A Statistical Analysis of Engagement in Arabic Language MOOCs

Thumbnail Image
Journal Title
Journal ISSN
Volume Title
As massive open online courses (MOOCs) gain popularity as an educational resource with minimal time, space and fee restrictions, many researchers invested in studying the engagement of learners with MOOCs. Such studies have enabled the identification of learners’ general characteristics and engagement patterns, which can help the MOOC’s providers to better understand the learners’ needs and accordingly enhance their learning experiences. In 2013, MOOCs reached the Arabic region when two MOOC’s platforms launched, Rwaq (in Saudi Arabia) and Edraak (in Jordon). They are now considered the leading Arabic MOOC platforms with more than 400 courses and over 70,000 daily visitors, combined. However, Arabic MOOC providers are only focusing on launching new platforms and delivering MOOCs that cover many subjects. The field of MOOC research lacks investigation of the adoption of MOOCs in the Arabic region. This might prevent the improvement of the learning experience of the Arabic-speaking learners. This research aims to study the adoption of MOOCs in the Arabic-speaking region. This can be achieved by: (1) identifying Arabic-speaking learners’ characteristics, (2) analysing their engagement with MOOCs’ contents, and (3) identifying differences and similarities in learner’s engagement between Arabic- and English-speaking learners. As the subjects of many studies, the leading MOOC platforms in the Western world have a good understanding of their learners and their needs. Following their lead, this research uses two fundamental works in the area of learner engagement with MOOCs. These are: Kizilcec et al. (2013) and Ferguson and Clow (2015) who analysed learner engage- ment with MOOCs in the Coursera and Futurelearn platforms, respectively. Kizilcec et al. (2013) introduced a classification method to classify learners based on their weekly interaction with the MOOC contents. They used two variables to compute learners’ weekly score, videos and assessments. Then they used these scores in a one-dimensional K-means clustering algorithm, which clustered their learners into four groups. This classification method was modified by Ferguson and Clow (2015) and applied to the Futurelearn platform. They included one more variable to compute learners’ weekly score, weekly comments. Then they used these scores in a multidimensional K-means clustering algorithm to prevent losing useful data when composing the weekly scores into a singular digit for the one-dimensional K-means clustering algorithm. Their classification method clustered their learners into seven groups, two of which are found in Kizilcec et al.’s result. We applied both classification methods (one- and multidimensional K-means clustering algorithm) to our data that we obtained from the Edraak platform. We found that the learner engagement results were similar. Each classification method produced the same three groups that represent Edraak’s learners’ engagement types, which are the following: • Sampling: represents learners who had no interaction with the MOOC contents or dropped out of the MOOC after the first week. • Disengaging: represents learners who had a good interaction with the MOOC contents at the beginning of the MOOC, but had dropped out by the middle of the course. • Completing: represents learners who had a good interaction with the MOOC contents, and completed the MOOC. Comparing our results with Kizilcec et al.’s (2013) and Ferguson and Clow’s (2015) showed that the engagement of Edraak’s learners with MOOCs is closer to Coursera’s learners than Futurelearn’s learners. Edraak’s and Coursera’s learners have a lower completion rate than Futurelearn’s learners, in addition to the poor use of discussion forum in Edraak’s and Coursera’s platforms compared to Futurelearn’s platform. Moreover, considering pedagogical approaches adopted by these platforms, we found that th