Safeguarding the Peaceful Transfer of Power: Pro-Democracy Electoral Frames and Journalist Coverage of Election Deniers During the 2022 U.S. Midterm Elections

Abstract

This study develops a new normative and analytical framework of “democracy-framed electoral coverage” grounded in literatures that stress the role of governmental and communicative institutions in protecting democracies from threats. We define “democracy-framed electoral coverage” as journalism that embraces fairly contested elections as an established norm and political ideal, and clearly alerts the public to threats to the peaceful transfer of power. Using the U.S. as a case study of a consolidated democracy facing autocratization threats, the study applies the framework to analyze a comprehensive dataset of 708 articles encompassing twenty one races during the 2022 U.S. midterm elections when candidates who denied the results of the 2020 presidential election were on the ballot. Additionally, we conducted interviews with twelve journalists to evaluate their perspectives and practices regarding election denying candidates. We find that journalists routinely failed to alert the public to the threat posed by candidates unwilling to embrace the legitimacy of U.S. elections. This paper demonstrates the necessity of a normative framework for pro-democracy election coverage, and the findings underscore the electoral fragility of the U.S.—a case of a democracy undergoing autocratization processes and facing threats to the legitimacy of its elections and the peaceful transfer of power.

According to the U.S.-based The Washington Post (Blanco et al. 2022), during the 2022 midterm elections in the U.S., 291 Republicans ran for office at the state and federal level while denying, contesting, or otherwise questioning the results of the 2020 presidential election. 179 of these candidates won their races.

This is a fundamental threat to democracy. If a country cannot have legitimate competitive elections despite all available evidence of their safety and security—if a political leader or party cannot lose—then there can be no peaceful transfer of power. The logical conclusion of this played out on Jan. 6, 2021 during the attempted coup (Cline Center 2022) at the U.S. Capitol led by former president Trump and his supporters. There was also the subsequent failure of Republican Party leaders to address, or in many cases even acknowledge, the danger of extra-judicial attempts to hold onto power. And, January 6th was only the most visible sign of the Republican Party leading the U.S. down a path of democratic backsliding, including through state-level efforts designed to limit electoral competition (Grumbach 2022).

The United States offers only one case of recent threats to competitive elections, albeit a particularly important and dramatic one. Over the past decade, democracy has been under threat globally. In 2020, autocratic tendencies were observed in twenty five countries, impacting 34 percent of the world’s population; in contrast, the sixteen countries moving toward democracy represent just 4 percent of people worldwide (Hellmeier et al. 2021). On Jan. 8, 2023 in Brasilia, Brazil, supporters of outgoing president Jair Bolsonaro attacked the country’s Supreme Court, National Congress, and Presidential Palace in the hope of overthrowing a newly seated, democratically elected president. Attempts to undermine the peaceful transfer of power are not always so violent. In countries such as Hungary (Bogaards 2018) and Turkey (Çalışkan 2018), leaders have consolidated power in part by limiting electoral competition through judicial and other means, including through measures cynically stated to protect the integrity of elections.

While historically the decline of democracy has been marked by sudden collapses in countries with low to moderate economic development, recent patterns indicate that backsliding is happening gradually and impacting a broader range of nations, including those believed to possess strong democratic institutions (Diamond 2021; Kaufman and Haggard 2019). For example, recently South Korea’s Yoon Administration has been showing autocratic tendencies (E. T. Kim 2023; M. Kim 2023), including preventing news media from fulfilling their watchdog roles, filing criminal defamation complaints against several news organizations, and dismantling the nation’s only fact-checking organization, the SNU FactCheck Center, a coalition of thirty two major news organizations in the country (Nyariki 2023)—all in advance of the 2024 general election.

While we now have a robust body of work on what is variously called “democratic backsliding” (Haggard and Kaufman 2021) or “democratic decay” (Daly 2019), we lack robust normative theory of journalism’s role in protecting democratic institutions, specifically in the context of elections. And, we lack an analytical framework and empirical evidence for assessing how well journalists respond to threats to elections—whether they entail systematic attempts by political leaders to undermine public trust in elections, attacks on election administrators and the legitimacy of the vote, refusals to concede, efforts to limit voting by one’s opponents, or violent attempts to overthrow a government after election loss. All of these actions are designed to undermine the accountability of political leaders, and their political parties, at the ballot box and the peaceful and legitimate transfer of power.

This article brings together various threads of literature on journalism’s role in democracy (e.g., Altay et al. 2023) to develop a new normative and analytical framework of “democracy-framed electoral coverage,” and provides an empirical test of journalism’s performance of this role through the case of the 2022 U.S. midterm elections. Specifically, we offer a normative and analytical framework of democracy-framed electoral coverage to delineate when journalism is in the service of protecting electoral institutions. Developed in collaboration with Shannon McGregor and Erik Peterson, we define democracy-framed electoral coverage as that which foregrounds fairly contested elections as both an established norm and a political ideal. This frame of coverage goes beyond pointing out that claims of widespread voter fraud are false and not substantiated (if, indeed, there is no evidence that irregularities occurred) —it also positions election denial as a violation of democratic norms with deleterious implications for democracy. It treats election denial—or ex ante assertions that a candidate will not accept the result of an upcoming election—as fundamentally different from other campaign issues. We then operationalize and apply this definition to a sample of 708 articles about twenty one races around the U.S. with an election denier on the ballot. To gain insights into our findings, we also interviewed twelve journalists at different city/local, regional, and national outlets to discern their thoughts about this type of coverage.

