Micro-narratives: A Scalable Method for Eliciting Stories of People’s Lived Experience

Research Background and Problem

• Problem or Challenge: Existing data collection methods face difficulties in balancing data quality and scale. Particularly in Human-Computer Interaction (HCI) research, collecting large-scale qualitative data from participants incurs significant resource costs. Methods such as interviews and diary studies provide deep insights but are time-consuming and costly, while surveys and Ecological Momentary Assessments (EMA) can scale up data collection but lack depth in qualitative information.
• Significance: Gaining a deep understanding of people's real-life experiences is critical for mental health research and addressing social issues. There is an urgent need for a tool that can achieve both data quality and low participation costs in large-scale applications, especially for young people or vulnerable groups.
• Research Motivation and Related Work: This study is inspired by vignette studies in psychology, aiming to collect participant-generated narratives tailored to specific topics while reducing cognitive burden. Additionally, recent advancements in generative AI, such as large language models (LLMs), provide technical support for such innovations.

Solution

• Method or Solution:
  • A micro-narrative collection method based on human-AI collaborative design is proposed, leveraging AI to assist participants in generating personal narratives that express their life experiences.
  • The method involves a workflow designed in three stages: problem posing, AI-generated narratives, and participant review and customization of the narratives.
• Innovations:
  • Utilizes large language models (LLMs) to break down the cognitive burden in the narrative generation process, enabling non-professional writers to create narratives fluently.
  • A unique human-computer interaction framework allows participants to comfortably express their experiences while controlling and modifying the resulting narratives, ensuring alignment with personal records.
• Implementation Steps and Techniques:
  1. Stage One: Collect contextual "fragments" (e.g., details of events and psychological feelings) through AI-facilitated Q&A.
  2. Stage Two: Use an LLM to integrate these "fragments" into coherent micro-narratives. The system generates narrative versions with different "tones" based on preset templates.
  3. Stage Three: Participants review the narratives, select the most suitable version, and further adjust them to ensure the narratives authentically reflect their personal feelings.

Research Outcomes

• Specific Outcomes:
  • Through three empirical studies (including two preliminary trials and one comparative experiment), the feasibility and acceptability of the micro-narrative tool were validated:
    • In the preliminary trials, participants generated narratives with AI support and provided a high proportion of positive feedback.
    • The comparative experiment showed that, compared to traditional open-ended questionnaires, the micro-narrative method performed better across multiple dimensions, such as ease of use and accuracy of emotional expression.
  • Informal expert consultations revealed the tool's potential for wide application in areas such as mental health intervention design, behavioral analysis, and clinical research.
• Advantages:
  • Compared to traditional open-ended questionnaires, micro-narratives were perceived as more effective in expressing emotions and contexts, and participants were more likely to recommend this method.
  • Participants reported that the method enhanced self-reflection and emotional organization during the trials.
• Experimental and Evaluation Results:
  • In randomized experiments, participants rated micro-narratives as more suitable for young people, better at capturing the "voice" of narratives, and providing a higher sense of personal benefit (e.g., "sense of meaning") compared to open-ended questionnaires.
• Limitations and Future Directions:
  • Ethical and Safety Issues: While no significant ethical concerns arose during the study, large-scale deployment in the future will require careful attention to privacy protection and data sensitivity, especially when engaging participants with potential mental health risks.
  • Open Questions: Further exploration is needed regarding the credibility of human-AI co-generated narrative data (whether it truly reflects participants' intentions) and methods for analyzing such narratives.
  • Extended Applications: The study suggests the potential use of the micro-narrative method as a psychological intervention tool, but this requires integration with clinical psychology theories and the establishment of rigorous development standards.

This study demonstrates the potential of micro-narratives as a novel qualitative data collection method and paves the way for expanding and optimizing this tool in other fields. It also highlights areas requiring further exploration in terms of ethics, safety, and technology.