arxiv: 2604.23914 · v1 · submitted 2026-04-26 · 💻 cs.SI

Recognition: unknown

#MakeBeefGreatAgain: A Cross-Platform Analysis of Early #MAHA Discourse

Haoning Xue , Yue Li , Benjamin A. Lyons , Andy J. King

Authors on Pith no claims yet

Pith reviewed 2026-05-08 04:47 UTC · model grok-4.3

classification 💻 cs.SI

keywords MAHAcampaign sloganssocial media discourseagenda meldingcross-platform analysishashtag reinterpretationpolitical communicationpublic agenda formation

0 comments

The pith

Most early #MAHA posts ignored the campaign's five official health priorities.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper examines 41,819 posts using the #MAHA hashtag from September 2024 to January 2025 to test whether public discussion matched the Make America Healthy Again campaign agenda. It finds that 81.3 percent of posts engaged none of the five stated priorities, with platforms instead forming three separate discourse clusters. A sympathetic reader would care because the results illustrate how a campaign slogan circulates as a loose symbolic frame rather than a focused policy signal in fragmented online spaces. The analysis draws on agenda-melding theory to track thematic structure and temporal shifts across platforms.

Core claim

The analysis shows a substantial disconnect between #MAHA public discourse and the stated MAHA agenda, with 81.3 percent of posts not engaging any of the five campaign priorities. Platforms clustered into three broad environments: grassroots partisan-support spaces, informational sources, and health-focused spaces. As a result, #MAHA functioned less as a unified campaign agenda than as a symbolic frame interpreted differently across platforms.

What carries the argument

Agenda-Melding Theory, which tracks how public and media agendas interact and diverge in digital settings, applied via structural topic modeling and AI-assisted annotation to classify whether posts addressed the campaign priorities.

If this is right

Campaign slogans often circulate in public discourse without direct ties to their original policy priorities.
The same hashtag produces distinct thematic clusters depending on the platform and user community.
Public agendas around political slogans form through reinterpretation rather than direct transmission of campaign messaging.
Fragmented digital spaces transform how campaign language is amplified and repurposed over time.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

Campaign teams may need platform-tailored messaging to reduce divergence between slogan use and intended priorities.
Similar patterns of slogan reinterpretation could appear in other political hashtags and reduce message coherence.
Over longer periods the hashtag might stabilize around meanings unrelated to the original five priorities.

Load-bearing premise

The five official MAHA campaign priorities were correctly identified from official sources and the AI-assisted annotation plus topic modeling accurately distinguished posts that engaged those priorities from those that did not.

What would settle it

An independent manual coding of a random sample of the 41,819 posts that finds substantially more than 18.7 percent of posts addressing at least one of the five priorities would undermine the central claim of widespread disconnect.

Figures

Figures reproduced from arXiv: 2604.23914 by Andy J. King, Benjamin A. Lyons, Haoning Xue, Yue Li.

**Figure 1.** Figure 1: Interrupted time-series analysis results of (A) daily #MAHA post count and (B) daily proportion of MAHA campaignrelevant posts from Sep 05, 2024 to Jan 10, 2025. Gray points indicate observed daily values. The vertical dashed line marks the presidential election day (i.e., November 5, 2024) as the interruption point. Colored solid lines represent fitted interrupted time-series trends with 95% CIs. The gra… view at source ↗

**Figure 2.** Figure 2: Interrupted time-series analysis results of STM-derived topic prevalence from Sep 05, 2024 to Jan 10, 2025. Gray points indicate observed daily values. The vertical dashed line marks the presidential election day (i.e., November 5, 2024) as the interruption point. Colored solid lines represent fitted interrupted time-series trends with 95% CIs. The gray dashed lines represent the expected pattern in the ab… view at source ↗

**Figure 3.** Figure 3: Distribution of MAHA campaign agenda categories across STM-derived #MAHA topics in a Sankey diagram. 2.4 Early #MAHA Public Discourse Across Seven Online Platforms To answer RQ3, we examined how the prevalence of STM-derived topics varied across online platforms using a Pearson chisquare test (see view at source ↗

