Evaluating User and System Models: Applying Scaling Techniques to Problems in Human-Computer Interaction

Abstract

A user's mental model of a system should be an important determinant of performance and as well as a means of understanding why particular user errors occur. In particular, experienced users' models should be in closer agreement with the system than less experienced users' models, and deviations of expert models from the system should correspond to difficulties in performance and suggest ways that system usability could be improved. The present study explored the utility of scaling techniques for defining and comparing user and system models. The results support the assertion that with experience users' mental models approach the system model. However, even experienced users had significant deviations from the system model, leading to predictions of where experts would have difficulty using the system and suggestions for improving usability.