Our normative and analytical framework of “democracy-framed electoral coverage” effectively assesses journalistic performance when faced with threats to democracy. Put succinctly, our results indicate that U.S. journalism clearly fell short of protecting democratic institutions. Our findings reveal the limited presence of democracy frames and a prevalence of weak statements that failed to condemn, correct, or call out election deniers. News articles clearly trailed opinion pieces in democracy-framed election coverage—indicating that journalists failed to take a clear and assertive stance against denialism as part of their routine coverage from the campaign trail—and journalists had limited use of election administrators as sources, particularly troubling given their nonpartisan role of being tasked with election security and vote fairness. And, we found no significant variation across races in terms of democracy-framed election coverage. It did not matter whether journalists were covering competitive or non-competitive races or working in districts with large numbers of Republicans or Trump supporters or Democratic-safe districts—media coverage across the board failed to protect against threats to democracy.

These findings underscore the urgency of addressing growing threats to democratic institutions, especially elections. They matter both in the context of this specific case and more broadly in offering a clear approach to assessing whether journalists and media fulfill roles of protecting democracy in elections globally (e.g., Humprecht et al. 2020).

Literature Review

There are very well-established bodies of literature about journalism’s roles in democratic societies. For example, there is a body of normative journalism literature, which in turn complements empirical studies of journalists’ own ideal roles as well as those posited in legal theories (Ananny 2018; Nielsen 2017; for a review and critique, see McDevitt 2022). To summarize, these roles include providing accurate information, holding powerful figures and institutions accountable, facilitating public debate and deliberation, encouraging participation in democratic life, seeking truth or trustworthy information, and representing the public and public opinion to those in power.

Despite this body of work, there is comparatively less work, and less consensus, on journalism during complicated democratization processes—which can last decades and which might feature periods of regression—or threats to deconsolidation after democracy is established (McDevitt et al. 2022; Sobel Cohen and McIntyre 2023; Voltmer 2019; Wasserman 2020). As scholars have suggested, democracy is not a steady-state condition, but a continually performed achievement (Cianetti et al. 2020; Sobel Cohen and McIntyre 2023). Presumptions of linear decline (or progress) often elide far more complicated political realities (Cianetti et al. 2020). In many veins of the literature on democratic backsliding, scholars focus on the role of political institutions (such as parties) and laws. For example, a central finding in this literature is that we have moved from outright coups to slower processes of constitutional erosion brought about by the efforts of anti-democratic leaders and parties (Cianetti et al. 2020; Jee et al. 2022; McDevitt 2022). The key actors in this drama are often elected leaders and parties in power that, over time, undermine competitive elections (such as passing laws to favor one political party), norms such as respecting a legitimate opposition (such as through “enemies of the people” language against opponents), and the rule of law (such as through the appointment of people with partisan interests to justice offices and the politicization of the judiciary).

Despite its considerable strengths, much of this comparative democracy literature has rather thin conceptualizations of “media” (which, in turn, inform the conceptions of political systems used by media scholars). When media are considered, they are often evaluated in structural terms (such as whether media are state-owned or unduly state influenced) and used as measures of civil liberties or independent checks on ruling parties—not as a potential source of backsliding or decay. For example, deeply influential literature presumes that freedom of expression and pluralist sources of information are cornerstones of democracy, such as the influential Varieties of Democracy framework and its V-DEM index which is the most significant typology of global varieties of democracy (Coppedge et al. 2020). Or, these indices rely on generalized expert assessments of press performance (see Supplemental Information file, Supplement A, for details).

And yet, it is exactly these expressive and media freedoms, presumed to be democratic in one literature, that are of growing concern to media, political communication, and journalism scholars. Free expression, free media, and alternative sources of information such as on social media facilitate mis- and disinformation and conspiracy theories that undermine democratic deliberation and public facts (Lukito et al. 2023; Persily and Tucker 2020), and enable the widespread circulation of false claims of election fraud (Chen et al. 2023). Non-state, commercial media can engage in coordinated campaigns that advance the interests of anti-democratic parties when they are in power, or strive to gain it back (Benkler et al. 2018; White 2018; Peck 2019; Williamson et al. 2011). Or, media outlets can engage in tireless baseless criticism of democratically elected governments or electoral processes in ways that undermine the legitimacy of those in power (Henrickson and Betz, 2023). Even a legacy press that is independent and commercial and adheres to professional journalism standards might fail to protect, or even actively undermine, electoral institutions given norms such as objectivity and balance that work against alerting the public to democratic threats (Adams 2020; Carlson et al. 2021), to say nothing of the vast economic incentives that shape commercial journalistic coverage toward grabbing and holding the public’s fleeting attention (Munger 2020).