**Figure 4.** Figure 4: STM-derived topic prevalence across seven online platforms. The disconnect between #MAHA public discourse and the MAHA campaign agenda priorities is consistent with AMT [10]. Rather than passively reproducing elite agendas, AMT suggests that individuals actively reinterpret, prioritize, and downplay issues from campaigns, media, and social groups. Early #MAHA public discourse did not simply echo the campai… view at source ↗

**Figure 5.** Figure 5: Flow diagram of data collection, ChatGPT-assisted relevance classification, and ChatGPT-assisted campaign agenda classification. This study draws on a cross-platform dataset of N = 41,819 posts related to #MAHA across seven online sources, including six social media platforms (i.e., TikTok, Instagram, Facebook, X, Reddit, and YouTube) and web articles (see details below). Posts from these sources were conn… view at source ↗

read the original abstract

Make America Healthy Again (MAHA) is a health-related campaign slogan proposed by Robert F. Kennedy Jr. and later incorporated into the political coalition of President Trump. While #MAHA quickly circulated beyond the campaign itself and became a prominent hashtag for public discussion, it remains unclear whether this public discourse reflected, reshaped, or diverged from the stated agenda of the MAHA campaign. This study presents a large-scale, cross-platform analysis of early #MAHA public discourse between September 2024 and January 2025, using the framework of Agenda-Melding Theory. Drawing on 41,819 #MAHA-related posts, this study combines structural topic modeling, interrupted time-series analysis, and AI-assisted data annotation to examine the thematic structure and temporal dynamics. The most prominent finding is the substantial disconnect between #MAHA public discourse and the stated MAHA agenda: 81.3% of posts did not engage any of the five campaign priorities of the MAHA campaign. There were also pronounced cross-platform differences, with online platforms clustering into three broad discourse environments: (a) grassroots partisan-support spaces, (b) informational sources, and (c) health-focused spaces. #MAHA functioned less as a unified campaign agenda than as a symbolic frame interpreted differently across platforms. More broadly, this study provides useful empirical insight into how campaign slogans are reinterpreted and how public agendas are formed, amplified, and transformed in the fragmented digital environments.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Desk Editor's Note private letter to a colleague

The paper quantifies a sharp disconnect in #MAHA posts from the campaign priorities but the 81.3% figure depends on AI annotation with no validation details shown.

read the letter

The main thing here is that 81.3% of the 41,819 #MAHA posts did not engage the five stated campaign priorities, and the platforms split into three distinct discourse clusters. That is a concrete new observation for agenda-melding work on this particular slogan and time window. The cross-platform breakdown into grassroots partisan, informational, and health-focused spaces gives a clear picture of how the hashtag got repurposed rather than staying tied to the original agenda. The scale of the data and the use of structural topic modeling plus interrupted time series to track themes and timing are straightforward and appropriate for the question. Those parts make the divergence claim worth looking at even if the exact percentage needs checking. The soft spot is the annotation process that produced the 81.3% number. The abstract gives no accuracy, precision, recall, or human agreement stats for the AI-assisted labeling, nor does it spell out how engagement was defined or how the five priorities were extracted from official sources. If the classifier has systematic errors, the disconnect result and the cluster analysis built on it lose force. The paper should include those checks and any robustness tests. This is for people working on political communication, hashtag dynamics, or how public health slogans travel in fragmented online spaces. A reader who wants empirical examples of agenda formation across platforms will get usable material from the clustering and temporal patterns. It deserves peer review because the data collection is large and the methods are established, even though the validation gap on the key classifier will need fixing before publication.

Referee Report

2 major / 2 minor

Summary. The paper analyzes 41,819 #MAHA-related social media posts collected between September 2024 and January 2025. It applies structural topic modeling, interrupted time-series analysis, and AI-assisted annotation to map thematic structure, temporal dynamics, and cross-platform variation. The central empirical claim is that public #MAHA discourse diverges sharply from the official MAHA campaign agenda, with 81.3% of posts engaging none of the five stated campaign priorities; platforms cluster into three discourse environments (grassroots partisan-support, informational, and health-focused), and #MAHA functions primarily as a symbolic frame rather than a unified agenda.

