Understanding Emotion Changes in Mobile Experience Sampling

Title of the Paper

Understanding Emotion Changes in Mobile Experience Sampling

Paper Information

• Field of Study: Human-Computer Interaction and Affective Computing
• Keywords: Emotion changes, mobile experience sampling, affective computing, psychological state, interaction design, mobile devices, contextual factors, self-report, psychological state collection
• Conference: CHI Conference on Human Factors in Computing Systems (CHI '22)
• Date: April 29 - May 5, 2022
• DOI: 10.1145/3491102.3501944

Research Background and Issues

• Identified Problems or Challenges:
  1. Mobile Experience Sampling Method (ESM) relies on self-reports to collect users' emotional states, but the random tasks of ESM may disrupt users, potentially affecting the accuracy of emotional state data.
  2. There is insufficient research exploring the potential impact of ESM tasks themselves on the validity of emotional sampling.
  3. Existing laboratory settings fail to capture diverse and natural emotional data in everyday contexts.
• Importance of the Issues:
  • ESM is a critical tool in affective computing, and validating its effectiveness is essential for obtaining accurate psychological state data.
  • Understanding how ESM disrupts users' emotions can help optimize data collection methods.
• Research Motivation and Related Work:
  • The study focuses on authenticity (reducing recall bias) and naturalness (not affecting participants' normal behavior).
  • Literature indicates that task disruptions may lead to adverse emotional changes, but this has not been linked to ESM sampling.
  • Previous studies have explored emotional changes in laboratory conditions but lack investigations into real-time emotional changes under ESM disruptions.

Solution

• Methods or Solutions:
  1. Develop a new ESM questionnaire with specific questions related to emotional changes.
  2. Collect 2,227 ESM samples from 78 participants, incorporating physiological signals and smartphone usage data to analyze factors related to emotional changes.
  3. Conduct repeated measures ANOVA and multilevel regression analysis to examine the effects of ESM tasks and contextual factors on emotional changes.
• Innovations:
  • The first systematic study of how ESM tasks impact users' emotional states.
  • Integration of diverse contextual information (e.g., location, usage time, physiological signals) to investigate their correlation with emotional changes.
  • Provide optimization recommendations for ESM in affective computing research.
• Implementation Steps and Key Techniques:
  1. Questionnaire Design: Develop a concise and comprehensible multidimensional psychological state assessment questionnaire, including emotional change questions.
  2. Data Collection: Use wearable devices and smartphones to record physiological signals, contextual information, and questionnaire feedback.
  3. Data Analysis:
     • Use ANOVA to identify key contextual factors affecting emotional changes.
     • Apply multilevel regression analysis to evaluate relationships between subjective and objective factors.
  4. Qualitative Analysis: Summarize participants' subjective feedback on emotional changes through interviews.

Research Findings

• Specific Findings:
  1. Approximately 38.6% of ESM responses reported changes in emotional states.
  2. Identified eight significant factors closely related to emotional changes:
     • Subjective factors: emotional valence, attention level, stress level, degree of task disruption.
     • Objective factors: recent location visits, smartphone usage time, heart rate changes, weekend time periods.
  3. Provided daily contextual classifications for positive and negative instances of emotional changes.
• Comparison with Existing Solutions:
  • This study highlights that ESM tasks are not merely data sampling tools but also influence individuals' emotional states, which was previously overlooked.
  • By combining quantitative and qualitative data analysis, the study offers more specific recommendations for improving ESM methods.
• Experimental and Evaluation Results:
  • Statistical tools validated the significant factors.
  • Interviews analyzed the contextual background of participants' emotional changes.
• Limitations and Future Directions:
  1. Sample Distribution Bias: Some users chose to ignore notifications, potentially introducing selection bias.
  2. Sample Population Limitation: Data primarily based on young university students, limiting generalizability.
  3. Device and User Burden: Frequent sampling may increase user burden; future research could explore methods to reduce survey frequency.
  4. Future Research:
     • Investigate the impact of different ESM settings (e.g., notification methods) on emotional changes.
     • Predict suitable sampling timings based on real-time emotional states.
     • Explore deeper relationships between personality traits and perceptions of disruption.

Conclusion

The study provides a novel perspective on emotional data collection in experience sampling, emphasizing the perceptual impact of ESM tasks and proposing optimization methods to enhance the accuracy of psychological data collection and user experience on mobile platforms.