In other words, freedom of expression and free media can undermine democracy itself. As Haggard and Kaufman (2021) argue, authoritarian leaders can “exploit deep liberal commitments to free speech” to manipulate and control media, intimidate and discredit opponents, undermine truth, create confusion and mask weakening institutional safeguards, and narrowly target vulnerable citizens.

As such, we need both a more clearly specified analytical framework for journalism’s role in protecting electoral institutions and an empirical way of evaluating press performance. Journalists, and journalism as an institution more broadly, potentially play significant roles in defending democratic norms and institutions, especially around elections, or weakening them. In comparatively consolidated democracies, threats to democracy might come from the rise of an anti-democratic party or political leader—and as such journalists have to put aside concerns over social solidarity or polarization in order to clearly alert the public to the democratic threat. This is required of a communicative institution that serves as a “gatekeeper” of democratic, and specifically electoral, institutions (Adams 2020; Alexander 2006; Levitsky and Ziblatt 2019). Adams (2020) argued that ideally journalism institutions should serve as “civil gatekeepers” and alert the polity to threats to democratic norms and institutions—but also that journalism’s fulfillment of this role must be analyzed empirically given competing pressures and norms (such as objectivity and balance).

In this study, we are focused more narrowly on a minimalist definition of “procedural democracy” (Strömbäck 2005) (what the democratization literature calls “liberal democracy”) in coupling elections with secured rights. We define “democracy-framed election coverage” as that which foregrounds competitive elections and the peaceful transfer of power as both an established norm and political ideal. In our normative view, journalism has a responsibility as a communicative institution of democracy to help secure elections, and especially to clearly and repeatedly alert the public to false claims of voter fraud, strategic campaigns to undermine public confidence in elections, the failures of candidates and parties to embrace election results, and other attempts to subvert legitimate processes for holding, or transferring, power. These are all dangerous violations of democratic norms resulting in threats to procedural democracy. As such, democracy-framed coverage requires journalists to interpret intentions, especially in the context of strategic attempts to manipulate public perceptions for electoral or power advantage (What exactly is democracy-framed election coverage can only be understood in particular national political contexts at distinct moments in time. See Supplemental Information file, Supplement B for further discussion.)

We view our approach sitting alongside and supplementing indices such as V-DEM in offering a research framework for conducting content analytic empirical tests of press performance during elections. Specifically, the role that journalists potentially play in staving off electoral threats, or failing to, is an empirical question that has to be evaluated in the context of actual press performance during the course of an election.

The U.S. Case of Threats to the Security of Elections and Peaceful Transfer of Power

We use the U.S. 2022 midterm elections as our case of democracy-framed electoral coverage. The U.S. provides a case of a consolidated democracy experiencing threats to its electoral system and fair electoral competitiveness. It has reasonably well-established electoral institutions since the enforcement of universal suffrage in the 1960s, which ended legally and institutionally constrained citizenship for Black people (however, there is the persistence of racialized categories of citizens without voting rights such as felons [Alexander 2011]). V-DEM classifies the U.S. as a democracy since 1921 after women gained the right to vote, and a liberal democracy (premised on securing individual rights) after 1969. The U.S. has faced considerable threats to free and fair elections in recent years—including systematic efforts on the political right to undermine public confidence in elections (Clayton et al. 2021), Republican partisan efforts to legally limit the participation of political opponents through electoral rules (such as around voter identification) (Grumbach 2023), and the express attempt by a sitting Republican president to prevent the peaceful transfer of power through an attempted auto-coup on Jan. 6, 2021 (Cline Center 2023). These dynamics in turn are fueled by white, and white Christian, backlash to political, social, and economic gains made by non-white groups (Gorski and Perry 2022).

The U.S. is also a particularly important case given the country’s support for global democracies (Lieberman et al. 2019), as are European countries which have undergone their own right-wing strains on democracy (e.g., Pintsch et al. 2022). The U.S. as a case looks like a set of other countries experiencing democratic threats, especially in the context of elections. The U.S.’s comparative length of time as a democracy, and its high rate of economic development, mitigate the likelihood of its democracy being downgraded (Treisman 2023). That said, there are clearly alarming threats to U.S. electoral and liberal democracy, and other studies argue that the country has been autocraticizing since 2016 (Boese et al. 2021). These assessments do a good job positioning the U.S. as a case on two levels. First, as a case of an autocratizing country broadly in the context of the erosion of the rule of law and the undermining of the peaceful transfer of power. Second, as a case of a comparatively consolidated liberal democracy undergoing autocratizing processes. (see Supplemental Information file, Supplement C for more information about the country’s autocratizing trend).