Significance. If the priority classification proves reliable, the study supplies concrete, large-scale evidence that campaign slogans are rapidly reinterpreted in fragmented digital environments, consistent with Agenda-Melding Theory. The combination of topic modeling, time-series methods, and cross-platform comparison on a sizable corpus is a methodological strength that could inform future work on agenda formation online.

major comments (2)

[Abstract and Methods] Abstract and Methods section on AI-assisted annotation: The 81.3% figure (and all downstream claims about thematic disconnect and platform clustering) rests on the binary classification of posts as engaging or not engaging the five MAHA priorities. No definition of 'engagement,' no enumeration of the five priorities, and no validation statistics (accuracy, precision/recall, human-AI agreement rate, or size of the validation sample) are supplied. Without these, it is impossible to assess whether the reported disconnect is a measurement artifact.
[Results] Results section reporting the 81.3% statistic: The claim that 81.3% of the 41,819 posts engage none of the five priorities is presented as a direct count from the annotated data. If the AI classifier was not validated against a human-coded gold standard, systematic mislabeling (e.g., over- or under-detection of priority-related language) would propagate into the structural topic model outputs and the interrupted time-series analysis, undermining the cross-platform clustering interpretation.

minor comments (2)

[Abstract] Abstract: The five campaign priorities are referenced but never listed; including a brief enumeration would allow readers to evaluate the scope of the 'disconnect' claim without consulting external sources.
[Introduction] The title references '#MakeBeefGreatAgain' while the body concerns #MAHA; a short clarifying sentence in the introduction would prevent reader confusion about the relationship between the two hashtags.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for their careful and constructive review. We agree that the AI-assisted annotation requires substantially more methodological detail to support the central claims. We will revise the manuscript accordingly and address each comment below.

read point-by-point responses

Referee: [Abstract and Methods] Abstract and Methods section on AI-assisted annotation: The 81.3% figure (and all downstream claims about thematic disconnect and platform clustering) rests on the binary classification of posts as engaging or not engaging the five MAHA priorities. No definition of 'engagement,' no enumeration of the five priorities, and no validation statistics (accuracy, precision/recall, human-AI agreement rate, or size of the validation sample) are supplied. Without these, it is impossible to assess whether the reported disconnect is a measurement artifact.

Authors: We concur that these details are essential for evaluating the classification. In the revised manuscript we will expand the Methods section to enumerate the five MAHA campaign priorities, supply an explicit operational definition of 'engagement,' and report full validation statistics (accuracy, precision, recall, human-AI agreement rate, and validation-sample size) from a human-coded gold standard. These additions will allow readers to judge whether the 81.3% figure reflects a genuine thematic disconnect. revision: yes
Referee: [Results] Results section reporting the 81.3% statistic: The claim that 81.3% of the 41,819 posts engage none of the five priorities is presented as a direct count from the annotated data. If the AI classifier was not validated against a human-coded gold standard, systematic mislabeling (e.g., over- or under-detection of priority-related language) would propagate into the structural topic model outputs and the interrupted time-series analysis, undermining the cross-platform clustering interpretation.

Authors: We acknowledge the risk of error propagation. The revised manuscript will include a dedicated validation subsection demonstrating the AI classifier's performance against human annotations. Reporting these metrics will substantiate the 81.3% statistic and support the reliability of the subsequent topic-model and time-series results. revision: yes

Circularity Check

0 steps flagged

No circularity: primary result is direct empirical count from annotated posts

full rationale

The paper's central claim (81.3% of 41,819 posts did not engage the five MAHA priorities) is produced by collecting #MAHA posts and applying AI-assisted annotation plus structural topic modeling to label engagement. This is an observational measurement on external data rather than any derivation that reduces by construction to fitted parameters, self-definitions, or self-citations. No equations, ansatzes, or uniqueness theorems are invoked that would make the percentage tautological; the result stands or falls on the validity of the annotation process and data collection, which are independent inputs.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The analysis rests on the domain assumption that agenda-melding theory is an appropriate lens and that standard computational social science techniques can measure engagement with campaign priorities.

