The role of the term “fake news” in U.s. alternative media: Shaping political discourse and hyperpartisanship

Summary

This study examines how the term “fake news” is rhetorically deployed in two ideologically extreme U.S. alternative outlets — Newsmax (right) and Occupy Democrats (left) — across nearly a decade (April 2015–October 2023). Combining manual content analysis grounded in Egelhofer et al.’s (2020) typology with LDA topic modeling interpreted through Hallin’s three-sphere model, the authors argue that “fake news” functions overwhelmingly as a weaponized label to delegitimize ideological opponents rather than to describe actual misinformation. Both left- and right-leaning hyperpartisan outlets repurpose the term in functionally similar ways, actively renegotiating the boundaries between consensus, legitimate controversy, and deviance, and thereby eroding shared epistemic norms.

Key Contributions

• Reframes “fake news” research away from audience-level misinformation effects toward the rhetorical deployment of the label itself.
• Extends Egelhofer et al.’s (2020) typology (disinformation genre, empty buzzword, weaponized label) beyond mainstream journalism to ideologically extreme alternative media.
• Integrates Hallin’s macro-level sphere model with Egelhofer et al.’s micro-level typology into a dual-level analytical framework.
• Demonstrates a mixed-methods workflow combining manual coding with LDA topic modeling for large hyperpartisan corpora.
• Draws practical implications for journalism ethics, urging professional restraint in using “fake news” to avoid further trivialization.

Methods

The authors scraped articles containing “fake news” from both outlets, yielding a corpus of 2,536 Newsmax and 513 Occupy Democrats pieces. A 30% stratified random sample (759 / 154) was hand-coded using a binary scheme derived from Egelhofer et al. (2020) covering the three rhetorical uses plus actor categories, with intercoder reliability (Cohen’s κ = .84–.94) established on a ~10% pretest. In parallel, LDA topic models were fit in R (k=11 for Newsmax, k=8 for Occupy Democrats, selected by perplexity), consolidated into higher-order themes, and mapped onto Hallin’s spheres of consensus, legitimate controversy, and deviance.

Findings

• “Fake news” was deployed as a weaponized label in 89.6% of Newsmax and 98.7% of Occupy Democrats articles (χ²=13.14, p<.001), and as an empty buzzword in nearly all articles.
• Occupy Democrats invoked the disinformation-genre framing more often (81.8%) than Newsmax (67.2%); it also reported actors spreading disinformation in 96.1% vs. 73.5% of articles.
• The term appeared incidentally (64–71%) more often than as a main focus, suggesting it operates as a habitual rhetorical reference.
• Newsmax themes targeted mainstream outlets (NYT, CNN, WaPo), framed Trump–press conflicts, and invoked conspiracy and election-interference narratives.
• Occupy Democrats themes centered on Trump-era media conflict, election integrity, Russian interference, and critiques of right-wing propaganda.
• Most themes in both outlets fell within Hallin’s deviance sphere, but identical issues (e.g., Russian interference, mainstream credibility) were classified differently across outlets — empirical evidence for the fluidity of Hallin’s spheres.

Connections

This paper sits at the intersection of hyperpartisan media research and the discursive study of disinformation labels, complementing work that examines how alternative and partisan ecosystems construct epistemic authority — see Frischlich2025-vn and Marwick2025-ov on alternative/partisan media practices, and Kalsnes2025-zb and Farkas2026-lr on the political instrumentalization of “fake news” discourse. Its dual focus on polarization and information disorder connects to broader debates on partisan asymmetry and media fragmentation in Humprecht2025-ml, Rossini2026-jn, and Hameleers2026-mc. The LDA-plus-coding methodology also resonates with computational-qualitative hybrids found in Kansaon2025-id and Nenno2025-xa.

Podcast

A research-radio episode discusses this paper: Listen