Behavioural Visibility in Data (BeViD): Crafting a new research methodology

  • Professor Irene Ng
  • Professor Tatiana Chameeva Bouzdine
  • Professor Roger Maull

Submitted to Frontiers in Service, 2018 (Austin, Texas)

The transformation to digital technologies in the world of service creates new opportunities for observing human social and behavioural activities due to data being more available for practical studies and in academic research.  To date, most of the studies in psychology, marketing and behavioural economics are intention or attitude based, and behavioural research explains human behaviour through the lens of social preferences, heuristics and norms. Scientific findings are mainly taken from surveys or behavioral laboratory experiments and not from the real life behaviours. Epistemologically, these studies frequently face the tension of being either clearly object–oriented or perspective-driven. Methods are often causality or quasi causality focused, although neuroscience and cognitive science are used to improve the reliability and precision of explaining human activities. 

Advancement in digital technologies and sensors is enabling data to be collected from actual physical day to day interactions in addition to activities on the Internet. There is now data from spending and social media postings, music listens on Spotify, Fitbit sleeps and activities, all of which provide opportunities to observe and analyse authentic human behaviours through data, without being intrusive, if the data can be shared with researchers. How could such data empower the development of research projects and new scientific research tools and methods in social sciences? This is the question we pose in our study as we craft the new methodology on Behavioural Visibility in Data (BeViD).

The ability for individuals to share their own data has only been possible recently, with the HAT (the Hub of All Things) technology that enables the donation of data by individuals themselves, and which allow individuals to collect their own data from a range of Internet services and IoT devices. Individuals can donate or exchange their actual behavioural data, resulting in the possibility of individual behavioural data becoming a source for a new scientific method. Using the BeViD method must include the classic steps of any scientific process: observation; putting forward hypothesis; making predictions as well as testing the predictions by the controlled analysis of other individual BeVID records and finally creating conclusions on the basis of the analysed information. Yet, even if the data is collected from its source (the individual that generated it), it still has the potential confounds of secondary data for example, the Fitbit could have been put on a dog, or the Facebook posts could have been faked. 

We will present the process of creating the BeVID methodology, with the aim of crafting it into a robust and useful research methodology for social sciences. We discuss the questions on how to set up respondent panels, create rules and put in place tests for reliability, and validity. We present different classification algorithms (based on decision- trees, neural networks or fuzzy logic techniques) to ensure higher levels of accuracy in monitoring relevant behavioural activities. The research rigour of the new BeViD methodology must ensure unbiased experimental design, analysis, interpretation, and reporting of results so that it can be a significant contribution to research and society.

Irene Ng