Mobile-Based Experience Sampling for Behaviour Research.

Veljko Pejovic, Neal Lathia, Cecilia Mascolo, Mirco MusolesiEmotions and Personality in Personalized Services 2015 · OpenReview

TL;DR

This paper surveys the evolution of Experience Sampling Methods (ESM) from traditional methods to modern mobile-based approaches (mESM). It highlights how smartphones and wearables are revolutionizing behavior research by enabling richer, more contextual, and scalable data collection, while also discussing the challenges and future directions of this field.

What Problem Does This Paper Address?

Understanding human behavior in its natural context is crucial for fields like psychology and behavioral therapy. Traditional methods like lab studies and retrospective interviews suffer from ecological validity issues and recall bias. The Experience Sampling Method (ESM) was developed to address this by prompting users for in-situ assessments. However, traditional ESM, relying on beepers and paper diaries, was cumbersome, expensive, and lacked the ability to capture "interesting" moments automatically. This paper addresses the challenge of making ESM more efficient, less intrusive, and more powerful by leveraging the capabilities of modern mobile computing devices.

Methods

The paper discusses the shift from traditional ESM methods to mobile-based ESM (mESM). Key advancements highlighted include:

  • Context Sensing: Utilizing smartphone sensors (GPS, accelerometer, light, etc.) to automatically infer the user's context and trigger surveys at opportune moments, rather than relying on pre-programmed times.
  • Ubiquitous Devices: Leveraging smartphones, which are already an integral part of users' lives, reducing the burden of carrying additional devices and increasing participant comfort.
  • Scalability: The ability to distribute mESM applications as mobile apps, enabling large-scale studies across numerous devices simultaneously.
  • Sensor Data Integration: Combining self-reported survey data with passively collected sensor data to provide a more holistic view of behavior and context.
  • Frameworks and Tools: The development of software frameworks (e.g., ESP, MyExperience, funf, ODK, UBhave) that abstract the technical complexities of building mESM applications, making them more accessible to researchers and therapists.
  • Adaptive Sampling: The concept of dynamically adjusting sampling schedules based on predicted user behavior or interruptibility.
  • Behavioral Interventions: The integration of mESM with digital behavior change interventions (dBCIs) to provide personalized feedback and support.

Datasets

Not discussed in the paper. The paper focuses on the methodology and frameworks for mESM rather than presenting results from specific datasets.

Key Insights and Ideas

  • The Smartphone as a "Super-Sensor": Mobile devices are not just for communication; their integrated sensors offer a rich stream of contextual data that can significantly enhance behavioral research.
  • Context is King: Moving beyond random or time-based sampling to context-aware triggers is a major leap in ecological validity for ESM.
  • Reducing Participant Burden: By using existing devices and automating data collection, mESM significantly lowers the barrier to participation and improves data quality.
  • The Rise of Computational Social Science: Mobile sensing is a key enabler for this emerging field, allowing for the analysis of large-scale, fine-grained human behavior data.
  • Beyond Tracking: Intervention is the Future: The ultimate goal is not just to observe behavior but to actively influence it, with mESM serving as a platform for personalized digital interventions.
  • The Challenge of "Stealth" Research: Designing studies that are minimally invasive and seamlessly integrated into users' lives is crucial for long-term engagement and data quality.

Key Results

The paper is a survey and vision paper, not an empirical study with quantitative results. However, it highlights the potential of mESM:

  • Unprecedented Granularity and Scale: mESM allows for the collection of highly detailed behavioral data from a vast number of participants.
  • Improved Ecological Validity: In-situ sampling with contextual awareness leads to more accurate representations of real-world behavior.
  • Reduced Bias: Automation of data collection and context inference minimizes self-report bias and recall issues.
  • Enabling New Research Domains: mESM opens doors to studying areas like location-dependent privacy, sleep monitoring, and physical exercise.

Limitations and Future Work

Limitations:

  • Participant Recruitment and Retention: While smartphones offer scale, ensuring diverse participation and maintaining engagement over long periods remain challenges.
  • Technical Hurdles: Battery life, data transfer costs, and the need for "always-on" applications pose technical constraints.
  • Contextual Bias: The choice of sampling strategy can influence the data collected and the inferences drawn, requiring careful consideration.
  • Privacy Concerns: Handling sensitive location, audio, and video data requires robust privacy-preserving mechanisms.
  • Interruption Load: The increasing number of notifications on smartphones can lead to user fatigue and reduced responsiveness to survey prompts.

Future Work:

  • Anticipatory mESM: Developing systems that can predict future user behavior and context to adapt sampling schedules and deliver interventions proactively.
  • Seamless Integration with Interventions: Creating robust frameworks that seamlessly integrate mESM with digital behavior change interventions (dBCIs).
  • Advanced Context Inference: Improving the accuracy and sophistication of inferring complex behavioral states and intentions from sensor data.
  • Large-Scale Data Mining and Visualization: Developing tools for efficient analysis, interpretation, and feedback of the massive datasets generated by mESM.
  • Ethical Considerations: Continued focus on privacy, consent, and the ethical implications of pervasive mobile sensing.

Summary updated

Abstract

The Experience Sampling Method (ESM) introduces in-situ sampling of human behaviour, and provides researchers and behavioural therapists with ecologically valid and timely assessments of a person's psychological state. This, in turn, opens up new opportunities for understanding behaviour at a scale and granularity that was not possible just a few years ago. The practical applications are many, such as the delivery of personalised and agile behaviour interventions. Mobile computing devices represent a revolutionary platform for improving ESM. They are an inseparable part of our daily lives, context-aware, and can interact with people at suitable moments. Furthermore, these devices are equipped with sensors, and can thus take part of the reporting burden off the participant, and collect data automatically. The goal of this survey is to discuss recent advancements in using mobile technologies for ESM (mESM), and present our vision of the future of mobile experience sampling.
BibTeX
@inproceedings{pejovic2015mobile,
  author = {Veljko Pejovic and Neal Lathia and Cecilia Mascolo and Mirco Musolesi},
  title = {{Mobile-Based Experience Sampling for Behaviour Research.}},
  booktitle = {Emotions and Personality in Personalized Services},
  year = {2015},
  url = {https://openreview.net/forum?id=8B1Nhk9ehHgQ},
}
Figure