Introducing openESM: A database of openly available experience sampling datasets

Area of Science:

• Psychology
• Data Science
• Computational Social Science

Background:

• Experience sampling via mobile devices offers deep insights into daily life.
• Individual studies are limited in scope, and scattered, unharmonized data hinder robust analysis.
• Challenges include data accessibility, standardization, and integration for large-scale research.

Purpose of the Study:

• To introduce openESM, an open-source database for harmonized experience sampling datasets.
• To facilitate research on the robustness, generalizability, and heterogeneity of psychological phenomena.
• To create a centralized, searchable resource for the scientific community.

Main Methods:

• Developed openESM, an open-source database with harmonized experience sampling data.
• Collected 60 datasets, >16,000 participants, >740,000 observations.
• Enabled data discovery and download via R and Python packages from openesmdata.org.

Main Results:

• Demonstrated openESM's utility by analyzing within-person affect correlations across 39 datasets.
• Found a significant negative momentary correlation between positive and negative affect ($-0.49$, 95% CI: [ $-0.54$, $-0.42$]).
• Highlighted the potential for large-scale meta-analyses using harmonized data.

Conclusions:

• openESM addresses critical data accessibility and standardization issues in experience sampling research.
• The database supports cumulative research by providing a unified platform for diverse datasets.
• Future development focuses on community-driven expansion and evolving design principles for sustained impact.