axioms (1)

domain assumption Agenda-Melding Theory supplies a valid framework for interpreting slogan reinterpretation across platforms
Invoked to frame the observed disconnect between campaign agenda and public discourse.

pith-pipeline@v0.9.0 · 5571 in / 1131 out tokens · 60789 ms · 2026-05-08T04:47:39.704684+00:00 · methodology

discussion (0)

Reference graph

Works this paper leans on

31 extracted references · 13 canonical work pages

[1]

Make America Healthy Again (MAHA) PAC: Official Site, March 2025

MAHA. Make America Healthy Again (MAHA) PAC: Official Site, March 2025. URL https://web.archive.org/web/ 20250308194651/https://www.maha.vote/

2025
[2]

Establishing the President’s Make America Healthy Again Commission, February 2025

The White House. Establishing the President’s Make America Healthy Again Commission, February 2025. URL https://www.whitehouse.gov/presidential-actions/2025/02/establishing-the-presidents-make-america-healthy-again- commission/

2025
[3]

State Food Additive Legislation Surged Across 38 States in 2025 (Plus Other Food Policy Trends We Saw Last Year)

Katherine Tschopp. State Food Additive Legislation Surged Across 38 States in 2025 (Plus Other Food Policy Trends We Saw Last Year). URL https://www.multistate.us/insider/2026/1/14/state-food-additive-legislation-surged-across-38- states-in-2025-plus-other-food-policy-trends-we-saw-last-year

2025
[4]

#maha, 2026

TikTok. #maha, 2026. URL https://www.tiktok.com/tag/maha

2026
[5]

#maha • Instagram, 2026

Instagram. #maha • Instagram, 2026. URL https://www.instagram.com/explore/search/keyword/?q=%23maha

2026
[6]

The MAHA sphere of influence.POLITICO, March 2026

Kelly Hooper and Chelsea Cirruzzo. The MAHA sphere of influence.POLITICO, March 2026. URL https://www. politico.com/newsletters/politico-pulse/2025/05/23/the-maha-sphere-of-influence-00366757

2026
[7]

Whole Foods, MAHA and the Battle Over Healthy Eating in America.The New York Times, October

Ben Ryder Howe. Whole Foods, MAHA and the Battle Over Healthy Eating in America.The New York Times, October
[8]

URL https://www.nytimes.com/2025/10/29/dining/whole-foods.html

ISSN 0362-4331. URL https://www.nytimes.com/2025/10/29/dining/whole-foods.html

2025
[9]

Watch: Is MAHA the New MAGA?, February 2026

Hannah Norman Sable-Smith, Bram. Watch: Is MAHA the New MAGA?, February 2026. URL https://kffhealthnews. org/news/article/watch-video-make-america-healthy-again-maha-maga-rfk-explainer/

2026
[10]

Tina Reed. RFK Jr. and MAHA’s not-so-secret weapon – moms, August 2025. URL https://www.axios.com/2025/08/31/ rfk-jr-vaccine-fluoride-moms-influencers

2025
[11]

Shaw, Maxwell McCombs, David H

Donald L. Shaw, Maxwell McCombs, David H. Weaver, and Bradley J. Hamm. INDIVIDUALS, GROUPS, AND AGENDA MELDING: A THEORY OF SOCIAL DISSONANCE.International Journal of Public Opinion Research, 11(1):2–24, March 1999. ISSN 0954-2892. doi: 10.1093/ijpor/11.1.2. URL https://doi.org/10.1093/ijpor/11.1.2

work page doi:10.1093/ijpor/11.1.2 1999
[12]

Cross-platform social dynamics: an analysis of ChatGPT and COVID-19 vaccine conversations.Scientific Reports, 14(1):2789, February 2024

