Depth Charges: Does "Deep State" Propagandizing Undermine Bureaucratic Reputations?

Tyler Johnson*

The University of Oklahoma

In the wake of prominent instances of bureaucratic defiance, supporters of Donald Trump's presidency have taken to describing said bureaucrats and the departments and agencies they represent as part of a "deep state" seeking to maintain and wield power behind the scenes. Such claims can be understood as an attempt at character assassination with the end goal of undermining the reputations of bureaucracy and bureaucrats alike. Efforts to disseminate this propaganda across varied forms of media have been both sustained and forceful. Do such attempts to shape public opinion lead Americans to think less of prominent agencies, cabinet departments, and their leaders? The author utilizes an original survey experiment to examine if learning about what a deep state is, reading media members debate its reality, or hearing the President's son declare it to be truth shapes attitudes toward the image of the CIA and the Departments of State, Justice, and Defense. Preliminary results reveal such propagandizing rarely changes how individuals think about bureaucracy. The rare instances in which it does affect attitudes reveal such arguments may be just as likely to improve bureaucratic reputation as they are to diminish it, with presidential approval at times conditioning outcomes.

Keywords: bureaucracy, conspiracy theory, character assassination, deep state, propaganda, reputation, Trump


Within days of the September 2019 revelation that a whistleblower had filed a complaint concerning President Donald Trump and a potential attempt on his part to leverage foreign aid to Ukraine in exchange for investigations into a political rival, Trump administration defenders immediately went on the attack. These defenders attacked former Vice President Joe Biden, his son Hunter, the media, and Ukraine itself, but some of the sharpest barbs were reserved for the intelligence community as well as the federal bureaucracy more generally (Bump, 2019). Representative Matt Gaetz (R-Florida) claimed there were "people in the intelligence community and other parts of the government who just have it out for the President," while Senator Josh Hawley (R-Missouri) went a step further, putting a familiar name to this cabal in declaring the whistleblower part of "another Deep State attack" (Bump, 2019).

Claims that a "deep state" (in which appointed leaders of departments and agencies, everyday bureaucrats, or both clandestinely pursue their own agenda rather than that of elected officials in order to maintain power, achieve goals, or promote values) exists and is wreaking havoc when it comes to implementing administration initiatives have been made since civil service reform was passed in the United States in the late 1800s (Friedman, G., 2017). Some say President Dwight D. Eisenhower was alluding to such an idea in describing a "military-industrial complex" in his 1961 farewell address, and supporters of presidents like George W. Bush and Barack Obama claimed such influences at times shaped foreign policy making (Taub & Fisher, 2017). However, discussions of the presence, power, and reach of a deep state have become increasingly commonplace during President Trump's time in the White House.[1] Moreover, such claims have been discussed and dissected ad nauseam by media members over the past several years, seeking to understand how Washington works and whether a deep state truly exists or is a conspiracy theory (and subsequently an effort at propaganda those in power use to distract and deflect with a "conscious intent of manipulation" (Silverstein, 1987, p. 51).

In his 1965 book Propaganda: The Formation of Men's Attitudes, Jacques Ellul describes propaganda as "techniques of psychological influence combined with techniques of organization and the envelopment of people with the intention of sparking action" (p. xiii). Those who spread propaganda consider both content (what ideas are likely to influence opinions?) and process (how might that influence be achieved?). One motivation underlying the research presented here is to consider what drives the use of the idea of a deep state as a stealthy culprit pulling the strings behind any bureaucratic defiance. In addition, this research endeavors to learn more about whether modern deep state claims have any serious and significant effect on how Americans look at their government. Why might such claims affect the reputation of the bureaucracy, and do they do so? The goal of some Trump supporters in claiming a deep state is behind moves like leaks or whistleblowing (or, to borrow from Ellul, the "action" they wish to "spark") appears to be to assassinate the character of departments, agencies, and those who work within them in the eyes of average citizens. We know next to nothing about whether this strategy works, nor do we know which types of information are most likely to erode the public standing of those under attack.

This manuscript proceeds in the following manner. After a discussion of the literature on how elected officials attempt to control the bureaucracy and when and why bureaucrats might defy such attempts, I briefly chronicle the difficulties the presidential administration of Donald Trump has had keeping the bureaucracy in line. Supporters of his administration have used varied tools to try to bring the bureaucracy to heel, one of which is to undermine the reputation of a variety of agencies by linking them to a deep state. The reach and potential power of such a claim are examined through the lenses of research on propaganda, character assassination, conspiracy theory, and public and private sector reputation management. I then chronicle a survey experiment in which respondents are offered varying types of information (from Wikipedia, media experts, and the President's son) on the concept and potential reality of a deep state. The experiment serves as an initial effort at answering three related research questions: 1) does exposure to information on a deep state shape attitudes toward bureaucracy and its leaders, 2) what types of information might be more potent than others, and 3) which types of people appear most susceptible to such claims? Preliminary findings reveal the rarity with which receiving such propaganda has significant impact on attitudes toward departments, agencies, and their leaders. In most instances, the reputation of the bureaucracy remains unaffected and efforts at character assassination fail. At times, deep state information leads to positive assessments of the bureaucracy, although at others, interaction between this information and approval of the president leads departments, agencies, and their leaders to face backlash. I close with a discussion of some of the limits of this research that might offer avenues for exploring the claim of a deep state further and some conjecture on the short-term future of the deep state argument.

Political Control (Or Lack Thereof) of the Bureaucracy

As long as there has been a bureaucracy to control, politicians have sought to control it. Scholars of agency theory like Moe (1984) have illustrated how leaders (in the White House, Congress, or even at state and local levels) create bureaucracies with internal hierarchies to facilitate such control. Said leaders regularly attempt to use tools such as appointments, dismissals, reorganization, oversight, personnel, and budgeting to keep departments and agencies under their purview in line (Wood & Waterman, 1991).

Just because a leader wields such power, however, does not guarantee fealty to that leader's mission. Bureaucrats in general are often depicted as focusing on growing their budgets and protecting their turf, doing whatever possible to stave off change (Golden, 1992). As Brehm and Gates (1999) note, bureaucrats have their own policy preferences, often make political decisions, have discretion in decision making, and listen to other voices beyond those of elected officials; such voices might include supervisors, peers, or clients. Moreover, bureaucracies also bring their own values to the process of making decisions on how to best represent the public (Meier & O'Toole, 2006). Such factors might lead to what O'Leary (2019) calls "guerilla government," whereby those in public service disobey the people (be they elected or appointed) in charge of them. Their capacity for doing so is at times facilitated by the ability to stay clandestine in their defiance (O'Leary, 2019). Bureaucratic willingness to be responsive to or resist might also be conditioned by the history of responsiveness or resistance within a department or agency itself (Golden, 2000). One potential outcome of a disconnect between what leadership wants and what bureaucrats want is sabotage, what Brehm and Gates describe as "negative output" (1999, p. 30). Not all negative output of this sort is of equal impact; some defiance is forever unknown by those in charge, while other actions can be quite public and damaging (O'Leary, 2010). When threatened, bureaucrats might band together behind the scenes to try and improve their footing or appeal to outsiders like Congress or the media to generate broader impact (Golden, 1992).

Such disconnects between presidents and the bureaucracy have happened with regularity throughout the history of the United States. However, it can be argued that throughout his first term in office, President Donald J. Trump has clashed with bureaucrats with a regularity, a breadth, and a depth perhaps never seen before in executive-bureaucratic relations. Within days of Trump's inauguration, the White House found itself battling what they perceived as "negative output" from bureaucrats big and small, at times openly and publicly and at others clandestinely. In late January 2017, Acting Attorney General Sally Yates refused to enforce the administration's travel ban. In July of 2017, the Senate Homeland Security committee reported nearly daily national security leaks, and at a rate higher than previous administrations Republican or Democratic (Committee on Homeland Security and Governmental Affairs, 2017; Crowley, 2017). Just one month later, Environmental Protection Agency officials tipped off The New York Times about the potential quashing of a report on climate change (Crowley, 2017; Friedman, L., 2017). The following year, in September of 2018, The New York Times published an anonymous essay entitled "I Am Part of the Resistance Inside the Trump Administration" that chronicled government officials ignoring presidential directives (Anonymous, 2018). These are only a handful of the many examples of moments where bureaucrats expressed a willingness to defy what the Trump White House wanted or stood for.

