The Role of User Models in CAT: Exploring Adaptivity Variables
Evangelos Triantafillou, Elisabeth Georgiadou and Anastasios Economides
This paper reviews all different variables that have been used in adaptive educational systems and then discuss their potential use to a hypothetical user model for Computer Adaptive Testing. From all the variables presented that triggers adaptation, modelling of ‘knowledge on the domain presented’, ‘background-experience’, ‘preferences’, ‘personal data’ and ‘mental models’ can produce more efficient CATs in terms of time, as fewer items will be needed to assess performance. Moreover, it could affect items’ quality, since items can be more complex taking into account user characteristics, resulting in testing sessions that can contribute to the learning process and not merely arrange examinees on a problem complexity scale.