham’ orchestrated by the ‘far-left.’ This narrative involved multiple levels of framing. The first of which was to paint long-standing Democrats like Nancy Pelosi (who has served in Congress for 33 years) and Adam Schiff (who has served for 11 years) as ‘liars’ (January 30, 2020) who were working as part of a ‘socialist movement’ (December 19, 2019). The second step was to delegitimize Trump’s first impeachment as a bogus allegation. This included multiple emails that described the impeachment as a ‘political farce’ (October 21 and November 20, 2019) or ‘a sham’ (October 31; November 4, 13; December 4, 5, 17, 2019; and February 3, 2020) and claimed that his accusers ‘don’t care about the facts’ (October 18, 2019).

While the emails framed the Trump administration as ‘strong’ in foreign policy, the impeachment was described as the ‘weakest in U.S. history’ (December 19, 2019 and January 27, 2020). Once the President was impeached, the White House claimed Trump’s impeachment was actually bad for Democrats, and not the President, because the trial had wasted millions of government dollars on a ‘witch-hunt’ (January 29; February 5, 11, 2020).

Despite a growing death-toll from the pandemic, the White House repeatedly claimed that America was ‘more prepared than ever to overcome it [COVID-19].’ Sent on the first of April, Trump urged 30 days of ‘patriotism and selflessness’ in order to ‘beat’ the virus. Other emails in the COVID-19 corpus, spoke highly of Trump’s handling of the pandemic and shifted the blame to ‘the news media’ noting that ‘A majority of Americans, 55 percent, disapprove of how the media has handled the Coronavirus response’ (March 28, 2020). Like emails that discussed foreign policy, the administration’s emails claimed that Trump would ‘safeguard’ the U.S. from ‘others’ by relying on racist dog-whistles (e.g., ‘China Virus’ – August 25 and October 19, 2020).

Amid a global pandemic, civil rights protests around the United States formed in response to the murder of George Floyd at the hands of Minneapolis police officer Derek Chauvin and continued through much of the summer and into the fall. Starting in June, The White House repeatedly sent emails that framed these events as ‘violent mobs’ and ‘riots’ that attacked the police and local businesses, leaving destruction in their wake (June 2, 3, 4, 6, 8, 9, 12, 16, 17, 18, 29; July 13, 20, 21, 27, 29; August 3, 6, 11, 18, 26, 28, 31; September 1, 3, 4, 16, 17, 22, 25, 30; October 16, 28; November 3, 2020). Emails in this corpus also mentioned ‘Antifa’ (June 2, 2020) and condemned ‘far-left’ groups for acts of ‘domestic terrorism’ claiming they had ‘hijacked’ George Floyd’s death for their own gains (June 1, 2020). The emails also threatened an increase in crime if calls to ‘defund the police’ were heeded (June 16 and June 20, 2020).

The administration explicitly framed the events in Kenosha, WI as ‘violent riots’ in three emails in September and one on August 31 and described Trump’s visit to Wisconsin as an opportunity to ‘survey property damage’ (September 1, 2020). Framing social justice protest as violent attacks on the police force and places of commerce may have even influenced future judicial outcomes. For example, during the trial of Kyle Rittenhouse, prosecutors were told they could not refer to those killed in the Kenosha protests as ‘victims’ (Sullivan, 2021).

Topic absences

Finally, we applied an analytical lens to what was missing from our data set to capture the topics the White House deemed unimportant for discussion. Despite a decade recording the hottest temperatures in history and top scientists’ warnings of ‘mass extinction’ (Weston, 2021), the administration never acknowledged the rapid rise of global warming via email. While the White House framed COVID-19 as ‘under control’ and praised the administration’s actions, there were few calls to action for people to social distance or wear masks. In the three rare cases where ‘masks’ were mentioned (July 20, July 23, and August 5, 2020), the emails only referred to how the administration was working with corporations to donate or manufacture personal-protective equipment (PPE) for health care workers. Of the 460 emails analyzed, only one indicates that face coverings should be used (July 22, 2020), but defers to ‘state’s rights,’ a common ideological reference for conservative voters.

Most importantly, not a single email acknowledged the electoral outcome in which Donald Trump lost the 2020 Presidential election. In an email titled ‘WATCH: The best is yet to come,’ sent the day before election day, WhiteHouse.gov recapped Trump’s accomplishments noting that ‘America’s future is in our hands’ and encouraged voters to get to the polls.' On November 13, November 20, November 25, December 4, December 11, and December 18, the White House continued to send subscribers ‘Five Stories President Trump Doesn’t Want You To Miss’ emails. None of the information or news stories in these emails addressed the election, even though the Associated Press, Fox News, CNN, and all other news networks had effectively called Pennsylvania for Joe Biden by November 7, providing him with the 270 electoral votes he needed to secure the presidency (Cheney and Montellaro, 2020). The last email sent from the Trump administration went out on December 24, 2020, wishing ‘all Americans a Merry Christmas and a wonderful holiday season’ but still failed to acknowledge Trump’s loss. On January 20, 2021 the White House email service effectively transitioned to the Biden administration and linked subscribers to President Biden’s inaugural address.