The Trump administration's willingness to fight back against bureaucrats seen as unwilling to carry out policy aims can be measured in a number of ways. As Wood and Waterman (1991) illustrate, the appointment power is crucial to control. No one can deny a willingness on the part of the Trump White House to use this power with great regularity (and sometimes multiple times for the same position) to get preferred results; in the first 2 and a half years of his presidency, Trump turned over more positions in the Cabinet (9 of 15 spots) than any of his 5 predecessors did throughout an entire term (Gregorian, 2019b). Another perspective on using appointment powers has been the Trump White House's choice to give bureaucrats acting or interim control, or to not appoint anyone to fill vacancies at all. By one account in the summer of 2019, the Trump administration had over a dozen acting officials in key bureaucratic leadership roles that had yet to be submitted for Senate confirmation (Rod, 2019). Additionally, according to estimates in spring 2019, over 100 high-level bureaucratic positions had no nominee at all, while another 100+ were awaiting confirmation by the Senate (Gregorian, 2019a). According to some accounts, an overall philosophy of shrinking the bureaucracy (in terms of both money and manpower) has led to lower morale and difficulty in hiring due to the unattractiveness of government work as a career (Rein, 2017).

Another path by which the Trump White House and its supporters throughout government, the media, and the public have attempted to rein in the bureaucracy is by trying to win a battle over reputation. As scholars have illustrated (e.g. Arnold, et al., 2003; Bankins & Waterhouse, 2019; Carpenter & Krause, 2012; Hutton, et al., 2001; Luoma-aho, 2007; Maor, 2014; Wæraas & Sataøen, 2015), reputation management is a key facet of the relationship constructed and maintained between citizens and entities from public sector government agencies and departments to private sector interest groups and corporations. With positive reputations come greater legitimacy (Wæraas & Sataøen, 2015), and greater autonomy (Carpenter & Krause, 2012). Bureaucratic departments and agencies may not have the resources to build enormously positive reputations, so doing enough to maintain a neutral reputation may be sufficient to hold onto the public's perception that an entity is professional, trustworthy, cooperative, willing to serve, and for the common good (Maor, 2014). Reputation may also be crucial in situations where external threats arise (Hood, 2011). One such external threat to bureaucratic autonomy might be a White House seeking to wield greater control.

Propaganda and Character Assassination as Bureaucratic Control

As described by Carmeli and Tishler (2004), reputation is about competitive advantages. Propaganda can be crucial to creating such advantages. Jowett and O'Donnell (2018, p.6) characterize propaganda as "the deliberate, systematic attempt to shape perceptions, manipulate cognitions, and direct behavior to achieve a response that furthers the desired intent of the propagandist." What that "attempt" (as Jowett & O'Donnell put it) constitutes can mean many things to many people. It might take the form of something seemingly benign, like advertising (Cunningham, 2010), or public relations (Gelders & Ihlen, 2010). It might be agitating the public to take action, or it might be reshaping thought processes moving forward (Ellul, 1965). In some instances, it involves something much more malign, like "lies, distortions, fabrications, and exaggerations…manipulation, and brainwashing" (Cunningham, 2010, p. 13).

Such competitive advantages can be gained by building one's own reputation up, or by tearing one's competitor's reputation down. The speed with which departmental and agency actions (overt or covert) that run counter to White House directives are labeled as related to a deep state can be viewed as an example of the latter. It is a form of, to use a term chronicled as of late by authors like Icks, Shiraev, Keohane and Samoilenko (2020), character assassination. Character assassination is typically thought of in terms of its effects on individual leaders (be they modern or historic), but it has also been studied in the context of specific nations, policies, forms of media, and facets of government (Icks & Shiraev, 2014; Icks et al., 2020). Attempts at character assassination, according to those who study it, have 5 pillars: the attacker, the target, the medium, the context, and the audience (Icks et al., 2020). When it comes to claims of a deep state, the attackers are many (but typically elected Republicans or media pundits), the targets are widespread (but the label has been applied to many an agency, department, or bureaucrat deemed to have undermined the White House), the medium varies (from speeches to social media to mainstream and alternative press), and the context as of late has been the first term of the Trump presidency and the administration's success or lack thereof in accomplishing goals. The audience appears to be the American people, but this research hopes to shed greater light on whether or not those making a deep state claim have a reach that exceeds their grasp.

Character assassination is also both intentional and public (Icks et al., 2020). Those claiming a link between bureaucratic independence and a deep state intentionally use the phrase "deep state" not only because it is familiar to some, but also because the spirit underlying the phrase is timely (a necessity for successful propagandizing, according to Ellul, 1965) and resonates with many. As a 2018 Monmouth University Poll showed, 37% of Americans were either very familiar or somewhat familiar with the term deep state, but after hearing a description of the idea 71% said that it definitely or probably exists (Monmouth, 2018). Even if one is not familiar with the term, one might buy into it if one is like a clear majority of Americans and feels they can't trust in government (Pew Research Center, 2019).

Proponents of the deep state claim also use the phrase publicly. Ellul (1965) notes that for propaganda to succeed in achieving its goals, the propagandist "must utilize all the technical means at his disposal" (p. 9). Those willing to further the idea of a deep state have taken to traditional forms of media (e.g. print and broadcast media) as well as new technologies (e.g. social media) to link this premise with bureaucratic defiance. As described earlier, Senator Josh Hawley made his claim that a deep state was behind the Ukraine whistleblower story in a Fox News interview. When the Director of the Central Intelligence Agency gave an exclusive briefing to a handful of senators on the killing of Washington Post reporter Jamal Khashoggi by Saudi Arabia, Senator Rand Paul (R-Kentucky) claimed said exclusivity was the result of the power of a deep state, and did so to both the Associated Press and his Twitter followers (Paul, 2018a; 2018b). Even the President himself has used such language, and with increased frequency over time during his presidency (Rohde 2020). Examples include mentioning the "Deep State Justice Dept" in a tweet regarding investigating the security practices of his 2016 opponent Hillary Clinton's aide Huma Abedin, and his calling the State Department the "Deep State Department" in a press conference (Trump, 2018; Woodward, 2020).

Elected officials directly attempting to undermine the character of the bureaucracy via such propaganda hope for powerful effects, but the extent of their influence might be limited to those who feel a kinship with the speakers themselves (Pornpitakpan, 2004). Asch's (1948) experiments reveal that individuals' interpretation and incorporation of information can be conditioned by who authored it. As such, getting a broader set of elites to talk about such statements gives attempts at propaganda greater depth and resonance. Key also to the public face (and perpetuation) of the deep state claim is the extent to which mainstream media have been willing to openly debate its reality, filling inches of columns, minutes within the televised 24-hour cable news cycle, webpages, and tweets in the process. As Ellul (1965) notes, the media do not become instruments of propaganda automatically. Key to getting a self-sustaining discussion of the concept going has been these repeated claims of the deep state's existence by a variety of elite sources across different forms of media. This begins a conversation between experts and pundits, which later begets more serious stories written by mainstream newspapers, television shows, magazines, and websites about the nature of the debate taking place. The regularity with which prominent journalists have discussed whether a deep state exists has gone far to elevate the concept in political circles and potentially expose the concept to mass audiences. Journalists in the pages of hard news publications like The New York Times, Newsweek, Time, Politico, and The Washington Post all discussed, in long form articles devoted solely to the topic, the potential existence of a deep state within the first year of the Trump presidency (Abramson, 2017; Crowley, 2017; Porter, 2017; Taub & Fisher, 2017; Weigel, 2017). Later, so too did authors in pop cultural venues like Rolling Stone (Hafford, 2018). Propaganda's broader success is facilitated by a variety of sources delivering the same message (Harkins & Petty, 1981). The higher the overall volume of stories that said variety builds to, the more likely propaganda is to flourish (Paul & Matthews, 2016). As Iyengar and Kinder (1987) illustrate, such elite media behavior can not only set the agenda for those consuming it, but it can also prime how judgments are made and frame which elements of a story are most essential.

