Assessing task difficulty in software testing using biometric measures

Abstract

In this paper, we set out to investigate the extent to which we can classify task difficulty in the software testing domain using psycho-physiological sensors. After reviewing the literature, we adapt the work of Fritz et al. (2014), and transpose it to the testing domain. We present the results of a study conducted with 15 professional software testers carrying out predefined tasks in a lab setting while we collected eye-tracking, brain (EEG) and skin (EDA) data. On average, each participant took part in a two-hour long data collection session. Throughout our study, we captured close to fourteen gigabytes worth of biometric data, consisting of more than a hundred and twenty million data points.

Using this data, we trained 21 Na 虉谋ve Bayes classifiers to predict task difficulty from three perspectives (鈥渂y participant鈥�, 鈥渂y task鈥� and 鈥渂y participant-task鈥�) and using the seven possible combinations of sensors. Our results confirm that we can predict task difficulty for a new tester with a precision of 74.4% and a recall of 72.5% using just an eye-tracker, and for a new task with a precision of 72.2% and a recall of 70.0% using eye-tracking and electrodermal activity. The results achieved are largely consistent with the work of Fritz et al. (2014) among software developers. We go on to provide insights as to what combinations of sensors provide the best results and how this work could be used to develop well-being and workflow support tools in an industry setting.