Conclusion

The findings from this study provide important insights into how executive correspondence frames public events. Our data demonstrate that most topics the Trump administration focused on prioritized subjects deemed most important by conservative voters (the nomination of conservative judges, immigration reform, and patriotism) - thus replicating the findings from earlier studies (Cormack, 2013, 2016; Grimmer, 2013, 2016). However, our analysis also counters what one might expect. Rather than ignore topics that would have otherwise damaged his reputation (his impeachment, a global pandemic, and civil rights protests), the Trump administration spent many emails focused on these subjects to try and control subscribers’ perceptions of these events.

Based on our research, it is also clear that email correspondence coming out of the White House overwhelmingly relied on an information ecosystem rife with contentious and speculative claims. By encouraging their readers to ‘do their own research’ but providing them the hyperlinks directly, the White House emails reveal an intricate structure whereby conservative news producers work in tandem with elected officials, bouncing signals throughout their information networks. While an overreliance on the RWME is not particularly surprising, the way in which the White House emails made audiences think they were clicking on one source only to link to another (e.g., referencing The Washington Post but linking to The Federalist) undermines the exploratory search process and normalizes fringe outlets as reputable news sources. Such a finding expands on existing research by demonstrating the ways in which elected officials can rely on alternative influence networks to frame events that could otherwise negatively impact a candidate’s campaign.

Future work could also analyze the legislative communication of other conservative politicians. This could determine the extent to which other politicians promote the RWME and if they used the same tactics of burying the hyperlink while signaling to sources outside of the network (e.g., CNN or The Washington Post). More work is also needed on other executive email correspondence, for example emails sent from gubernatorial offices. Analyzing these data sources could provide sociological insight into how different states handled the COVID-19 pandemic and the extent to which they framed the virus as either a public health threat or as a hoax.

The topic prevalence chart revealed temporal trends that coincide with social movements, and intriguing trends that deserve a closer examination (see Appendix). For example, emails about COVID-19 peaked between April and May 2020 and dwindled in the months leading up to the election, at which point the White House began focusing on the confirmation of Judge Amy Coney Barrett. Understanding the temporal structure of frames could expand on existing literature that analyzes the role email correspondence plays in mobilizing political movements.

By rooting our analysis in a sociological understanding of frames, our study also reveals what the administration deemed unimportant. Failing to acknowledge the growing threat of global warming, the importance of using masks to stop the spread of COVID-19, and eventually his electoral loss, created important signals to his supporters. Among the most important, the White House’s failure to acknowledge Trump’s loss created a conceptual tool for those who supported him, fueling the conspiracy that the election was stolen. Moreover, not a single email condemned the attempted coup on January 6, 2021. This is telling when we compare this void to the number of emails sent which framed BLM protests as ‘riots’ attended by ‘domestic terrorists.’ While the Trump administration repeatedly sent electronic correspondence assuring his supporters that he would keep America safe from threats both abroad and homegrown, he failed do so. Not only was the Capitol ransacked on his watch, his administration never sent a single email denouncing the insurrection. Given the important role electronic correspondence plays in quickly mobilizing political groups (Rohlinger & Bunnage, 2015), the administration’s decision not to acknowledge Trump’s loss nor address the attempt to overthrow the United States government on January 6, 2021 demands more investigation.

Disclosure statement

No potential conflict of interest was reported by the authors.

Additional information

Notes on contributors

Francesca Tripodi

Francesca Tripodi is an Assistant Professor in the School of Information and Library Science and Senior Researcher at the Center for Information Technology and Public Life at UNC-Chapel Hill.

Yuanye Ma

Yuanye Ma is a PhD candidate in the School of Information and Library Science at UNC-Chapel Hill.

Appendix

Figure A1. Topic prevalence over time.

Figure A2. Topic one word cloud ‘activated patriotism’.

Figure A3. Topic two word cloud ‘confirmation of judge Amy Coney Barrett’.

Figure A4. Topic three word cloud ‘foreign policy’.

Figure A5. Topic four word cloud ‘the impeachment of president Trump’.

Figure A6. Topic five word cloud ‘civil rights protests’.

Figure A7. Topic six word cloud ‘working-class jobs’.

Figure A8. Topic seven word cloud ‘COVID-19’.

Figure A9. Semantic and exclusivity score for different numbers of topics (K).