Improving Inference from Passively Collected Smartphone Data (Smart Inference)

Research question/goal: 

The widespread use of smartphones creates an enormous amount of digital trace data from log files about smartphone activities (e.g. calls and text messages, app usage) and from smartphone built-in sensors about their everyday behaviours (e.g. mobility, physical activity). Detailed behavioural measures open the possibility for a modernised assessment of social integration, social networks, and stress at the workplace. A sample of 4,293 participants of a nationally representative large-scale panel survey were asked to install a research app (IAB-SMART) on their smartphones, which passively collected novel data for social science research. Beginning in January 2018, 687 (15.9 percent) participants installed the app and contributed data on geolocation, physical activity, app usage, call and SMS logs, and phonebook contacts over the course of half a year. This project builds on and expands preliminary methodological work to improve population inference from the data and to provide access to such data for other research groups. The three objectives of this project are to (1) develop weights that adjust for coverage and nonparticipation error in order to produce unbiased population estimates on the measured constructs such as social integration, social networks, and work-related stress, (2) evaluate sources of measurement error for the different types of sensors and log file data and compare the passively measured data to self-reports, and (3) evaluate ways to anonymise the passively collected smartphone data of the project and make them available to the research community.

Fact sheet

2022 to 2025
Data Sources: 
passively collected smartphone data, survey data
Geographic Space: