In Adjunct Proceedings of Ubicomp/ISWC '15, MCSS 2015: 4th ACM Workshop on Mobile Systems for Computational Social Science. September 8, Osaka, Japan, pp. 723–732.
The mobile phone-based Experience Sampling Method (ESM) enables in situ recording of human behaviour and experience by querying users, via their smartphones, anywhere and anytime. Sampling can happen on a previously unimaginable scale, and across a diverse pool of participants. Therefore, mobile ESM is not limited to capturing users’ manual responses, as the surrounding context can be automatically captured by mobile sensors. However, obtaining high quality data with ESM is challenging, as users may fail to respond honestly, or may even ignore the questionnaire prompts if they perceive the study as too burdensome. In this paper, we discuss the potential of using interruptibility prediction models to deliver mobile ESM questionnaires at opportune moments, and thus improve the effectiveness of a study. We examine context prediction and interruptibility inference, which are fundamental challenges that need we need to overcome in order to make mobile ESMs better aligned with a user’s lifestyle, and consequently paint a truthful picture of a user’s behaviour.