Measuring Learnability in Human-Computer Interaction

It is well accepted that learnability is a crucial attribute of usability that should be considered in almost every software system. A good learnability leads within a short time and with minimal effort to a high level of proficiency of the user. Therefore, expensive training time of complex systems is reduced. However, there is only few consensus on how to define and evaluate learnability. In addition, gathering detailed information on learnability is quite difficult. In todays books on usability evaluation, learnability gets only few attention, research publications are spread to several other fields and the term learnability is also used in other context.

The objective of this thesis is to give an structured overview of learnability and methods for evaluation and additionally assist in the evaluator’s individual choice of an appropriate method. First of all, several definitions of learnability are discussed. For a deeper understanding psychological background knowledge is provided. Afterwards, methods to asses learnability are presented. This comprises nine methods that seem particularly appropriate to measure learnability. As this methods are very diverse, a framework based on analytical hierarchy process is provided. This framework aims to classify presented methods with respect to certain criteria and assess practitioners in selecting an appropriate method to measure learnability.