As such, analyzing how journalists covered 2022 candidates who denied the results of the election in 2020 provides an opportunity to evaluate journalism’s role vis-à-vis electoral institutions. The Electoral Integrity Project noted specifically that the 2022 midterm elections were an improvement on 2020 in terms of candidates and parties challenging legitimate results, even as the U.S. remained “a democracy struggling to maintain equilibrium” when it came to electoral management and rule of law (Garnett et al. 2023: 9). And, while the U.S. is most directly comparable to countries undergoing autocraticizing strains (as detailed above), press coverage of electoral threats (to legitimate election outcomes, election administration, and the peaceful transfer of power) matters to democracies of all stripes—especially given that autocrats and would-be autocrats can arise even in comparatively established democracies (Guasti 2020).

Hypotheses and Research Questions

The empirical research question of this study is: does U.S. election coverage include pro-democracy electoral frames when journalists cover candidates who have denied the results of free and fair elections? To answer it, we conducted a content analysis of media coverage of 2022 candidates who denied the results of the 2020 elections, including both news articles and opinion pieces. Based on previous literature, we had three hypotheses about democracy electoral frames in press coverage.

First, we anticipated that the utilization of pro-democracy electoral frames in news articles would be limited. Professional norms such as objectivity and balance likely hinder journalists from effectively alerting the public to democratic threats (Carlson et al. 2021; Adams, 2020 Hughes et al. 2023; Hughes et al. 2023). In addition, commercial incentives that limit professional autonomy might lead journalists to avoid alienating parts of their audience as well as focus their time and energy on attention-grabbing aspects of the campaign trail (Pickard 2018).

Second, we expected that the number of weak statements about election denialism will exceed strong statements in articles. Weak statements—such as mentioning a candidate’s denialism without refuting it—allow journalists to acknowledge the existence of denialism without actively challenging or refuting it, thus protecting objectivity norms. Even more, journalists may feel obligated to present false election claims simply as a viewpoint or opinion held by a segment of the population (Egelhofer et al. 2020, 2022; McDevitt et al. 2022). Lastly, strong statements—such as stating the truth of the outcome or analyzing actions of election denial as a political tactic designed to unfairly gain power outside of the electoral process—typically require journalists to invest scarce space, time, and effort into researching and providing evidence to contextualize denialist claims (Braun and Eklund 2019). While strong statements do not rise to the level of democracy-framed election coverage (which has the threat of denialism to established democratic procedures as the central theme of articles), they are nevertheless still important in alerting the public to unfair attempts to gain or hold onto power.

Third, we believe that news articles will be less likely to use pro-democracy electoral frames compared to opinion pieces. Journalists writing news articles may be hesitant to take a strong stance against denialist claims due to accusations of bias or partisanship (e.g., Pajnik and Hrženjak 2022). As such, we anticipated that writers of opinion pieces would find it comparatively easier to take a clear and assertive stance against denialist claims given opinion pieces are not bound by the same requirements of journalistic news norms, which may allow authors to actively challenge and refute false claims.

In addition to the hypotheses above, we examined whether election administrators were included as sources given their civic role and institutional role serving democratic values of impartiality, accuracy, and accountability, potentially enhancing public trust in the electoral process (e.g., Udani et al. 2018).

Finally, we analyzed a few factors we believe might have an impact on the coverage of election denying candidates. We tested whether the partisan composition of the district (majority Republican or Democrat or split based on the Cook Political Report 2022) brings about noticeable variations in news coverage (given potential fears of backlash or journalistic concerns about the trust of their audiences). For the same reasons, we tested whether differences in Trump’s vote share during the 2020 election in the relevant district, specifically below 50 percent versus over 50 percent, would lead to differences in coverage. And, we asked whether the competitiveness of the race, ranging from closer margins (48–52%) to solid wins above 55 percent, would result in meaningful differences in the news coverage of election denying candidates (with the idea that a closer race might potentially mean fewer strong statements about election denialism given journalists’ concerns about alienating some voters). Trump’s vote share and competitiveness of the race were both based on NBC News’ webpage for 2022 state election results (NBC News 2022).

Methods

Sampling of the Races

To determine our sample, we used The Washington Post’s tracking of 2020 election deniers (Blanco et al. 2022) as a reference. Initially, we selected twenty eight races via purposive sampling, considering their regions/districts, varying partisan compositions, levels of support for former President Trump in the 2020 elections, and competitiveness. To ensure thorough analysis, we assessed the availability of local news coverage for the twenty eight races using the ProQuest database, subsequently excluding from the final sample seven races that lacked any local news coverage in the database. As a result, the final sample included twenty one races. Refer to Supplemental Information file, Supplements D and E for the details on the sampling process for the races.

Sampling of News Articles

For our dataset, we chose those outlets that would likely be the highest in professionalism and autonomy compared with the broader U.S. media sector (Supplemental Information file, Supplement F includes information about the broader U.S. media sector). Despite discussion of the demise of the liberal model of the press, general interest legacy news outlets still reach comparatively large audiences with general interest news products produced according to professional standards of reporting in the U.S. and many democracies (Toff et al. 2023). This includes both national and local news outets producing routine reporting on-the-ground during elections, alongside their opinion pieces. These outlets also, in turn, often inform the reporting of cable news and other outlets in a media system such as the U.S. As such, we focus here on the professional, liberal model press in the U.S. We are less interested in the extensive U.S. right-oriented press because the role of Fox News in promoting election denialism has been well-documented, including by the courts (Barbas et al. 2023), and the broader right-wing media ecosystem’s disinformation is extensively studied (e.g., Benkler et al. 2018; Freelon et al. 2020).

