Content analysis reveals an inverted U-shaped link between visual sensationalism and user engagement in trending short videos, indicating optimal thresholds for online attention.
Key Points
To examine how textual and visual sensationalist features in video headlines and thumbnails influence online user engagement across different types of media producers.
Conducted a hybrid human-large language model content analysis coding 15 sensationalist features across N=1,100 trending videos from Douyin.
Evaluated differences in textual and visual sensationalism between legacy news outlets and non-journalistic content creators.
Analyzed the linear and nonlinear relationships between multimodal sensationalist elements and audience engagement metrics.
Legacy news outlets used significantly fewer textual sensationalist elements, particularly internet slang, hashtags, and positive emotions, compared to non-journalistic producers.
Visual sensationalist features displayed a nonlinear, inverted U-shaped relationship with user engagement.