Shayan Alipour, Alessandro Galeazzi, Emanuele Sangiorgio, Michele Avalle, Ljubisa Bojic, Matteo Cinelli, and Walter Quattrociocchi. Cross-platform social dynamics: an analysis of ChatGPT and COVID-19 vaccine conversations.Scientific Reports, 14(1):2789, February 2024. ISSN 2045-2322. doi: 10.1038/s41598-024-53124-x. URL https://doi.org/10.1038/ s41598-024-53124-x

work page doi:10.1038/s41598-024-53124-x 2024
[13]

Pevehouse, Yibing Sun, Hyerin Kwon, Ruixue Lian, Rui Wang, Xinxia Dong, and Dhavan V

Mengyu Li, Jiyoun Suk, Yini Zhang, Jon C. Pevehouse, Yibing Sun, Hyerin Kwon, Ruixue Lian, Rui Wang, Xinxia Dong, and Dhavan V . Shah. Platform affordances, discursive opportunities, and social media activism: A cross-platform analysis of #MeToo on Twitter, Facebook, and Reddit, 2017–2020.New Media & Society, 28(1):119–147, January 2026. ISSN 1461-4448. d...

work page doi:10.1177/14614448241285562 2017
[14]

Daniel J Sude and Shira Dvir-Gvirsman. Different platforms, different uses: testing the effect of platforms and individual differences on perception of incivility and self-reported uncivil behavior.Journal of Computer-Mediated Communication, 28(2):zmac035, March 2023. ISSN 1083-6101. doi: 10.1093/jcmc/zmac035. URL https://doi.org/10.1093/jcmc/zmac035

work page doi:10.1093/jcmc/zmac035 2023
[15]

Ideological frag- mentation of the social media ecosystem: From echo chambers to echo platforms.PNAS Nexus, 4(9):pgaf262, September

Edoardo Di Martino, Alessandro Galeazzi, Michele Starnini, Walter Quattrociocchi, and Matteo Cinelli. Ideological frag- mentation of the social media ecosystem: From echo chambers to echo platforms.PNAS Nexus, 4(9):pgaf262, September
[16]

doi: 10.1093/pnasnexus/pgaf262

ISSN 2752-6542. doi: 10.1093/pnasnexus/pgaf262. URL https://doi.org/10.1093/pnasnexus/pgaf262

work page doi:10.1093/pnasnexus/pgaf262
[17]

Pelletier, Alexandra Krallman, Frank G

Mark J. Pelletier, Alexandra Krallman, Frank G. Adams, and Tyler Hancock. One size doesn’t fit all: a uses and gratifica- tions analysis of social media platforms.Journal of Research in Interactive Marketing, 14(2):269–284, June 2020. ISSN 2040-7122. doi: 10.1108/JRIM-10-2019-0159. URL https://doi.org/10.1108/JRIM-10-2019-0159

work page doi:10.1108/jrim-10-2019-0159 2020
[18]

Michael D. Slater. Reinforcing Spirals: The Mutual Influence of Media Selectivity and Media Effects and Their Impact on Individual Behavior and Social Identity.Communication Theory, 17(3):281–303, August 2007. ISSN 1050-3293. doi: 10.1111/j.1468-2885.2007.00296.x. URL https://doi.org/10.1111/j.1468-2885.2007.00296.x

work page doi:10.1111/j.1468-2885.2007.00296.x 2007
[19]

agendamelding

Chris J. Vargo. Public “agendamelding” in the United States: assessing the relative influence of different types of online news on partisan agendas from 2015 to 2020.Journal of Information Technology & Politics, 19(3):284–301, July 2022. ISSN 1933-1681. doi: 10.1080/19331681.2021.1972893. URL https://doi.org/10.1080/19331681.2021.1972893. eprint: https://...

work page doi:10.1080/19331681.2021.1972893 2015
[20]