The data collection method for the content analysis involved a systematic process of searching, downloading, and cleaning local news articles from the Proquest database. Search strings included a list of news outlets covering the state the race took place in and the name of candidates anywhere in the text. For the national news coverage, we collected data from three national news outlets: The New York Times, The Washington Post, and USA Today. These news outlets are all included in the top five largest daily newspapers in the U.S. by average print circulation as of March 2023 (NYT: 29,633,000; WP: 13,923,000; USA Today: 13,264,000) (Statista 2023). For the local news coverage, we built the list of news outlets for each state using the News Deserts Database—a proprietary database owned by the UNC Hussman School of Journalism and Media, the University of North Carolina at Chapel Hill. ¹

Using each search string, we were able to download a total of 2,788 news articles, which included false positives and duplicates. One of the researchers and a research assistant then examined the articles to exclude those that were not written in English, not published by one of the news outlets we included in the search strings, and those that were only lists of candidates’ names endorsed by the news outlet.

The three national news outlets did not yield much coverage on the races in our sample, so we coded the full data (see Supplemental Information file, Supplement G). On the other hand, the local news coverage data sometimes required a different approach. For the attorney general, House congressional, and secretary of state races, we coded all articles. For the governor and Senate races, we did a stratified random sample for each race because they included 1,778 and 654 articles, respectively. Fifty random articles (a representative sample), or the total number of articles (whichever was smaller), were included in the sample for every governor and Senate race.

Our novel conceptual framework necessitated an in-depth understanding of the nuances and contexts of these articles. Therefore, recognizing the value of deep, iterative training discussions about discrepancies among coders with subject matter expertise, we opted for end-to-end manual content analysis over automated or computer-assisted methods. Coders were also asked to identify duplicate and irrelevant articles, thereby reducing the dataset of local news coverage to 589 news articles and 119 opinion pieces (see Supplemental Information file, Supplement G).

Codebook and Variables

Beyond the baseline details of the articles (headline, date, author, races), we structured the codebook into two sections. The first section contained variables pertaining to the news story as a whole, while the second focused on variables specific to individual races covered within the news story.

The first section included pro-democracy electoral frames. Pro-democracy electoral frames are a powerful tool for media to protect democracy. For example, in framing election denial as a threat to democratic norms and institutions, journalists can convey the importance of accepting election results, respecting the will of the voters, and ensuring the peaceful transition of power. Pro-democracy electoral frames highlight the necessity of fairness, the rule of law, and the necessity of a party being able to lose. Through pro-democracy electoral frames, journalists could also provide a counter-narrative to disinformation by clearly refuting false claims of voter fraud and conveying the implications of election denial for democratic governance. Thus, we coded whether a pro-democracy electoral frame was the central theme in the narrative and operationalized this concept as a binary variable: either the narrative was primarily structured around the threat election denial poses for democracy, or it was not.

Election Administrators as Sources

For each news article, we measured, as a binary variable, whether election administrators were included as sources.

The second section included variables coded for each race mentioned in the article (up to three races). First, type of race was coded to identify the race among the twenty one in our sample.

Strong and Weak Statements About Election Denialism

As detailed above, for strong statements we were looking for journalistic assertions of truth, which means correcting or asserting that the claim is false (including through the use of sources, including opposition party members). We coded all strong statements, including those paired with false or weak statements in the same article. Weak statements encompass mentions of election denialism without any correction. Both variables were coded using four nominal categories: “Yes, headline,” “Yes, body,” “Yes both headline and body,” and “not at all.” Due to the scarcity of both strong and weak statements in the article, we decided to recode these variables into a binary format.

Reliability

Reliable data can be operationally defined in two ways: (a) as data that remain constant throughout variations in the measuring process (Kaplan and Goldsen 1965, 83–4) and (b) as data that can be replicated elsewhere, demonstrating that coders concur on the readings and interpretations of the given text (Krippendorff 2018). To ensure reliable data for our content analysis, we developed a codebook with communicable and practical instructions, had coders go through a training program which in turn fed back into the codebook and instructions, and produced reliability data independent from the actual study.

We trained twelve undergraduate coders on a training dataset, which included ten articles separate from the actual content analysis study. ² To assess the agreement among coders, we utilized the Krippendorff’s alpha coefficient, a widely used measure of intercoder reliability. The results indicated a high level of agreement among coders, with an average Krippendorff’s alpha coefficient of 0.837, with all variables showing more than 90 percent agreement. An Alpha coefficient of 0.837 falls within “a good agreement” range (Krippendorff 2004). This demonstrates a high level of consistency among the coders in applying the coding scheme to the articles.

