Systematic partisan content skews in TikTok during the 2024 US elections

Area of Science:

• Political Science
• Computational Social Science
• Media Studies

Background:

• Social media platforms significantly influence political information exposure.
• Algorithmic curation's role in shaping political exposure is debated, especially on platforms with user-controlled feeds.
• TikTok's 'For You' feed, driven by algorithms, offers a unique setting to study constrained user agency.

Purpose of the Study:

• To investigate partisan imbalances in political information exposure on TikTok's algorithmic feed.
• To determine if algorithmic recommendations create asymmetric exposure to co-partisan and cross-partisan content.
• To identify specific content domains and account types contributing to these asymmetries.

Main Methods:

• Conducted 323 audit experiments using controlled 'sock puppet' accounts across three US states.
• Seeded accounts with either Democratic or Republican content to track algorithmic recommendations.
• Collected over 280,000 recommendations over 27 weeks during the 2024 US presidential election campaign.

Main Results:

• Republican-seeded accounts received 11.5% more co-partisan content than Democratic-seeded accounts.
• Democratic-seeded accounts were exposed to 7.5% more cross-partisan content (largely anti-Democratic material).
• Asymmetries were concentrated in high-reach Republican channels and specific policy domains (e.g., immigration, crime, abortion).

Conclusions:

• Algorithmic recommendations on TikTok create significant, asymmetric partisan imbalances in political information exposure.
• These findings have critical implications for platform governance and the health of democratic discourse.
• The study highlights the need for greater transparency and accountability in algorithmic content curation.