Engaging effective learning in a large-scale open online course : a self-regulated perspective from xMOOC

Abstract

This thesis explores learning engagement from the view of self-regulatory learning (SRL) among the participants of Massive Open Online Courses (MOOCs).

First, it investigates the SRL phases (forethought, performance, and self-reflection), sub-processes (self-motivational beliefs, task analysis, self-observation, self-control, self-judgment, and self-reaction), and learning strategies (e.g., goal-setting, task strategies, and causal attribution), focusing on the aspects of affection, behavior, cognition, and context. A fine-grained SRL-engagement framework is then developed based on previous models to probe the indicators of those four aspects of SRL processes.

Second, the thesis attempts to detect online learning behavioral indicators that influence the learning effectiveness of MOOC participants. Two studies were conducted: one qualitative, to demonstrate a comprehensive framework of SRL-engagement; and another quantitative, to evaluate behavioral patterns and learning sequences associated with learning outcomes.

In Study One, 164 MOOC learners participated in in-depth email interviews to report their SRL processes, in order to develop a strategy-aspect SRL framework. The interview script was adopted from a previous study (Zimmerman, 2000). In the light of the framework approach, it was possible to discover specific learning strategies to construct the fine-grained strategy-aspect SRL model based on a predefined framework. This framework was the phase-area SRL model (Pintrich, 2004).

The results of Study One found all aspects of engagement to be reflected in all three SRL phases. The sub-processes and strategies of SRL were indicated by at least one aspect of engagement. According to the descriptive statistics, the completers reported more SRL strategies from the behavioral aspect, such as strategic planning (81%), than the non-completers (59%). The completers also reported more SRL processes from the affective aspect, such as causal attribution (85%), than the non-completers (73%).

Surprisingly, some specific SRL strategies and SRL processes were mentioned more frequently by the non-completers than the completers. Task strategies were mentioned by 28% of completers and 47% of non-completers. Metacognitive monitoring was mentioned by 44% of completers and 65% of non-completers. The underlying rationales were discussed.

Study Two involved over 5,800 participants in an edX MOOC. Clustering analysis and sequential analysis were carried out based on their clickstream data. Using data from three major activities (video-watching, quiz-taking, and forum discussions), four clusters of participants were identified: (1) Drop-outs, (2) Learning-oriented learners, (3) Performance-oriented learners, and (4) All-rounders. Behavioral and sequential patterns differed markedly between participants of the four clusters. Participants demonstrated different patterns of self-controlled learning pace, information-seeking from videos, test-driven studying, scrutiny, and lurking in forums.

This study offers a theoretical contribution to the characterization of SRL processes and learning strategies by indicating the four aspects of engagement. It expands the role of SRL, and provides evidence that self-regulatory engagement can support learning effectiveness. The results, obtained from in-depth email interviews and data mining, have educational implications for policy-makers, educators and designers involved in large-scale open online learning.