Regular meetings with the coders took place to address any questions, clarify ambiguities, and discuss challenging cases. We also implemented consistency checks throughout the coding process. Coders received a random subset of articles in their assigned batches that was assigned to another coder as well. We compared the coding results among the coders to assess intercoder reliability and identify areas of disagreement. With the multiple pairs of coders, the average Cohen’s Kappa coefficient was 0.861 across variables, with all variables showing more than 90 percent agreement.

Qualitative Data

We were able to conduct interviews with twelve journalists from a range of city/local (3), state (4), and national (5) media outlets (see Supplemental Information file, Supplement H for outreach efforts). Of those, nine journalists were from newspaper or media sites, and three were from broadcast outlets (one each of city, state, and national outlets). Two were opinion commentators (one national and one city outlet.) All interviews were conducted “on-the-record” with journalists able to declare any statement “off-the-record,” “on background,” or “not for attribution” at their discretion. However, because of the few journalists willing to speak at all, and conditions requested by a number of the journalists we secured interviews with, we made the decision to report all of this data not-for-attribution.

Trained teams of two undergraduate researchers conducted the interviews following a collectively developed interview map (see Supplemental Information file, Supplement I). We performed two rounds of coding. For the first round, the entire research team inductively coded either five or six transcripts, but in a way that was sensitive to the larger themes of the study and the analytically guided interview questions. We then had a research meeting to discuss the themes and codes (n = 115) we developed, as well as pull exemplar quotes. For the second round, we used more deductive coding (based on the themes and codes from the first round) while also doing another round of inductive coding. What resulted was eight themes and 44 subthemes, which we combined to create the narrative of findings below.

We present the qualitative data in the discussion section here because they should be understood as a suggestive complement to the content analysis data. Perhaps not surprisingly, all the journalists we spoke to stated their outlet did coverage related to election deniers well, which suggests to us social desirability bias in both the sample and results. At best, these interviews are illustrative of what journalists perceive are best practices for how to cover election deniers, as well as the perceived tensions that they faced over things such as conveying democratic threats and maintaining the trust of Republican audiences.

Results

The Utilization of Pro-Democracy Electoral Frames in News Articles was Limited

For our first hypothesis, we examined the percentage of news articles that used pro-democracy electoral frames. A simple descriptive statistic showed that only six percent (36 out of 589 articles) of the news articles included pro-democracy electoral frames in their election coverage. None of the national news coverage used pro-democracy electoral frames; all thirty six news articles that used pro-democracy electoral frames were from city/local news outlets.

The Number of Weak Statements Exceeded That of Strong Statements in the News Articles

We tested our second hypothesis with two different units of analyses. First, we tested if the number of weak statements in news articles exceeded that of strong statements per race. Here, each race was considered as the unit of analysis, and an article could be coded up to three times for different races. A two-sample Z-test showed that the proportions of the strong and weak statements were significantly different. The test statistic (z = −2.02) showed a p value of .044, providing evidence of the significant difference. The effect size was found to be small (h = 0.11), indicating a minimal magnitude of difference between the proportions. Thus, although our results indicated a statistically significant higher prevalence of weak statements (12.03%) compared to strong statements (8.74%), the disparity observed was relatively small compared to the whole sample. In other words, both strong and weak statements had a low presence in our sample of news articles.

Next, we tested the same hypothesis but with the news articles as the unit of analysis. Because we coded up to three mentioned races for each news article, if any of the races included in the piece included a strong statement, it was considered that the news article had a strong statement. The percentage of news articles that included at least one strong statement was 9.51 percent (56 out of 589), while for weak statements the percentage was slightly higher with 13.58 percent (80 out of 589). The difference between these two percentages were statistically significant with a z score of −2.188 and a p value less than the .05 threshold (p = .03).

More Opinion Pieces Than News Articles had Pro-Democracy Electoral Frames and Strong Statements

We had two hypotheses related to the comparison between the percentage of pro-democracy electoral frames in news coverage versus that in opinion pieces. First, the percentage of pro-democracy electoral frames was higher for the opinion pieces at 19.33 percent (23 out of 119), compared to the six percent observed in news articles. A two-sample Z test showed that with a value of z = 4.72, and with a p value less than .0001, there was a significant difference between the two samples.

Next, we also compared the percentage of strong statements within news articles with those of opinion pieces. As we expected, 15.13 percent (18 out of 119) of opinion pieces had at least one strong statement, higher than the news articles (9.51%). The difference, however, lacked power to be statistically significant with a p value of p = .067 (z statistic = 1.83).

Few Articles Included Election Administrators as Sources

We considered the use of election administrators as sources regardless of whether they directly addressed election denialism or democracy. Out of the total sample of 589 news articles, only thirty five articles (5.94%) had election administrators as a source.

District Differences Did Not Matter for the Presence of Pro-Democracy Electoral Frames

As detailed above, we also considered a set of other variables we believed to be potentially relevant to democracy-framed electoral coverage: the partisan composition of district, Trump’s vote share in 2020, and the competitiveness of the race. We counted the occurrences of each variable by race (because up to three races were coded for each article, the number of the races is bigger than the number of the articles coded.) Independent T-tests indicated there were no differences in how election deniers on the ballot were covered across races that varied along these dimensions. Nonetheless, we found some interesting correlations among the key variables: see Supplemental Information file, Supplement J.

