Examining the Impact of YouTube's Video Recommendation Algorithm on Pro- or Anti-Tobacco Messaging

Area of Science:

• Digital Health
• Media Studies
• Public Health Communication

Background:

• YouTube is a primary information source for youth, with its recommendation algorithm influencing 70% of views.
• Content quality on YouTube varies, necessitating an understanding of algorithmic impact on tobacco information seeking.
• The study examines how YouTube's algorithm affects the tobacco-related content viewed by users.

Purpose of the Study:

• To analyze the influence of YouTube's recommendation algorithm on information journeys related to tobacco products.
• To investigate the role of different content sources in tobacco information seeking.
• To identify patterns in how users are directed between pro- and anti-tobacco content.

Main Methods:

• A dataset of 5182 YouTube video journeys was created using popular tobacco-related search queries.
• Videos were analyzed for tobacco relevance, stance (pro/anti), and source type (e.g., news, medical experts).
• The study analyzed viewpoint and source type transitions within recommendation "journeys".

Main Results:

• Most recommendation journeys (56.1%) led to videos with the same stance as the starting video.
• 10.4% of pro-tobacco journeys led to anti-tobacco content, and 13.3% of anti-tobacco journeys led to pro-tobacco content.
• Self-Described Medical Experts (SDMEs) were significant drivers of anti-to-pro tobacco content journeys (81.6%).

Conclusions:

• Self-Described Medical Experts (SDMEs) significantly contribute to users seeking anti-tobacco information being directed to pro-tobacco content.
• This algorithmic influence may expose health information seekers to content that contradicts their health objectives.
• Further research is crucial to understand recommendation feed impacts on tobacco video exposure.