Much, M., Rutherford, K., Greenfield, J., Tucker, J. A., & Nagler, J. (2026). Did podcasts help Trump win young men? Gendered media spaces and gender gaps. APSA Preprints. https://doi.org/10.33774/apsa-2026-cdb6x
Summary
This paper asks whether entertainment podcasts helped drive the sharp 2024 widening of the youth gender gap, when young men swung markedly toward Trump. The authors introduce the concept of gendered media spaces: non-political entertainment environments where audiences sort on masculine/feminine content preferences rather than ideology. Because masculinity is cognitively linked to conservatism (and femininity to liberalism), this “gendered ideological sorting” exposes men and women to systematically different incidental political messages even when neither is seeking politics. Combining two survey datasets with a large-scale transcript analysis of 36,659 episodes from the 276 most popular U.S. podcasts, they show that podcast masculinity correlates with conservative content, that Trump selected venues on gender rather than ideology, and that conservative podcast exposure is associated with higher Trump support — though it accounts for only a small share of the overall gender gap.
Key Contributions
- Introduces “gendered media spaces” and “gendered ideological sorting” as a political communication mechanism distinct from partisan selective exposure.
- Delivers the first transcript-level, large-scale empirical mapping of the top U.S. podcast ecosystem along both gender expression and ideology.
- Develops a validated LLM-based two-stage (detection then directional scoring) measurement approach for gender and ideological content.
- Empirically documents contrasting candidate venue-selection strategies (Trump on gender, Harris on ideology).
- Links incidental podcast exposure to vote intention/choice using both consumption-panel (Edison) and nationally representative panel (YouGov) data.
Methods
- Transcript-level analysis of 36,659 episodes from the top 276 U.S. podcasts (Jan 2024–Jan 2025), selected via ListenNotes and validated against Edison rankings.
- Audio transcribed/diarized with WhisperX; ~8.5 million speaker chunks scored for gender expression and ideology by Gemini 3 Flash Preview, collapsed to ternary directional scores validated against human coders.
- Episode/podcast indices built as word-count-weighted net gender and net ideology measures (gender blended 90% transcript / 10% host self-ID).
- Edison consumption data used at show-profile level (all 276 podcasts) and individual level (3,544 respondents across 129 non-political podcasts).
- Original YouGov panel (1,167 respondents) modeling Trump vote probability conditional on Joe Rogan and other candidate-interview listenership, controlling for party ID and prior vote.
- Selection addressed by controlling for partisanship throughout and prior Trump vote in the YouGov models, following the talk-radio literature.
Findings
- Political-genre podcasts average 81% political content; non-political podcasts average 15% (median 7%) — enough for meaningful incidental exposure.
- Among non-political episodes with political content: 47% lean liberal, 27% conservative, 26% mixed; among political-genre episodes 72% lean conservative.
- Liberal-leaning podcasts are more numerous overall (192 podcasts, 19,842 episodes vs. 79 conservative, 16,812), but conservative political podcasts produce more content within the political subset.
- Masculine gender expression positively correlates with conservative ideology; nearly all conservative podcasts are heavily masculine.
- Trump’s 16 non-political appearances aligned with podcast gender expression (persuasion of less-sorted masculine audiences); Harris’s 5 aligned with ideological lean (mobilization).
- Young men received the most exposure to conservative content in non-political podcasts of any age-gender cohort.
- Conservative exposure is associated with higher Trump vote intention (concentrated among Independents), with similar per-unit effects across groups but far more exposure among young men.
- Counterfactuals: giving young men young women’s podcast diet lowers predicted Trump support; the reverse modestly raises women’s.
- Self-reported Joe Rogan listening predicts higher Trump vote probability net of party ID and prior vote.
- The overall association is small and cannot explain the entirety of the youth gender gap.
Connections
This work sits within the partisan-media and selective-exposure literatures while pushing them toward incidental, non-ideological sorting, resonating with research on how partisan media and audience communities shape political attitudes such as Arceneaux2026-xk and Munger2025-cz. Its focus on masculine online cultures and gendered content spaces connects to work on manosphere and reactionary media communities like Marwick2026-qd and Copland2025-em. Its LLM-based transcript scoring of ideology and its analysis of candidate messaging strategies relate methodologically to computational content-classification efforts such as Le-Mens2025-qz and Votta2025-xz.
Podcast
A research-radio episode discusses this paper: 🎧 MP3 · Spotify