Abstract:
It is vital to understand students' Self-Regulatory Learning (SRL) processes, especially in Blended Learning (BL), where students need to be more autonomous in their learning processes. We are seeing higher student enrolment rates with students who come from different backgrounds, have different knowledge bases, and need different things from higher education, we are also observing many dropouts. Therefore, it is essential to support students in a timely manner. Due to the nature of BL environments, the lecturers do not have close relationships with students to be able to take precautionary measures. Thus, it is vital to give insights to lecturers to help students. Learning Analytics (LA) allows us to analyse, understand, and optimise learning processes. This analysis needs to be grounded in learning theories. Currently, LA lacks studies that have a theoretical foundation and are based on empirical evidence. Furthermore, students' motivational studies have not yet been sufficiently considered for analysis in LA. Therefore, this study used Activity Theory (AT) and SRL theory to understand how students' perceptions, motivations, and learning strategy use inform LA. The aim of LA in general and this thesis, in particular, is to help the learning process. In LA, students are central to the analysis. Accordingly, we collected both quantitative and qualitative data from two groups of students (freshmen and upper-level). We collected quantitative data by running the Motivated Strategies for Learning Questionnaire (MSLQ) three times in three 12-week courses (N=419). We also collected qualitative data by interviewing 42 students. Two quantitative studies (five papers) and one qualitative study (three papers) are included in the thesis. This thesis, by running three studies based on learning theories, added empirical evidence to LA and contributed to its theoretical foundation by linking LA with SRL and AT. This study also contributes to LA by focusing on the students’ conditions (motivation and learning strategy use) mentioned in Winne’s version of SRL (COPES model) and exploring the level of agency (students’ perceptions regarding tool use), which are currently lacking in the field. This thesis also brings empirical evidence to LA and informs theory and practice. Through predictive and cluster analysis, the study contributed to one of the most important aims of LA, identifying at-risk students. By identifying the constructs that help us predict students’ final scores through stepwise regression analysis early in the course, the lecturer can apply appropriate interventions in order to help students and prevent dropouts. Identifying constructs that have the highest correlation with the final score, the lecturer could promote them in the course. Also, by identifying different SRL profiles through applying the K-Means clustering algorithm and examining students' SRL profile adaptation longitudinally the study contributed to SRL theory and addressed the challenge identified regarding the cyclical nature of SRL. The study's aim was not only to identify at-risk students or students’ SRL profiles but also to use data to improve the learning process and support personalised learning. For this reason, learning theories, including AT and SRL, were applied to understand students' perceptions regarding the usefulness of tools through applying thematic analysis. We contributed to AT by identifying contradictions in students’ perceptions regarding using educational tools in classes for their learning process and the changes that needed to be applied to educational settings. We also contribute to SRL when we analysed students’ perceptions regarding how each tool supported a specific stage of SRL and tried to open Winne’s black box.