Character assassination of this sort may at first blush be about tearing down the reputation of one's opponents and undermining them, perhaps out of revenge (Icks et al., 2020). It may be an attempt to make bureaucrats or bureaucracy look like part of something sinister, shadowy, and power hungry. It may make departments, agencies, and their leaders appear as if they are enemies looking out for themselves or their entity instead of the executive branch or the American people. It has the potential to diminish trust, confidence, and approval as well. However, there are often other related goals is mind. Often there is a prize of some sort that the attacker and the attacked are in direct competition for. As discussed earlier, power and control are often at stake in the back and forth between a White House and the bureaucracy. Propagandists see information dominance as necessary to winning this sort of back and forth (Jowett & O'Donnell, 2018). Leaving rogue actions (actual or otherwise) by bureaucrats publicly unanswered has the potential to threaten that dominance. Soon after inauguration, the Trump administration saw bureaucratic leaking as a roadblock in accomplishing goals to rival President Obama's successes during his first 100 days in office (Rucker, Costa, & Parker, 2017). Months later, the bureaucracy (especially those entities related to intelligence gathering) was seen as conspiring to prevent President Trump from succeeding in attempts to keep his 2016 campaign promise of "draining the swamp," perhaps undermining the outcome of the election itself in the process (Porter, 2017). It is no surprise then that both of these moments (as well as several others since then) necessitated, as Ellul (1965) describes, a coherent response, one that takes events that might be vague or unclear and strives to organize them in the minds of members of the public. The idea of a deep state brings coherence to seemingly unrelated efforts on the part of bureaucrats across varied agencies with varied policy interests and influences to challenge a president's power. It also allows for blame to be shifted and scapegoating to take place, another set of weapons wielded by character assassins (Icks et al., 2020).

Conspiracy Thinking as Conduit of Propaganda

Why might those who promote the deep state claim feel such propaganda will resonate with a certain percentage of the public? One answer might lie in, as Goldwag (2012, p.14) described it, "America's long-standing penchant for conspiracy thinking." According to Barkun (2003), individuals who believe in conspiracies see everything happening by design and nothing happening randomly. Conspiracy theories also often have elements of powerful individuals or groups secretly wielding control over politics and society (Fenster, 2008). Conspiracy theorizing is done by individuals from all walks of life, across characteristics like level of education and political ideology (Oliver & Wood, 2014; Olmsted, 2009; Pipes, 1998; Uscinski & Parent, 2014); however, at times specific conspiracies do flourish with specific subgroups (Oliver & Wood, 2014). For example, partisans who feel threatened believe in conspiracy theories about their opponents (DiGrazia, 2017). Regardless of who believes, conspiracy theories, in the minds of those who believe them, often evoke a battle between dangerous, power-hungry conspirators and democracy itself (Dean, 1998; Sobiech, 2014).

What is it about the concept of a deep state that resembles what we think of as conspiracy theorizing? Sobiech (2014) argues that conspiracy theories require facts and a meaningful plot, and the idea of a deep state has both in great supply. Stories of bureaucrats acting contrarily have been commonplace during the Trump presidency, and the idea that they're doing so for selfish reasons (be it to maintain power or undermine the president) makes for a compelling and potentially impactful story. The aforementioned "I Am Part of the Resistance Inside the Trump Administration" essay published in 2018 is an excellent example of the sort of plot (someone is secretly trying to undermine the president) and facts (The New York Times has verified the author's identity) to which Sobiech alludes. To borrow from Barkun (2003), such conspirators are often not acting randomly, but instead are part of a broader design within departments and agencies we know have power regardless of who holds the White House. Some of the agencies associated with a deep state (e.g. the CIA) are also known for secrecy, further lending credence to the idea they might have the means to undermine a presidency in clandestine fashion. Additionally, conspiracies often have an element of government mismanagement to them as well (Olmstead, 2009), a claim that could be at the core of those who believe government departments and agencies have erred in contradicting the White House.

This propensity toward conspiracy on the part of some of the public is potentially powerful when fused with existing feelings about government, the expectations we have about who bureaucrats are in power to serve, and the extent to which the actions that draw deep state accusations run counter to those expectations. The bureaucratic actions that some have attributed to a deep state may trigger a mental "surveillance system" (Marcus, Neuman, & MacKuen, 2000), wherein individuals sense novel threats and search for ways to emotionally process such experiences. Belief in a deep state may allow some to maintain greater consistency in their attitudes about multiple facets of government as well, minimizing cognitive dissonance (Festinger, 1957). It may also be easier than undertaking the difficult task of contemplating the specifics of each new bureaucratic development; as Rosenberg (2008) discusses, individuals regularly struggle to learn and comprehend complex information in modern society.

The idea of a deep state may have a familiarity to it that makes it worth considering. In some ways, it evokes the long-standing concept of a military-industrial complex that President Dwight Eisenhower mentioned in his farewell address, but that scholars studied before his departure from the White House and in the six decades since (Baack & Ray, 1985; Ledbetter, 2011; Mills, 1956; Moskos, 1974). Such a concept had become so mainstream in American culture in the middle of the 20th century that, according to Pilisuk and Hayden (1965), more than 20 books were published on the topic in the 1950s and 1960s alone. Rohde (2020) argues that this facet of the deep state idea has appeal to liberals, while conservatives are more likely to buy in due to their desire for a powerful presidency (and the view that bureaucracy provides a roadblock to that). Modern Americans of all political persuasions also have grave concerns about whether they can trust government to do what is right (Pew Research Center, 2019). Only 17% of Americans feel government does what is right "just about always" or "most of the time." As such, specific elements of government must continuously fight a battle to earn and then maintain a reputation of trustworthiness amidst an atmosphere of skepticism. This battle takes place amidst others over power, goals, and territory that threaten the ability of departments and agencies to fulfill their missions (Bendor, Taylor, & Van Gaalen, 1987).

According to Carpenter (2010), the bureaucracy considers several facets of reputation that might matter to the public. One facet of agency reputation, according to Carpenter, is procedural, the extent to which it follows rules and norms. If a bureaucrat, an agency, or a department is leaking information or blowing the whistle on what it considers malfeasance, the possibility exists that such an action will be seen as straying from how bureaucracy normally functions. Another of these is what Carpenter calls moral reputation, wherein outsiders might consider the extent to which an agency protects stakeholder interests. Who an agency's stakeholders are, however, might differ from one person to the next. Some might see the American people as stakeholders when it comes to the entirety of government, while othersmight believe agencies first and foremost serve at the pleasure of the president.

To learn more as to whether attempts at disseminating deep state-related propaganda undermine the reputation of the bureaucracy, I conducted a survey experiment. This experiment endeavored to answer three related research questions:

RQ1: Does exposure to information on a deep state shape attitudes toward bureaucracy and its leaders?

RQ2: What types of information might be more potent than others?

RQ3: Which types of people appear most susceptible to such claims?