Harrison

Yumeng Luo and Teresa M. Harrison. How citizen journalists impact the agendas of traditional media and the government policymaking process in China.Global Media and China, 4(1):72–93, March 2019. ISSN 2059-4364. doi: 10.1177/ 2059436419835771. URL https://doi.org/10.1177/2059436419835771

work page doi:10.1177/2059436419835771 2019
[21]

Social Media and Political Agenda Setting.Political Communication, 39(1):39–60, January 2022

Fabrizio Gilardi, Theresa Gessler, Ma ¨el Kubli, and Stefan M ¨uller. Social Media and Political Agenda Setting.Political Communication, 39(1):39–60, January 2022. ISSN 1058-4609. doi: 10.1080/10584609.2021.1910390. URL https: //doi.org/10.1080/10584609.2021.1910390. eprint: https://doi.org/10.1080/10584609.2021.1910390

work page doi:10.1080/10584609.2021.1910390 2022
[22]

Desiree Steppat, Laia Castro Herrero, and Frank Esser. Selective exposure in different political information environ- ments – How media fragmentation and polarization shape congruent news use.European Journal of Communication, 37(1):82–102, February 2022. ISSN 0267-3231. doi: 10.1177/02673231211012141. URL https://doi.org/10.1177/ 02673231211012141

work page doi:10.1177/02673231211012141 2022
[23]

Pablo Porten-Che ´e and Christiane Eilders. Fragmentation in high-choice media environments from a micro-perspective: Effects of selective exposure on issue diversity in individual repertoires.Communications, 44(2):139–161, June 2019. ISSN 1613-4087. doi: 10.1515/commun-2018-2013. URL https://www.degruyterbrill.com/document/doi/10.1515/ commun-2018-2013/html

work page doi:10.1515/commun-2018-2013 2019
[24]

Active audience and uses and gratifications

Helen Wood. Active audience and uses and gratifications. InThe SAGE Handbook of Television Studies. SAGE Publica- tions, Limited, London, UNITED KINGDOM, 2014. ISBN 978-1-4739-1108-6. URL http://ebookcentral.proquest.com/ lib/utah/detail.action?docID=4013166

2014
[25]

Uses and Gratifications as Media Choice

Marina Krcmar and Yuliya Strizhakova. Uses and Gratifications as Media Choice. InMedia Choice : A Theoretical and Empirical Overview. Taylor & Francis Group, Oxford, UNITED KINGDOM, 2009. ISBN 978-0-203-93865-2. URL http://ebookcentral.proquest.com/lib/utah/detail.action?docID=425515

2009
[26]

Research API|TikTok for Developers, 2024

TikTok. Research API|TikTok for Developers, 2024. URL https://developers.tiktok.com/products/research-api/

2024
[27]

API Reference|YouTube Data API, 2025

YouTube Data API. API Reference|YouTube Data API, 2025. URL https://developers.google.com/youtube/v3/docs

2025
[28]

Newswhip Spike, 2026

Newswhip. Newswhip Spike, 2026. URL https://www.newswhip.com/spike-real-time-media-monitoring/

2026
[29]

Real-time web articles monitoring, 2026

Newswhip. Real-time web articles monitoring, 2026. URL https://www.newswhip.com/real-time-web-article- monitoring/

2026
[30]

Roberts, Brandon M

Margaret E. Roberts, Brandon M. Stewart, and Dustin Tingley. stm: An R Package for Structural Topic Models.Journal of Statistical Software, 91(2):1 – 40, October 2019. doi: 10.18637/jss.v091.i02. URL https://www.jstatsoft.org/index.php/ jss/article/view/v091i02

work page doi:10.18637/jss.v091.i02 2019
[31]

clean food

Jos ´e Pinheiro, Douglas Bates, and R Core Team. nlme: Linear and Nonlinear Mixed Effects Models, 2026. URL https: //CRAN.R-project.org/package=nlme. Supplementary Information for #MakeBeefGreatAgain: A Cross-Platform Analysis of Early #MAHA Discourse Post count Agenda-aligned post % b (SE) p b (SE) p Intercept 47.96 (129.64) .712 0.131 (0.03) .000 Time 5...

2026