Scrolling in the Deep: Analysing Contextual Influences on Intervention Effectiveness during Infinite Scrolling on Social Media

Research Background and Problem

• Identified Problems or Challenges:
  The design of infinite scrolling on social media platforms drives prolonged user engagement, often causing individuals to spend more time than intended. This can lead to various negative emotions, including habitual use, unconscious behavioral loops, and feelings of regret.
  Although some existing interventions have shown initial effectiveness, users gradually become desensitized due to a lack of contextual relevance in the intervention content, resulting in diminished effectiveness.

• Why This Problem is Important:
  Infinite scrolling is considered an attention-capturing "dark pattern" that can stimulate behavioral addiction and cause negative consequences. More importantly, it not only reduces users' well-being but may also lead to poor sleep quality and other health issues. Therefore, developing effective interventions can help improve digital well-being.

• Research Motivation and Related Work:
  While research on behavior change and user context has advanced, these studies primarily focus on active interaction forms (e.g., typing performance). There has been little in-depth exploration of the contextual impact on passive interaction forms like infinite scrolling. Previous studies have suggested that tailoring interventions to users' specific contexts may yield more significant effects, but there is a lack of empirical research in this area.

Solution

• Proposed Solution:
  The authors developed an Android application called InfiniteScape and conducted a 7-day longitudinal user study. The application monitors users' infinite scrolling behavior on social media platforms and displays an intervention overlay prompting users to pause after 15 minutes of continuous scrolling.

• Innovative Aspects of the Solution:
  The authors measured intervention effectiveness along two dimensions: responsiveness (i.e., the time it takes for users to stop scrolling) and resistance (i.e., the level of user opposition to the intervention). Unlike traditional one-size-fits-all interventions, this study focused on contextual factors during user activity, such as mood (valence), being at home, and multitasking. This context-aware approach to analyzing intervention effectiveness is relatively innovative.

• Implementation Steps and Key Techniques:

  1. Develop the InfiniteScape application, which uses Android's Accessibility Service to detect whether users are engaging in infinite scrolling.
  2. Display an intervention overlay prompting users to pause after 15 minutes of continuous scrolling.
  3. Trigger a questionnaire after users stop scrolling to record their current context (e.g., fatigue level, social situation, environment) and resistance to the intervention.
  4. Analyze the collected data using a linear mixed model (LMM) to explore the impact of contextual factors on intervention effectiveness.

Research Outcomes

• Specific Findings:

  • Contextual factors interact with each other, and different scenarios significantly influence the responsiveness and resistance to interventions. For example:
    • Users exhibit significantly lower resistance to interventions when fatigued, but this does not result in faster responsiveness.
    • At home, low emotional states lead to slower responsiveness to interventions, while multitasking (e.g., scrolling while engaging in other activities) results in faster responsiveness.
    • Work-related contexts or other significant external distractions can alleviate negative emotions and improve responsiveness.

• Advantages Over Existing Methods:
  This study provides more detailed insights into the effectiveness of interventions by incorporating contextual information, suggesting that context-aware interventions may be more effective than traditional ones. Existing interventions primarily focus on single measures, such as setting timers or locking screens. In contrast, this study emphasizes the importance of designing interventions based on users' current circumstances.

• Experimental or Evaluation Results:
  The experiment revealed significant contextual interaction effects. For instance, users at home with poor emotional states responded more slowly, while multitasking could serve as an effective distraction to stop infinite scrolling. Additionally, the linear mixed model identified significant interaction effects, such as the "sleepiness × social context" interaction significantly influencing response speed.

• Limitations and Future Directions:

  1. Limitations:
     • Data was collected exclusively from Android users, excluding iOS users.
     • The study duration was 7 days, which did not capture long-term effects.
     • The intervention used pop-up reminders, potentially overlooking other forms of intervention (e.g., visual or behavioral design friction).
     • Some data was missing, such as responses from users who did not stop scrolling.
  2. Future Directions:
     • Broaden the user sample to include iOS users.
     • Increase the diversity of intervention forms, such as progressive interventions or integration with external devices.
     • Explore additional contextual factors, such as whether the content being consumed influences intervention responsiveness.
     • Utilize objective sensing technologies (e.g., emotion recognition or multitasking detection) to enhance the accuracy of real-time context analysis, rather than relying on self-reported data.

Conclusion

This study provides empirical support for designing more effective digital well-being measures by combining context awareness with interventions for infinite scrolling. While the research highlights important contextual interaction effects, the practical complexity and privacy concerns of its application require further exploration. Future research can build on this foundation to investigate broader intervention methods and technical implementations.