From Adolescents' Eyes: Assessing an Indicator-Based Intervention to Combat Misinformation on TikTok

Document Title

From the Perspective of Adolescents: Evaluating Indicator-Based TikTok Misinformation Intervention Research

Document Information

• Subject Area: Human-Computer Interaction, Social Media, Misinformation, and User Studies
• Keywords: Misinformation, Fake News, User Intervention, Adolescents, Social Media, Video Sharing Platforms, TikTok
• Publication Source: CHI '24 (Proceedings of the 2024 Conference on Human-Computer Interaction)
• DOI: https://doi.org/10.1145/3613904.3642264

Research Background and Issues

• Issues and Challenges:

  • Video sharing platforms (VSPs) like TikTok present difficulties in distinguishing the quality of information, with misinformation being rampant and causing significant user misguidance, especially during crises (e.g., the Russia-Ukraine conflict).
  • Adolescents, as a primary user group, are often considered "digital natives" but are susceptible to deception due to excessive trust in platform content and a lack of critical thinking skills.
  • Existing text-based content interventions face unique challenges when addressing multimodal video content (e.g., emotional music and misleading sound effects).

• Research Motivation:

  • Address the gap in understanding user responses to multimodal misinformation interventions (e.g., indicator-based tagging and evaluation).
  • Provide user-centered intervention solutions to enhance transparency and reduce negative user emotions (e.g., distrust or resistance to interventions).

• Research Questions:

  1. How do adolescents evaluate the comprehensibility and practicality of indicators addressing misinformation on TikTok?
  2. How can these indicators be integrated into smartphone applications to assist adolescents in combating misinformation on TikTok?
  3. What are adolescents' perceptions of the potential and limitations of indicator-based digital interventions?

Solutions

• Methods and Design:

  1. Two-Stage User Study:
     • Stage 1: Explore how adolescents independently assess misinformation and understand scientifically proposed misinformation indicators through focus group discussions.
     • Stage 2: Design and evaluate a smartphone application prototype demonstrating indicator-based TikTok misinformation interventions.
  2. New Intervention Design:
     • Propose transparent, user-centered misinformation indicators, such as emotional sound effects, manipulative video layouts, and data claims without sources.
     • Indicators aim to prompt users to understand potential misinformation, enabling them to independently judge content credibility.

• Innovations:

  • Develop a generalized indicator system targeting multimodal video features (e.g., emotional music, mismatched audio and video).
  • Design a real-time TikTok misinformation sharing check function, allowing users to access a transparent auxiliary evaluation system via the share button.

• Implementation Steps:

  1. Extract and categorize potential indicators for video misinformation evaluation from source literature.
  2. Define transparent indicators that are scientifically sound and user-comprehensible, validated through experiments and surveys.
  3. Test the designed indicative interface in simulated TikTok scenarios, triggering misinformation evaluation via the "share" button.
  4. Analyze user behavior to refine application design based on dimensions like psychological trust.

Research Outcomes

• Specific Results:

  1. Indicator Evaluation:
     • Identified eight major indicators suitable for video content (e.g., music layering, emotional expression, comment section reactions).
     • Indicators like "manipulative audio" and "attention-grabbing techniques" demonstrated high comprehensibility.
  2. Prototype Design Results:
     • The smartphone application successfully integrated transparent indicators, with overall positive user feedback (SUS satisfaction average score of 81.7).
     • Users were able to quickly improve their evaluation of TikTok videos through detailed indicator prompts.
  3. Knowledge Transfer Effects:
     • Users particularly appreciated the transparent explanation of indicators, suggesting potential applicability to other platforms (e.g., Facebook).
  4. Provided insights into adolescents' perceptions of "social trust" and the risks of "blind trust in algorithms."

• Advantages:

  • Significantly outperformed pure classification or automated tagging (e.g., "true"/"false") in terms of transparency and interpretability.
  • Helped adolescents better understand potential misinformation and independently enhance critical thinking skills.
  • Demonstrated high transferability to other social networking platforms beyond TikTok.

• Limitations and Future Directions:

  1. Limitations:
     • Instances of adolescents "over-trusting" algorithmic prompts.
     • Insufficient understanding of the impact of misclassification (e.g., false positives or false negatives).
     • Challenges in covering all complex multimodal video content.
  2. Future Directions:
     • Conduct cross-platform experiments in real-world scenarios to verify long-term effectiveness.
     • Expand research to larger target groups (e.g., elderly individuals or those with low digital literacy); explore scalability to other social content platforms (e.g., Instagram or YouTube).
     • Extend the indicator system in more contexts, such as detecting "image authenticity details."

Conclusion

This study thoroughly explores the complexity of misinformation on video sharing platforms, proposing a transparent indicator-based user intervention solution validated through real-world adolescent user testing. It provides a solid scientific foundation for assessing the credibility of complex multimodal content in HCI research while emphasizing the importance of balancing adolescents' social roles in digital interactions and safeguarding their rational participation capabilities.