My PhD thesis is titled Sensor-supported, Unsupervised Observation Techniques for Field Studies and presents a set of unobtrusive field observation techniques. The thesis was reviewed and accepted for publication by Prof. Dr. Susanne Boll and Prof. Charlotte Magnusson. I successfully defended on July 17, 2014. A PDF version can be downloaded: PDF (300 dpi image resolution, 16.5 MB) or PDF (150 dpi image resolution, 5.2 MB).
We live in a mobile era and ubiquitous applications and interactive services surround us everywhere. To design and develop targeted and successful applications for these environments a detailed understanding of the users and their mobile context is needed.
Traditional lab studies and lab-based observation methods are mostly unsuited to investigate mobile applications, their users and the users’ experiences in such mobile and highly dynamic settings. Consequently, more and more research is done in field studies, where the users, applications and services are investigated in their natural environments. Common field observation methods, such as shadowing, are facing three essential challenges, i.e., situatedness, obtrusiveness, and scalability, and are in most cases unable to meet all of them. While mobile technologies continue to disappear, field observation techniques often still require the presence of human observers and observation tools, like video cameras.
In the last years, research has identified unsupervised observation techniques as promising approach for field studies. Popular examples for these techniques are logging, the Experience Sampling Method (ESM), and diary studies. In such techniques the human observer is replaced by sensors or self-reporting techniques, which are typically hosted on a mobile phone. However, although they inherently overcome many limitations of traditional field study observation techniques, they still fail to meet the three key challenges fully.
Approach & Outcome
In this thesis I investigate to what extent the sensing and computation capabilities of smart phones can improve unsupervised observation techniques. I demonstrate that the information gain of logging can be increased, if statistical analysis is applied and the sensor data is combined with static, environmental information. Further, I show that situation-aware self-reporting techniques, such as ESM, can trigger inquiries in opportune moments and, therefore, are perceived less obtrusive and are more likely to be answered. Finally, I illustrate that properly prepared and presented in situ information can serve as a valuable key resource for post hoc studies, like follow-up interviews.
I contribute a novel, sensor-based observation approach, which has shown its benefits and advantages in a set of structured field studies of different complexities and dynamics. Further, I contribute the Virtual Observer, a holistic observation framework and reference implementation of my observation approach, which facilitates use and application of my method in other contexts and domains. Both, the novel observation approach and its reference implementation, will enable researchers to conduct unobtrusive, scalable field studies that feature convincing situational details.