The survey was constructed in Qualtrics and conducted via Amazon Mechanical Turk in September of 2019.[2] After consenting to take part in the survey, participants completed 3 screening questions on income, party identification, and whether they lived in an urban, suburban, or rural area. The goal of this section was to guarantee partisan diversity in the respondent pool without tipping the respondent off as to the survey's intent.[3]

The 322 survey respondents were then randomly sorted into 1 of 4 treatment groups. 3 of these groups received varying amounts and levels of information on the concept and potential presence of a deep state in the United States; the 4th group served as a control. What I will call the "Definition Group" received the first paragraph of the Wikipedia entry on the concept of a deep state. This paragraph delivered a straightforward definition of a deep state ("a form of clandestine government made up of hidden or covert networks of power operating independently of a state's political leadership, in pursuit of their own agenda and goals"), a list of what types of departments or agencies might take part, and possible motivations for deep state activity (Wikipedia, 2019).[4]

What I will call the "Expert Group" also read this Wikipedia definition to begin, but then read excerpts from a February 2017 article in The New York Times on how leaks within and around the Trump White House have led some to compare the situation to a deep state (Taub & Fisher, 2017). In this article, the question of whether the United States is seeing its own deep state is pondered by discussing the normality and abnormality of leaks and the conflict between White Houses (present and past) and the bureaucracy. Barack Obama and George W. Bush-era struggles with leaks are mentioned, the perspective of a political science professor on the topic is mentioned, and the escalation of the issue in the Trump White House is addressed. This treatment group is meant to mimic one of the paths propaganda typically takes, wherein elites discuss the message in the mass media.

Finally, what I will call the "Partisan Group" began by reading the Wikipedia definition just like the Definition Group and the Expert Group did, but they then read a July 2017 tweet from Donald Trump Jr., son of President Donald Trump. In this tweet, the president's son retweets a Drudge Report tweet that states "Admin Hit With AT LEAST One National Security Leak a Day" (Drudge, 2017). Trump Jr. adds "If there was ever confirmation that the Deep State is real, illegal & endangers national security, it's this. Their interests above all else" (Trump Jr., 2017). This treatment group is meant to mimic another path propaganda typically takes, wherein elite co-partisans take their message directly to the people. In this instance, the path by which that message is conveyed is via a social media following.

Participants in all 3 of these treatment groups (plus the control group) were then asked questions about the performance of prominent departments and agencies. These departments and agencies included the Central Intelligence Agency (CIA), Department of Defense, Department of State, and Department of Justice. Departments and agencies like these, to one extent or another, have either modern ties to the concept of a deep state based on claims made by elected officials, administration staffers, or media pundits. Questions specifically asked included trustworthiness, whether these departments and agencies prioritized preserving their own power and status or helping serve the American people, and approval of the secretary or director in charge. Respondents were also asked more generally if unelected or appointed officials have too much influence in determining federal policy and to what extent the president should have the power to replace existing bureaucrats. Finally, participants in the Definition, Expert, and Trump groups were asked if they were familiar with the concept of a deep state before they took the survey, as well as if they believe a deep state exists.[5] The survey closed with a demographic and political questionnaire in which respondents were asked about gender, sexual orientation, race and ethnicity, age, income, education, religiosity, ideology, party identification, news consumption, and presidential approval.[6] Such questions revealed that the sample was 74% white and 58% male. About 1/3 of respondents were under the age of 30, while about 15% were 50 or older. The modal educational category (37%) had earned a bachelor's degree. In total, we can see that our sample is whiter, more male, younger, and more educated than the general public. The modal income category (32%) lived in a household that earned between $25,000 and $49,999 a year. 86% of respondents did not identify as gay, lesbian, or bisexual. Nearly a third of those surveyed identified as liberal or very liberal, while about a fifth identified as conservative or slightly conservative. Attitudes toward politics (as measured through party, ideology, and feelings about President Trump) were relatively similar across treatment groups. Overall, the data also align with a vast array of research (see Footnote 2 for examples) that illustrates how mTurk samples regularly differ from the general public in predictable but rarely problematic ways.


We begin by considering general attitudes of our sample on the bureaucracy writ large. Such survey findings might offer insight into the public's overall relationship with government beyond just elected officials and prominent institutions like Congress or the Supreme Court. Only 31% of our respondents felt that unelected or appointed officials have too much influence; the other 69% believed there is the right balance between elected and unelected officials. On the question of how much latitude presidents of one part should have in replacing government officials who took their positions under the administration of an opposing party, respondents were quite middling in nature. The modal response was "a moderate amount," chosen by 46% of those surveyed. Respondents were a bit more likely to select "a lot" or "a great deal" (31% total) than they were to say "a little" or "none at all" (23% total).[7]

As stated earlier, we focus our analysis of the relationship between reading about the deep state and attitudes toward bureaucracy on three specific dependent variables: trust in a department or agency, job approval of the secretary or director charged with leading the department or agency, and whether the department/agency preserves their own power and status or serves the American people. The first two of these variables are ordinal in nature, and as such ordered logit modeling is used to determine whether significant statistical relationships exist. The third variable is dichotomous and a logit model is used there. We focus our inquiry on 4 departments and agencies regularly associated with the conversation about a deep state: Defense, State, Justice, and the CIA.

Tables 1 and 2 reveal the impact of key variables of interest on attitudes toward departments, agencies, and their leaders. The clearest takeaway from these models in total is that there are rarely significant relationships between exposure alone to information on a deep state and attitudes toward departments and agencies. The Expert treatment (where respondents read the media story in addition to the definition) is the most successful, significant in 3 of 12 cases. In all 3 cases, the Expert treatment was linked with more positive feelings toward departments and their leaders. In general, those exposed to this treatment are more likely to trust the Department of Defense and approve of the job Secretary of Defense Mark Esper is doing. They are also more likely to believe the CIA's priority is to serve the American people. Such findings illustrate a potential for the deep state argument to backfire; institutions and their leaders alike were viewed significantly more positively in this handful of instances. Those exposed to the Partisan treatment (the tweet from Donald Trump, Jr.) were less likely to approve of the job CIA Director Gina Haspel is doing. In no cases was exposure to the Definition treatment significantly linked with questions of bureaucratic trust, leader approval, or department or agency priorities in terms of who is served.

In some cases, how a respondent feels about how the president is doing his job is related to how that same respondent feels about departments, agencies, and their leaders. This is clearest when it comes to individuals appointed to lead. Perhaps this is because these appointees are more directly tied to the president than departments and agencies themselves. Those who approve of Donald Trump's job performance as president also approve of Mark Esper's performance as Secretary of Defense, Mike Pompeo's performance as Secretary of State, and William Barr's performance as Attorney General. In fact, attitudes toward the Justice Department are positively linked to presidential approval across the board.

Perhaps, however, the interaction of these two variable types, receiving information about the deep state and feelings about the job Donald Trump is doing as president, might offer additional insight into how such messages matter. As discussed earlier, concepts like the deep state, in that they resemble other conspiracy theories, might thrive amongst specific political subgroups. Interactions at the bottom of Tables 1 and 2 illustrate how, in several cases, reading an informational treatment on the deep state and approving the job Donald Trump is doing as president is significantly linked with backlash against departments, agencies, and their leaders. This effect was most prominent amongst those who read the Expert treatment. Those who read the Expert treatment and also approved of the president were less likely to trust the Department of Defense and Secretary of Defense Mark Esper, and more likely to believe the Department of Justice and CIA were looking out for their own power and status. Decreased trust in the Department of Justice was also linked with the combination of exposure to the Definition treatment and approval of the job the president is doing. Interestingly though, a tweet from Donald Trump Jr. was not once linked with negative attitudes about bureaucracy from Trump advocates.

Tables 1 and 2 also reveal how rarely covariates explained attitudes toward the 4 departments and agencies under examination. Race, gender, sexual orientation, education, religiosity, and party identification are insignificant across all 12 models. Older individuals are more likely to believe the State Department is serving the American people. Individuals in households with higher incomes are more likely to trust the Department of Justice. More conservative individuals are more likely to approve of the job Attorney General William Barr is doing, while those who watch more news are more likely to disapprove.[8]

Additional data from this experiment allow us to consider if participants reported familiarity with the idea of a deep state before taking the survey, as well as whether they believed in the existence of a deep state. Of the respondents in the 3 treatment groups, 55% were "somewhat familiar" with the term "deep state," while the remaining 45% were split nearly equally between "not familiar" and "very familiar." Nearly three-quarters (74%) stated that a deep state probably exists or definitely exists.[9] Noticeably more of these individuals (close to double, in fact) were in the "probably" group than were the "definitely" group though. Only a little over 5% said a deep state definitely does not exist, leaving just over 20% in the "probably does not exist" group. There were no significant differences on either of these questions, however, across treatment types.











