Different Audiences, Different Logics Platform-Based Political Engagement on YouTube in the 2020 U.S. Election

Özet

This study examines how the linguistic, emotional, and structural features of political campaign videos are associated with engagement on candidates’ official YouTube channels during the 2020 U.S. presidential election. Because YouTube makes audience responses publicly visible, engagement is conceptualized hierarchically views indicate lower-effort attention, likes reflect evaluative endorsement, and comments capture higher-effort expressive participation. The analysis covers 3,962 videos from the official campaign channels of Joe Biden and Donald Trump. Using natural language processing and separate negative binomial regression models, the study estimates how transcript-based features are associated with engagement outcomes. The findings reveal systematic asymmetries across candidate-centered digital audiences. For Biden, anger-related language, shorter videos, and business-hour posting were positively associated with engagement, whereas for Trump, joy, self-centered discourse, and weekend posting were more consistently associated with higher engagement. Function words also showed divergent patterns, being negatively associated with engagement for Biden but positively associated for Trump. Overall, the results show that engagement on official campaign channels is best understood as resonance within candidate-centered digital audiences, shaped by candidate-audience alignment and platform context rather than uniform message effects.