One Metric for All: Calculating Interaction Effort of Individual Widgets

Automating usability diagnosis and repair can be a powerful assistance to usability experts and even less knowledgeable developers. To accomplish this goal, evaluating user interaction automatically is crucial, and it has been broadly explored. However, most works focus in long interaction sessions, which makes it difficult to tell how individual interface components influence usability. In contrast, this work aims to compare how different widgets perform for a same task, in the context of evaluating alternative designs for small components, implemented as refactorings. For this purpose, we propose a unified score to compare the widgets involved in each refactoring by the level of effort required by users to interact with them. This score is based on micro-measures automatically captured from interaction logs, so it can be automatically predicted. We show the results of predicting such score using decision trees.