Discussion

The aim of this study was to develop a normative and analytical framework of pro-democracy-framed electoral coverage that could be applied globally in assessments of the press’s ability to protect against democratic threats, specifically in the context of elections given their centrality to securing all other freedoms (Dahl 2008), including expression and assembly. We believe that pro-democracy election framed coverage offers a minimalist, yet clear, conceptualization of journalism’s normative role in democratic states, and provides an analytical and empirical framework for evaluating press performance in the context of threats to the peaceful transfer of power.

In our application of this framework to the case of the 2022 U.S. midterm elections, which took place in the context of a country undergoing autocratization pressures, our findings reveal that journalism clearly fell short. Given the urgency to confront mounting threats to electoral institutions in countries around the world, one question we had throughout this project is why—why do we see so little coverage of anti-democratic threats to elections? Turning to our interview data, we now suggest what journalists perceive to be behind successful, or unsuccessful, pro-democracy-framed election coverage.

Editorial Processes

Five of the journalists in our interviews cited having an express editorial process around, or at the very least a set of explicit conversations about, how to cover election deniers on the ballot in 2022. These journalists were from different outlets and did not include opinion writers (who all stated they were on their own with how they approached election deniers). The editorial processes described by these journalists ranged from having editorial team meetings devoted to fleshing out how to cover election deniers to having individual editors guiding coverage for journalists. One journalist described their outlet having “boilerplate” (personal communication) text (a strong statement) that was inserted into every story the newspaper was producing about election deniers on the ballot. Journalists who discussed these types of interactions with editors all cited in various ways that this process was there to guide individual journalists covering elections and standardize the approaches of the outlet across races.

The remainder of journalists cited being more on their own when it came to how to handle coverage of election deniers. Some of these journalists reported having conversations with editors, but they were less structured and practices seemed to be more emergent and contingent on the evolution of these races. In this subset of responses, decision-making was less deliberative or intentional. As one participant (personal communication) reported, “[journalists and editors] discussed it [election deniers] privately, but not in a robust way as you would expect on such an important topic.” Another (personal communication), stated that “I don’t remember specific conversations. I think everyone was just making it up as they went, because it was so unprecedented.”

Journalistic Orientations

The news journalists, almost universally, grounded their professional understanding of their roles on epistemological grounds. Namely, they saw themselves and their institution in terms of creating and conveying public facts. Although they centered democracy as a concern, and resoundingly espoused the idea that democracy was important and under threat, they squarely grounded their own role in information provision terms. In other words, none of these journalists mentioned seeing their roles in terms of making an “affirmative” case for democracy, such as educating the public about the importance of the peaceful transfer of power or legitimate elected rule. There was an implied sense that the audience would share these thoughts with journalists, but it was implied.

Nearly all of the news journalists we spoke with framed their, and their outlet’s, orientation in terms of being “truthful,” “accurate,” and “objective.” We take these to be journalists’ perceived ideal orientations—they relayed their commitment to the facts, including asserting those facts when it came to candidates that threatened democratic institutions and processes. At the same time, however, when asked about their practices we received a more nuanced picture that threatened these ideals and might help explain the low prevalence of democracy frames and strong statements in the sample. Even while espousing ideals such as a commitment to facts, journalists also discussed wanting to be “balanced,” and how difficult that was when election deniers were on the ballot.

For example, journalists described adopting reporting practices to navigate tensions between being truthful about threats to democracy while striving to be balanced (i.e., nonpartisan) in their coverage. One journalist cited avoiding “partisan” words and phrases about democracy, while also still conveying democratic threats. For example, this journalist (personal communication) cited how since “democracy”-related themes were a Democratic [Party] talking point, they deliberately avoided using similar language while still working to convey a similar message (i.e., that elections are safe and secure). Another journalist (personal communication) cited not using the label of “election denier” which, in their view, was unnecessarily pejorative and political as a label. Better, in this person’s view, to describe empirically what the candidate did and said in the context of the actual safety and security of elections (an example of adopting strong statements). Another news journalist stated that being “balanced” meant not continually repeating the threat to democracy that some candidates represented. Another cited that they made the decision to not continually remind their audiences of election denial and instead focused on other issues. Still another suggested that they highlighted concerns about denialism, but balanced it against other, more positive, things about the candidate.

Concerns Over Trust

All of the journalists in the sample touched on another important concern in various ways: their relationship with, and maintaining the trust of, their audiences. A majority of the news journalists we spoke with cited maintaining, losing, or regaining the trust of audiences as a main concern, and thought about their coverage through this lens. Journalists referred to “polarization” being a significant concern in various ways and having to navigate entrenched partisans interested in only believing their side. As one journalist (personal communication) put the challenge: “And so I think, you know, as much as partisan media has a place, I think our role is to continue to be like as objective as possible while documenting like the objectively insane things that are happening.”

