Assessing Wearable Sensor Data for Small Group Research. A Benchmarking and Validation Study

Abstract

Wearable sensors are providing exciting new research opportunities for social scientists. The scholarly community has invested considerable effort to assess the validity and reliability of gathered data over the recent years (Kayhan et al., 2018; Parker et al., 2018). The grand majority of these initial studies has relied on laboratory experiments or field studies with single groups. At the same time, contributions are spread out across different strands of the social-, behavioral- and computer science literature. Findings, therefore, are scattered, and mostly limited to one specific group or field situation without means to assess the influence of wider contextual conditions on sensor based data and insights based on them.

This paper addresses the problem by analyzing and comparing wearable sensor data of ten, relatively small Research and Development (R&D) teams in the context of the H2020 GEDII project (2015-2018). Inter-group variance of sensor measures are explored in the light of complementary data collected, including socio-demographics of team members, gender stereotype, personality traits, and three round-robin ratings regarding advice seeking, friendship and psychological safety. By examining how important sensor measures vary between comparable teams, a more fine-tuned picture regarding the context-sensitive nature of supposedly “objective” sensor measures starts to appear. Our research contributes to the important task of validating sociometric, sensor-based data as new, quantitative measurement tool for social scientists; a methodological proposal for research design, data pre-processing and analysis is included.