Application of Learning Curves for Didactic Model Evaluation: Case Studies

The success of (online) courses depends, among other factors, on the underlying didactical models which have always been evaluated with qualitative and quantitative research methods. Several new evaluation techniques have been developed and established in the last years. One of them is ‘learning curves’, which aim at measuring error rates of users when they interact with adaptive educational systems, thereby enabling the underlying models to be evaluated and improved. In this paper, we report how we have applied this new method to two case studies to show that learning curves are useful to evaluate didactical models and their implementation in educational platforms. Results show that the error rates follow a power law distribution with each additional attempt if the didactical model of an instructional unit is valid. Furthermore, the initial error rate, the slope of the curve and the goodness of fit of the curve are valid indicators for the difficulty level of a course and the quality of its didactical model. As a conclusion, the idea of applying learning curves for evaluating didactical model on the basis of usage data is considered to be valuable for supporting teachers and learning content providers in improving their online courses.