One way four of our interviewees resolved this tension was to clearly delineate between “reasonable” and “unreasonable” audience members. One journalist (personal communication) explicitly cited that they would not be able to gain, or maintain, the trust of “hardcore election denialists,” but they did think that there was a broader “middle” that was open to factual appeals, including corrections to election denialism. It was this group—the reasonable, sensible, and open-to-being-informed audience—that these four journalists cited imagining and writing for. As another journalist (personal communication) put it, “I think as long as you are telling the truth, and you’re not trying to put any sort of slanted coverage together, then an objective person should be able to recognize that.”

Only one journalist spoke about this in expressly partisan terms, but others clearly implied that unreasonableness applied mostly to the Republican portion of their audience (if only because of the clear evidence that election denialism was rooted in the Republican Party). We discerned that these journalists were implicitly seeking to maintain the trust of these reasonable Republicans. Interpreting their words, these journalists were suggesting that maintaining the trust of this audience, in particular, would preserve the important epistemological dimensions of American democracy: a shared set of facts that people believe in and that must be the basis for political decision making (and, by extension, the legitimacy of those who provide these facts).

What Was Missing from These Accounts

As important as what was said was what was not. Of particular note, very few journalists discussed anything related to political power. In keeping with much public discourse, our interviewees spoke about election denialism through the lens of an “information deficit” model—where people just did not have enough information or were not reasonable about that information. This stands in contrast with what might be thought of as a “power” model—that politicians and their supporters might be embracing denialism to gain or hold onto power, and that this benefitted them in clear ways. There was also very little mention of race across these interviews, even as many political and social scientists center this in their understandings of contemporary political conflicts in the U.S. (Jardina and Mickey 2022; Takahashi et al. 2023). Indeed, there was in general journalistic reliance on ideas such as “polarization” to the exclusion of an analysis of the stakes of politics and contemporary political conflicts (Kreiss and McGregor 2024).

Conclusion

This study has significant implications for the role of journalism in safeguarding democratic elections. Our study underscores the significant democratic limitations of even ostensibly “free” journalism (in the sense of autonomy from the state) in terms of its failure to protect the peaceful transfer of power. As we showed, despite growing threats to electoral institutions in the U.S., journalists failed during 2022 to clearly refute, and condemn, illegitimate power grabs. As such, we believe that U.S. journalism has eroded expectations for the peaceful transfer of power through journalists’ relative silence, and the norm that losers of elections concede and work to gain power in the next one. We believe that these findings clearly underscore the need for empirical analysis as to the role of journalism at times of democratic consolidation, transition, or instability—analysis of media that accounts not only for structures or perceptions of press performance. Our study demonstrates the need for empirical analyses of press performance during the course of elections according to minimalist conceptions of democracy. The picture we paint here is much darker than we might have otherwise guessed in a society marked by media pluralism.

With the global trends of democratic backsliding and threats to democracies, we hope future research extends this analysis of democracy-framed election coverage to different countries and contexts. As detailed above, the U.S. case presented here is clearly relevant for consolidated democracies undergoing autocratization pressures, or countries that face threats to the peaceful transfer of power more broadly (e.g. India, Mazhar and Goraya, 2022). We also see the potential of our framework extending to single-party regimes such as China, Vietnam, and Laos. Truong (2022), for example, argues that even for societies that are not democracies, moving further away from it should be considered democratic backsliding or a threat to democracy. Future research, for instance, can evaluate democracy-framed electoral coverage in single-party regimes that often limit liberal freedoms and circumscribe elections (Levitsky and Way 2010) while tolerating citizen journalism (Luo and Harrison 2019; Truong 2022).

Meanwhile, in democracies journalists have the ability, and the responsibility, to alert the public as to democratic threats. They must develop the processes, orientations, values, and practices to fulfill this democratic role. Embracing the challenge of pro-democracy-framed election coverage as a core guiding professional principle would likely require a reorientation for journalists; away from conceiving their role in narrowly informational terms, and toward an embrace of protecting the very democratic institutions they rely on and educating citizens as to the value of the peaceful transfer of power and legitimate elections—as well as alerting them when they are under threat. We believe that by doing so, media organizations can strengthen the bonds of trust with their audiences—especially through focusing on people that share commitments to democratic values, are willing to lose elections, and respect shared facts and knowledge-producing institutions as one important basis for the peaceful transfer of power.

Supplemental Material

sj-docx-1-hij-10.1177_19401612241235819 – Supplemental material for Safeguarding the Peaceful Transfer of Power: Pro-Democracy Electoral Frames and Journalist Coverage of Election Deniers During the 2022 U.S. Midterm Elections

Supplemental material, sj-docx-1-hij-10.1177_19401612241235819 for Safeguarding the Peaceful Transfer of Power: Pro-Democracy Electoral Frames and Journalist Coverage of Election Deniers During the 2022 U.S. Midterm Elections by Heesoo Jang and Daniel Kreiss in The International Journal of Press/Politics