Pompeo Approval
Definition .58 (.66) .09 (.77) .14 (.65) .95 (.64) .66 (.73) .86 (.68)
Expert 1.61* (.69) 1.31 (.79) 1.43* (.71) .43 (.70) 1.17 (.77) .65 (.73)
Partisan .04 (.68) .27 (.79) .02 (.68) -.61 (.68) .11 (.76) .05 (.72)
Age .01 (.01) .02 (.01) -.00 (.01) .01 (.01) .03* (.01) .02 (.01)
Income .17 (.09) -.02 (.10) -.02 (.09) .05 (.08) -.09 (.10) -.00 (.09)
Race -.12 (.26) .07 (.29) .14 (.27) .01 (.26) -.07 (.28) .24 (.27)
Gender .07 (.23) -.14 (.27) -.10 (.24) .04 (.23) .11 (.25) -.04 (.24)
Sexual Orientation .40 (.37) .45 (.42) .31 (.36) .24 (.36) .38 (.40) .36 (.38)
Education -.01 (.08) .01 (.09) .02 (.08) .12 (.07) .01 (.08) .05 (.08)
Religiosity -.12 (.09) -.14 (.10) -.14 (.09) -.13 (.09) -.11 (.10) -.15 (.09)
Ideology .06 (.10) .09 (.12) .11 (.11) -.14 (.10) -.14 (.12) .11 (.11)
Party ID .06 (.10) .12 (.12) .17 (.11) .10 (.11) .17 (.12) .11 (.11)
News -.01 (.07) -.12 (.08) .05 (.07) .03 (.07) -.05 (.07) -.11 (.07)
Trump .50 (.27) .57 (.31) .70* (.28) .47 (.27) .44 (.30) 1.05* (.28)
TrumpXDefinition -.10 (.32) -.07 (.39) -.13 (.33) -.44 (.32) -.30 (.36) -.58 (.34)
TrumpXExpert -.69* (.33) -.65 (.39) -.74*(.35) -.36 (.34) -.69 (.38) -.47 (.35)
TrumpXPartisan .22 (.35) -.12 (.39) -.04 (.34) .25 (.33) -.02 (.37) -.22 (.35)
N=322. * = p < .05. Trust is a 3-point scale ranging from untrustworthy to trustworthy. Serve is a 2-point scale ranging from "preserving their own power and status" to "helping serve the American people." Approval is a 3-point scale ranging from disapprove to approve. Trust and Approval are ordered logit models; Serve is a logit model. On race, white=1. On gender, male=1. On sexual orientation, heterosexual=1.

Table 1 Effects of Deep State Information on Attitudes Toward Defense and State












Haspel Approval
Definition .89 (.64) .56 (.73) .77 (.70) -.05 (.66) .57 (.72) 1.01 (.66)
Expert .78 (.71) 1.36 (.77) 1.10 (.74) .38 (.70) 1.66* (.77) -.16 (.69)
Partisan -.13 (.68) -.26 (.77) .11 (.73) -.21 (.68) .02 (.77) -1.52* (.68)
Age -.00 (.01) .01 (.01) .02 (.01) .01 (.01) .02 (.01) .02 (.01)
Income .16* (.08) .01 (.10) .03 (.09) .14 (.09) .12 (.09) .04 (.09)
Race -.04 (.25) .19 (.29) .01 (.27) -.17 (.25) -.34 (.28) .16 (.26)
Gender .00 (.23) .16 (.26) -.01 (.24) -.06 (.23) -.20 (.25) .27 (.23)
Sexual Orientation .01 (.36) -.41 (.40) .20 (.38) .33 (.37) .54 (.41) -.00 (.36)
Education .04 (.07) .11 (.08) .05 (.08) -.02 (.08) -.03 (.08) .05 (.07)
Religiosity -.15 (.09) -.11 (.10) -.14 (.09) -.15 (.19) .08 (.10) -.10 (.09)
Ideology -.19 (.11) .19 (.12) .23* (.11) .02 (.10) -.06 (.12) .00 (.11)
Party ID .14 (.11) -.13 (.12) .08 (.11) .01 (.11) .14 (.12) .08 (.10)
News .01 (.07) .00 (.07) -.18* (.07) .01 (.07) .02 (.08) .01 (.07)
Trump .70* (.27) .69* (.30) 1.07* (.29) -.06 (.27) .35 (.30) .26 (.28)
TrumpXDefinition -.70* (.32) -.56 (.36) -.45 (.35) .07 (.33) -.22 (.35) -.45 (.34)
TrumpXExpert -.49 (.35) -.80* (.38) -.61 (.36) -.26 (.34) -.98* (.39) -.23 (.35)
TrumpXPartisan -.02 (.34) .12 (.39) -.05 (.37) .13 (.33) -.11 (.37) .66 (.35)
N=322. * = p < .05. Trust is a 3-point scale ranging from untrustworthy to trustworthy. Serve is a 2-point scale ranging from "preserving their own power and status" to "helping serve the American people." Approval is a 3-point scale ranging from disapprove to approve. Trust and Approval are ordered logit models; Serve is a logit model. On race, white=1. On gender, male=1. On sexual orientation, heterosexual=1.

Table 2 Effects of Deep State Information on Attitudes Toward Justice and CIA


In total, the findings presented in the previous section offer preliminary evidence that the effect of the deep state concept on attitudes toward government is limited in nature. Exposure to deep state information itself is rarely significant. The same goes for the interaction between exposure and presidential job approval. Generally speaking, efforts at disseminating deep state propaganda to assassinate the character of bureaucratic agencies and their leaders appear, for the most part, to fail.

There are, however, some trends worth noting when relationships are significant. Of the three treatment types, the Expert information from the media is more likely to have an impact on attitudes than the straightforward Definition information or the Partisan information from Donald Trump, Jr.'s Twitter account. When exposure to the Expert information is significant, it is always linked with stronger support for bureaucracy broadly defined. When it is interacted with presidential job approval, however, it always signals a backlash against bureaucracy for those who support the job the president is doing. Ellul's (1965) research on propaganda notes the central role of mass media in accomplishing goals. Exposure to the media account of the deep state has the most power, and that power can cut in either direction. The nature of the significant interactions reveals that for those pushing the idea of the deep state in an attempt to undermine the character of departments and agencies, this only seems to work as intended on a subset of the population (Trump supporters). The use of this propaganda, however, risks broader backlash in some cases as well. The fact that, at times, information on the deep state does work on such a specific partisan subset lends credence to arguments made by Pornpitakpan (2004) regarding in-group source credibility, van Prooijen and van Dijk (2014) regarding conspiracy theory acceptance in the face of threats to one's group, as well as Stanley (2015), who stated more specifically that the American two-party system has created strong group identities that allow for susceptibility to propaganda.

What emerges from this pattern of findings in terms of the literature and theories discussed earlier is that the belief that tying federal departments and agencies to the idea of a deep state will regularly diminish the reputation of various entities within the bureaucracy appears to be misguided. Instead, the impacts seem irregular and unpredictable. In some cases, reading about the idea of a deep state (especially via media experts) was linked with respondents supporting departments, agencies, and their leaders. If some hope to use the idea of a deep state as part of a propaganda effort to damage the standing and assassinate the character of bureaucracy and bureaucrats, the ability to do so might be dependent on whether the individual being targeted has positive feelings toward President Trump. This suggests a narrow playing field, perhaps one in which media discussion of the idea of a deep state solely builds antigovernment sentiment amongst those already on the side of the White House.

Future Directions

One caveat to these findings lies in the experimental design itself, which only asks about attitudes toward departments, agencies, and their leaders after the informational treatments. A future variation on this experiment could ask respondents for their thoughts on the deep state and on bureaucratic agencies, then offer information, and repeat the same survey questions (either immediately or after some time) to determine if and how opinions are changed. Such a variation would allow research to speak more deeply to the idea of opinion change. We know that changes in partisan control are linked with belief in conspiracies (DiGrazia, 2017), so considering who is in the White House and for how long is also something to account for in future research. The survey in this research was conducted just over 2.5 years into the Trump presidency; had it been executed in 2017 or mid-2020, responses might have differed. Were the Democrats to win back the White House following the 2020 presidential election, findings might change entirely.

Another possible path for future research on this topic to consider is the power of varying volumes of deep state messaging on public opinion toward bureaucracy. The survey experiment presented here is one in which a single definition, news story, or tweet is given to respondents. We know, however, that discussions of the deep state ebb and flow over time. New controversies appear and new conflicts arise between presidents and federal departments and agencies. Scholars conducting longitudinal public opinion surveys could regularly ask questions about the deep state or about bureaucracy, then link that to the flow of discussion about the deep state in media or on social media. Varying survey experiments could test the impact of hearing about the state from different sources other than the ones used here, or multiple sources simultaneously. Such studies would give us even richer and deeper knowledge on the impact of this concept than what is able to be derived in this research. Subsequent versions of the work presented here would be wise to consider how other attitudes toward politics and the state of the nation today might influence attitudes toward bureaucracy as well. One's feelings about the present and future might make one more susceptible to the arguments made by those claiming a deep state exists and has impact. Work like that of DiGrazia (2017) on individual-level feelings of threat and insecurity and propensity to believe in conspiracies could be integrated into future iterations of survey design. Finally, future research might test the effect of information arguing against the presence of a deep state as well, to see if it converts those who once believed or only serves to harden their opinions, akin to findings of those like Festinger, Riecken, and Schachter (1956).

Subsequent attempts at research of this sort could also consider the survey sample itself. Future research might utilize other survey populations beyond that of Amazon Mechanical Turk for the purpose of replication, allowing us to confirm these initial findings. There are also ways in which more focused analyses on survey subsamples might also prove interesting regarding this phenomenon. A survey focused solely on presidential co-partisans, for example, would give researchers greater statistical power how efforts to attack the character and undermine the reputation of departments and agencies via links to a deep state have potency amongst those on the right. Ellul (1965) notes how propaganda "standardizes current ideas, hardens prevailing stereotypes, and furnishes thought patterns in all areas" (p.163). One could perceive how discussion of the deep state might exacerbate previously held feelings about government of this sort amongst those who support the president no matter what. This might also prove to be fertile ground for seeing if different experts from the ones utilized here might have even more or less impact. One could easily envision a story on the deep state coming from The New York Times or The Washington Post being deemed "fake news" and not shaping the opinions of Trump supporters, but a message from Fox News being particularly potent.


What then of the conversation about the deep state moving forward? Even though the findings here suggest that it mostly fails to move the needle in terms of shaping the reputation (let alone assassinating the character) of most of its bureaucratic targets, we should undoubtedly expect it to live on (and probably even after Donald Trump's time in the White House comes to an end). In fact, during the COVID-19 pandemic in early 2020, government experts like Dr. Anthony Fauci were accused of being a member of the deep state (Broderick, 2020), and the deep state was described as a co-conspirator (along with China, George Soros, the World Health Organization, Bill Gates, and Bill and Hillary Clinton) in a plot to control the world's population by controlling potential coronavirus cures (Fried, 2020). Its future as an implement in the propaganda and character assassination toolkit may lie in its simplicity. It appears flexible enough to be applied to any department, agency, bureaucrat, or politician. It can be rooted in existing feelings about government, American history, or current events. It provides a natural foil for politicians to rally against. It makes for an interesting story to a media seeking to gain and maintain audiences. It attracts the portion of the public that is intrigued by conspiracy theorizing. What we shouldn't expect, however, is for it to do lasting damage on the jobs that those in federal departments and agencies do every day in executing the laws and executive orders authorized by presidents past and present.


Abramson, A. (2017, March 8). President Trump's allies keep talking about the 'deep state.' What's that? Time.

Anonymous. (2018, September 5). I am part of the resistance inside the Trump administration. The New York Times.

Arnold, J., Coombs, C., Wilkinson, A., Loan-Clarke, J., Park, J., & Preston, D. (2003). Corporate images of the United Kingdom National Health Service: Implications for the recruitment and retention of nursing and allied health profession staff. Corporate Reputation Review, 6(3), 223-238.

Asch, S. (1948). The doctrine of suggestion, prestige, and imitation in social psychology. Psychological Review, 55(5), 250-276.

Baack, B., & Ray, E. (1985). The political economy of the origins of the military-industrial complex in the United States. The Journal of Economic History, 45(2), 369-375.

Bankins, S., & Waterhouse, J. (2019). Organizational identity, image, and reputation: Examining the influence on perceptions of employer attractiveness in public sector organizations. International Journal of Public Administration, 42(3), 218-229.

Barkun, M. (2003). A culture of conspiracies: Apocalyptic visions in contemporary America. Berkeley, CA: University of California Press.

Bendor, J., Taylor, S., & Van Gaalen, R. (1987). Stacking the deck: Bureaucratic missions and policy design. American Political Science Review, 81(3), 873-896.

Berinsky, A., Huber, G., & Lenz, G. (2012). Evaluating online labor markets for experimental research:'s Mechanical Turk. Political Analysis, 20(3), 351-368.

Brehm, J., & Gates, S. (1999). Working, shirking, and sabotage: Bureaucratic response to a democratic public. Ann Arbor, MI: University of Michigan Press.

Broderick, R. (2020, April 15). A Youtube video accusing Dr. Anthony Fauci of being part of the deep state has been viewed over 6 million times in a week. Buzzfeed News.

Buhrmester, M., Kwang, T., & Gosling, S. (2011). Amazon's Mechanical Turk: A new source of inexpensive, yet high-quality data? Perspectives on Psychological Science, 6(1), 3-5.

Bump, P. (2019, September 23). Trump's allies have found the real culprit in this Ukraine mess: Everyone but Trump. The Washington Post.

Carmeli, A., & Tishler, A. (2004). The relationships between intangible organizational elements and organizational performance. Strategic Management Journal, 25(13), 1257-1278.

Carpenter, D. (2010). Reputation and power: Organizational image and pharmaceutical regulation in the FDA. Princeton, NJ: Princeton University Press.

Carpenter, D., & Krause, G. (2012). Reputation and public administration. Public Administration Review, 72(1), 26-32.

Committee on Homeland Security and Governmental Affairs. (2017). State secrets: How an avalanche of media leaks is harming national security.

Crowley, M. (2017, August/September). The deep state is real. But it might not be what you think. Politico.

Cunningham, S. (2002). The idea of propaganda: A reconstruction. Westport, CT: Greenwood Publishing Group.

Dean, J. (1998). Aliens in America: Conspiracy cultures from outerspace to cyberspace. Ithaca, NY: Cornell University Press.

DiGrazia, J. (2017). The social determinants of conspiratorial ideation. Socius: Sociological Research for a Dynamic World, 3(1), 1-9.

Drudge, M. [@drudge_report]. (2017, July 7) Admin hit with at least one national security leak a day. [Tweet].

Ellul, J. (1965). Propaganda: The formation of men's attitudes. New York, NY: Alfred A. Knopf.

Fenster, M. (2008). Conspiracy theories: Secrecy and power in American culture. Minneapolis, MN: University of Minnesota Press.

