| dc.description.abstract | Massive Open Online Courses (MOOCs) are a highly popular, relatively new phenomenon in the world of online education. However, due to their massive scale they encounter many problems such as low completion rates and a lack of personalised learning at scale. Several reasons for low completion rates are discussed in literature, among which are course difficulty and course workload. In addition, some authors have suggested adaptive learning as a solution for personalised learning. However, little research has been performed towards the combination of MOOCs and adaptive learning.
Therefore, this research project aimed to investigate the effect of adaptive learning on learner satisfaction, learner engagement, and ultimately completion rates in the context of MOOCs. It was hypothesised that adaptive learning would increase all three variables. To investigate this, an online adaptive learning system was designed using an approach based on the principles of design science. The designed system was then applied in practice in a case study. The designed system represented a challenge-based MOOC on cyber security and contained three experimental conditions: a randomised condition to calibrate the adaptive system, a linear condition, and an adaptive condition. The system collected quantitative usage data as well as qualitative data by means of two surveys.
The system was deployed in practice at five educational institutions. A total of 156 users registered on the system, of which 131 users participated actively. Adaptive learning was found to significantly reduce learner dropout and completion time when compared to the linear condition. However, adaptive learning also significantly reduced learner satisfaction. No significant effect of adaptive learning on engagement or completion rates was found. Additionally, no effect of satisfaction or engagement on completion rates was found, though satisfaction and completion were found to be highly correlated. It is concluded that implementing adaptive learning in MOOCs is a viable option for MOOC developers, but that it depends on the mission and context of a specific MOOC whether this is desirable. Concrete recommendations to support decision-making are provided. | |
