Logo: Deutsches Institut für Internationale Pädagogische Forschung



show results

Yue, Zaho:

Learning analytics technology to understand learner behavioral engagement in MOOCs

Delft : University of Technology (2019) , 164

URL des Volltextes:


1. Monographien (Autorenschaft); Monographie

Kurs, Open Education, E-Learning, Universität, Lernverhalten, Zeit, Lernprozess, Aktivität, Analyse, Leistungsbeurteilung, Verhaltensänderung, Lernumgebung, Mobile Computing, Video, Aufmerksamkeit, Feedback, Technologie, Empirische Untersuchung, Niederlande

As one of the most prominent examples of technology-enhanced learning, massive open online courses (MOOCs) have attracted extensive attention of learners, educators, and researchers since 2012. However, a low completion rate is a ubiquitous and severe problem in MOOCs, which means that only a small portion of learners got scores higher than or equal to the course requirements in MOOCs. Learner engagement is commonly presumed to be highly related to the completion rates of MOOCs. In traditional classrooms, learner engagement can be observed by experienced educators. They can keep learner engagement by adjusting the course content and the way they teach. However, educators cannot observe learner engagement in MOOC learning the same way they usually do in traditional classrooms, while many learners lack skills to keep their engagement by themselves, which leads to high dropout rates of MOOCs. To observe learner engagement in MOOCs and provide learners feedback about their learning progress, learning analytics technology has been used by educators and researchers on MOOC platforms. Learner engagement is usually investigated in three dimensions: behavioral engagement, emotional engagement, and cognitive engagement. In this thesis, we focus on using learning analytics technology to understand learner behavioral engagement in MOOCs. While many activities related to learner emotional engagement and cognitive engagement happen outside of MOOC platforms, most activities related to learner behavioral engagement are captured by the technology of MOOC platforms. Specifically, we study learner behavioral engagement on three time scales: throughout a course, in a learning session, and in a short period of time. First, we explore learner behavioral engagement throughout a course based on learning analytics technology with large-scale trace data. To investigate the change of learner behavior after clinching a passing grade, we define a set of pre-passing and post-passing behavior patterns in our study. We present a data-driven approach which analyzes trace data from four thousand learners whose scores met the course requirements and find a certain subset of learners who heavily reduced their engagement in question answering after clinching a passing grade. Our study suggests that the course structure and grading schema of MOOCs should be designed to assign a certificate to learners only when they display mastery of an entire course subject. Second, to investigate learner behavioral engagement in mobile learning sessions, we measure the impact of divided engagement and real-world environments on learner performance and interactions. We conducted a study which requires learners to have mobile learning sessions while sitting in the lab and walking on campus. To measure the impact of multitasking and divide attention, the trace data of learners and their answers in the questionnaire are analyzed. We find that learning on-the-go contributed to lowered learning performance and learners show different time arrangement in video watching and question answering while walking with learning. Third, we investigate learner behavioral engagement in a short period of time. Specifically, we focus on tracking learner attention during video watching. Many MOOCs are centered around video lectures and learners can easily lose their attention while watching videos. If learner inattention can be detected automatically and in real-time, interventions can be provided to MOOC learners once they are being disengaged. We first propose an eye-tracking based method and our lab study indicates that it is possible to deploy a large-scale application of the webcam-based inattention detection in MOOCs. To avoid a high detection lag, low accuracy, and the complexity of design and maintenance in the eye-tracking method, we propose another method with face-tracking. We deploy our face-tracking based inattention detection method as a widget IntelliEye in real MOOCs. Through the deployment of IntelliEye, we find that most learners have capable setups to run our widget and one-third of them are willing to use it. Based on analyzing learner trace data, we observe high levels of learner inattention and their adaption toward our attention tracking technology. (DIPF/Orig.)

Informationszentrum Bildung


last modified Nov 11, 2016