Festinger, L. (1957). A theory of cognitive dissonance. Stanford, CA: Stanford University Press.

Festinger, L., Riecken, H., & Schachter, S. (1956). When prophecy fails. Minneapolis, MN: University of Minnesota Press.

Fried, I. (2020, May 11). Coronavirus misinformation is a tricky foe for tech. Axios.

Friedman, G. (2017, March 15). The deep state is a very real thing. HuffPost.

Friedman, L. (2017, August 7). Scientists fear Trump will dismiss blunt climate report. The New York Times.

Gelders, D., & Ihlen, Ø. (2010). Government communication about potential policies: Public relations, propaganda, or both? Public Relations Review, 36(1), 59-62.

Goldwag, A. (2012). The new hate: A history of fear and loathing on the populist right. New York, NY: Pantheon Books.

Golden, M. (1992). Exit, voice, loyalty, and neglect: Bureaucratic responses to presidential control during the Reagan administration. Journal of Public Administration Research and Theory, 2(1), 29-62.

Golden, M. (2000). What motivates bureaucrats?: Politics and administration during the Reagan years. New York, NY: Columbia University Press.

Gregorian, D. (2019a, May 15). Trump revolving door creates 'unprecedented' vacancy mess in government. NBC News.

Gregorian, D. (2019b, September 23). Off the charts: White House turnover is breaking records. NBC News.

Hafford, M. (2018, April 22). Deep state: Inside Donald Trump's paranoid conspiracy theory. Rolling Stone.

Harkins, S., & Petty, R. (1981). The multiple source effect in persuasion: The effects of distraction. Personality and Social Psychology Bulletin, 7(4), 627-635.

Hood, C. (2011). The blame game: Spin, bureaucracy, and self-preservation in government. Princeton, NJ: Princeton University Press.

Huff, C., & Tingley, D. (2015). Who are these people? Evaluating the demographic characteristics and political preferences of MTurk survey respondents. Research & Politics, 2(3), 1-12.

Hutton, J., Goodman, M., Alexander, J., & Genest, C. (2001). Reputation management: The new face of corporate public relations? Public Relations Review, 27(3), 247-261.

Icks, M., & Shiraev, E. (2014). Character assassination throughout the ages. New York, NY: Palgrave Macmillan.

Icks, M., Shiraev, E., Keohane, J., Samoilenko, S. A. (2020). Character assassination: Theoretical framework. In S. A. Samoilenko, M. Icks, J. Keohane, & E. Shiraev (Eds.), Routledge handbook of character assassination and reputation management (pp. 11-25). London/New York: Routledge.

Iyengar, S., & Kinder, D. (1987). News that matters: Television and American opinion. Chicago, IL: University of Chicago Press.

Jowett, G., & O’Donnell, V. (2018). Propaganda & persuasion. Thousand Oaks, CA: Sage Publications.

Ledbetter, J. (2011). Unwarranted influence: Dwight D. Eisenhower and the military industrial complex. New Haven, CT: Yale University Press.

Levay, K., Freese, J., & Druckman, J. (2016). The demographic and political composition of Mechanical Turk samples. SAGE Open, 6(1), 1-17.

Luoma-Aho, V. (2007). Neutral reputation and public sector organizations. Corporate Reputation Review, 10(2), 124-143.

Maor, M. (2014). Theorizing bureaucratic reputation. In A. Wæraas & M. Maor (Eds.), Organizational reputation in the public sector (pp. 31-50). Abingdon, UK: Routledge.

Marcus, G., Neuman, W., & MacKuen, M. (2000). Affective intelligence and political judgment. Chicago, IL: University of Chicago Press.

Meier, K., & O’Toole Jr., L. (2006). Political control versus bureaucratic values: Reframing the debate. Public Administration Review, 66(2), 177-192.

Mills, C. (1956). The power elite. Oxford, UK: Oxford University Press.

Moe, T. (1984). The new economics of organization. American Journal of Political Science, 28(4), 739-777.

Monmouth University Poll. (2018, March 19). National: Public troubled by deep state.

Moskos Jr., C. (1974). The concept of the military-industrial complex: Radical critique or liberal bogey?. Social Problems, 21(4), 498-512.

Mullinix, K., Leeper, T., Druckman, J., & Freese, J. (2015). The generalizability of survey experiments. Journal of Experimental Political Science, 2(2), 109-138.

O’Leary, R. (2010). Guerrilla employees: Should managers nurture, tolerate, or terminate them?. Public Administration Review, 70(1), 8-19.

O’Leary, R. (2019). The ethics of dissent: Managing guerrilla government, third edition. Washington, DC: CQ Press.

Oliver, J., & Wood, T. (2014). Conspiracy theories and the paranoid style(s) of mass opinion. American Journal of Political Science, 58(4), 952-966.

Olmsted, K. (2009). Real enemies: Conspiracy theories and American democracy, World War I to 9/11. Oxford: Oxford University Press.

Paolacci, G., & Chandler, J. (2014). Inside the Turk: Understanding Mechanical Turk as a participant pool. Current Directions in Psychological Science, 23(3), 184-188.

Paolacci, G., Chandler, J., & Ipeirotis, P. (2010). Running experiments on Amazon Mechanical Turk. Judgment and Decision Making, 5(5), 411-419.

Paul, R. [@randpaul]. (2018a, December 4). Do you want to know what the deep state is? The CIA director is coming to the US Senate and only briefing a select few members of the Senate. [Tweet].

Paul, R. (2018b, December 10). Senator compares exclusive Khashoggi briefing to deep state. The New York Times.

Paul, C., & Matthews, M. (2016). The Russian `firehose of falsehood' propaganda model: Why it might work and options to counter it. Rand Corporation.

Pew Research Center. (2019, April 11). Public trust in government, 1958-2019. Pew Charitable Trust.

Pilisuk, M., & Hayden, T. (1965). Is there a military industrial complex which prevents peace?: Consensus and countervailing power in pluralistic systems. Journal of Social Issues, 21(3), 67-117.

Pipes, D. (1999). Conspiracy: How the paranoid style flourishes and where it comes from. New York, NY: The Free Press.

Pornpitakpan, C. (2004). The persuasiveness of source credibility: A critical review of five decades’ evidence. Journal of Applied Social Psychology, 34(2), 243-281.

Porter, T. (2017, August 2). Deep state: How a conspiracy theory went from political fringe to mainstream. Newsweek.

Rein, L. (2017, December 31). How Trump's first year has decimated federal bureaucracy. The Independent.

Rod, M. (2019, July 3). The Trump administration's highest ranks are riddled with officials in 'acting' roles. CNBC.

Rohde, D. (2020). In deep: The FBI, the CIA, and the truth about America's "deep state." New York, NY: W.W. Norton & Company.

Rosenberg, S. (2008). The not so common sense: Differences in how people judge social and political life. New Haven, CT: Yale University Press.

Rucker, P., Costa, R., & Parker, A. (2017, March 5). Inside Trump's fury: The president rages at leaks, setbacks, and accusations. The Washington Post.

Silverstein, B. (1987). Toward a science of propaganda. Political Psychology, 8(1), 49-59.

Sobiech, M. (2014). The ethos of conspiracy argument: "Character" as persuader in conspiracy rhetoric. (Paper No.1359) [Doctoral dissertation, The University of Louisville]. The University of Louisville’s Institutional Repository.

Stanley, J. (2015). How propaganda works. Princeton, NJ: Princeton University Press.

Taub, A., & Fisher, M. (2017, February 16). As leaks multiply, fears of a 'deep state' in America. The New York Times.

Trump, D. [@realdonaldtrump]. (2018, January 2). Crooked Hillary Clinton's top aid, Huma Abedin, has been accused of disregarding basic security protocols. She put classified passwords into the hands of foreign agents. [Tweet].

Trump Jr., D. [@donaldjtrumpjr]. (2017, July 7). If there was ever confirmation that the deep state is real, illegal & endangers national security, it's this. Their interests above all else. [Tweet].

Uscinski, J., & Parent, J. (2014). American conspiracy theories. Oxford, UK: Oxford University Press.

Van Prooijen, J., & van Dijk, E. (2014). When consequence size predicts belief in conspiracy theories: The moderating role of perspective taking. Journal of Experimental Social Psychology, 55, 63-73.

Wæraas, A., & Sataøen, H. (2015). Being all things to all customers: Building reputation in an institutionalized field. British Journal of Management, 26(2), 310-326.

Weigel, D. (2017, March 7). Trump and Republicans see a 'deep state' foe: Barack Obama. The Washington Post.

Wikipedia. (2019). Deep state.

Wood, B., & Waterman, R. (1991). The dynamics of political control of the bureaucracy. American Political Science Review, 85(3), 801-828.

Woodward, A. (2020, March 20). Trump pretends he doesn't know Sean Spicer, refers to "deep state department" and snaps at reporter during tense coronavirus conference. The Independent.


Appendix A: Experimental Prompts

Prompt 1: Definition

A deep state (from Turkish: derin devlet), also known as a state within a state, is a form of clandestine government made up of hidden or covert networks of power operating independently of a state's political leadership, in pursuit of their own agenda and goals. Examples include organs of state, such as the armed forces or public authorities (intelligence agencies, policy, secret police, administrative agencies, and government bureaucracy). A deep state can also take the form of entrenched, career civil servants acting in a non-conspiratorial manner, to further their own interests. The intent of a deep state can include continuity of the state itself, job security for its members, enhanced power and authority, and the pursuit of ideological objectives. It can operate in opposition to the agenda of elected officials, by obstructing, resisting, and subverting their policies, conditions, and directives. It can also take the form of government-owned corporations or private companies that act independently of regulatory or governmental control.

Prompt 2: Expert (Definition + News Story)

WASHINGTON – A wave of leaks from government officials has hobbled the Trump administration, leading some to draw comparisons to countries like Egypt, Turkey, and Pakistan, where shadowy networks within government bureaucracies, often referred to as "deep states," undermine and coerce elected governments.

So is the United States seeing the rise of its own deep state?

Though leaks can be a normal and healthy check on a president's power, what's happening now extends much further. The United States, those experts warn, risks developing an entrenched culture of conflict between the president and his own bureaucracy.

Though American democracy is resilient enough to resist such clashes, early hints of a conflict can be tricky to spot because some push and pull between a president and his or her agencies is normal.

In 2009, for instance, military officials used leaks to pressure the White House over what it saw as the minimal number of troops necessary to send to Afghanistan.

Leaks can also be an emergency brake on policies that officials believe could be ill-advised or unlawful, such as George W. Bush-era programs on warrantless wiretapping and the Abu Ghraib detention facility in Iraq.

"You want these people to be fighting like cats and dogs over what the best policy is, airing their views, making their case and then, when it's over, accepting the decision and implementing it," said Elizabeth N. Saunders, a George Washington University political scientist. "That's the way it's supposed to work."

"Leaking is not new," she said, "but this level of leaking is pretty unprecedented."

Institutional conflicts under Mr. Trump, she worried, had grown into something larger and more concerning.

Officials, deprived of the usual levers for shaping policies that are supposed to be their purview, are left with little other than leaking. And the frenetic pace of Mr. Trump's executive orders, which the agencies would normally review internally over weeks or months, has them pulling that lever repeatedly.

They have leaked draft executive orders, inciting backlashes that led the orders to be shelved. And they have revealed administration efforts to circumvent usual policymaking channels, undermining Mr. Trump's ability to enact his agenda.

Prompt 3: Partisan (Definition + Trump Jr. Retweet)

Donald Trump Jr.


If there was ever confirmation that the Deep State is real, illegal & endangers national security, it's this. Their interests above all else.


Admin Hit With AT LEAST One National Security Leak a Day…

5:18 PM 7/7/17 Twitter for iPhone

Appendix B: Survey Questions of Interest

"As it stands right now, do you think that unelected or appointed officials in the federal government have too much influence in determining federal policy or is there the right balance of influence between elected and unelected officials?"

  1. a) unelected or appointed officials have too much influence
  2. b) right balance of influence between elected and unelected officials

"How much latitude should new presidents of one party have in replacing government officials who took their positions under the administration of an opposing party?"

  1. a) none at all
  2. b) a little
  3. c) a moderate amount
  4. d) a lot
  5. e) a great deal

"Do you approve or disapprove of the job these federal appointees (Secretary of State Mike Pompeo, Attorney General William Barr, Central Intelligence Agency (CIA) Director Gina Haspel, Secretary of Defense Mark Esper) are doing?"

  1. a) disapprove
  2. b) neither approve nor disapprove
  3. c) approve

"To what extent do you find these federal departments and agencies (Department of State, Department of Justice, Central Intelligence Agency (CIA), Department of Defense) trustworthy?"

  1. a) untrustworthy
  2. b) neither trustworthy nor untrustworthy
  3. c) trustworthy

"What do you believe is a higher priority for these federal departments and agencies (Department of State, Department of Justice, Central Intelligence Agency (CIA), Department of Defense): preserving their own power and status, or helping serve the American people?"

  1. a) preserving their own power and status
  2. b) helping serve the American people

"Do you believe a deep state in the federal government definitely exists, probably exists, probably doesn't exist, or definitely doesn't exist?"

  1. a) definitely doesn't exist
  2. b) probably doesn't exist
  3. c) probably exists
  4. d) definitely exists

"Before taking this survey, to what extent were you familiar with the term 'deep state' as it applies to the federal government?"

  1. a) not familiar
  2. b) somewhat familiar
  3. c) very familiar


* Email:

[1] As evidence of this, Google Trends data reveal that between January 1, 2008 and February 29, 2020, the 10 months with the highest levels of searches for the phrase "deep state" have all occurred during Donald Trump's time as president.

[2] Amazon Mechanical Turk (or mTurk for short) has been a popular resource for public opinion research by academics for over a decade. With this popularity has come great scrutiny as to best practices for conducting mTurk surveys (see Berinsky, Huber, & Lenz, 2012; Buhrmester, Kwang, & Gosling, 2011; Huff & Tingley, 2015; Levay, Freese, & Druckman, 2016; Mullinix, et al., 2015; Paolacci & Chandler, 2014; Paolacci, Chandler, & Ipeirotis, 2010). Much of this literature has praised mTurk for providing high quality respondents, while at the same time suggesting researchers be conscious of how mTurk participants differ from the public at large. I heed such advice in the construction of my survey and in the analysis of my findings, controlling for many of these aforementioned differences.

[3] Quotas were used to ensure both Democrats and Independents were not drastically oversampled; previous mTurk surveys have revealed respondent pools that leaned more Democratic than the population at large.

[4] The text of all treatments used in this experiment can be found in Appendix A.

[5] Specific questions used in the analysis portion of this manuscript can be found in Appendix B. Questions pertaining to unelected/appointed officials' influence and respondent belief in a deep state were drawn from a March 2018 survey (Monmouth, 2018).

[6] Demographic and political questions are available from the author upon request.

[7] Multivariate modeling not presented here reveals that the only variable that significantly predicts how one feels about replacing government officials appointed by another party's administration is approval of President Trump.

[8] The inclusion of interactions between age and receiving the treatments revealed no instances in which younger respondents reacted differently than older respondents to the Deep State information. The inclusion of interactions between education and receiving the treatments revealed a single instance (out of 36 possible) where the highly educated reacted differently than the poorly educated to the Deep State information; in this lone instance, the results presented in this paper are not affected in terms of significance or lack thereof by the inclusion of this interaction.

[9] As stated earlier, only respondents in the 3 treatment groups were asked general questions about the deep state concept.

ISSN: ISSN 2398-5836

Copyright (c) 2021 Tyler Johnson