Evaluating the Potential of Data-Driven Surveys for Fitness-Tracking Research

Online surveys are used extensively in fitness-tracking research. One common limitation is that researchers cannot assess the veracity of participants' self-reported physical activity data. One promising response to this limitation is using data-driven surveys that integrate participants' online account data (e.g., Fitbit). In this paper, we evaluate using data-driven surveys for fitness-tracking research by examining how participants perceive them, how participants' self-reported data compares to their Fitbit data, and what monetary incentives could motivate participation. To this end, we integrated the participants' Fitbit data in a survey and conducted a three-group study with 𝑁 = 300 participants. We discuss the two main findings: First, although participants were receptive to data-driven surveys, the groups that had to share their data had lower completion rates. Second, there was a large discrepancy between the self-reported data and the data in the participants' Fitbit accounts (e.g., only 38% self-reported a consistent typical weekly exercise-time). We provide suggestions for researchers who conduct online surveys, including data-driven ones, that collect online-account data.

CCS Concepts: • Human-centered computing → Interactive systems and tools; Empirical studies in ubiquitous and mobile computing; Empirical studies in HCI .