Exploring temporality in learning analytics: a comprehensive framework for temporal educational studies

With the increasing focus on the temporal aspects of learning as an emerging area of study in education research, there is a new need for a guiding framework to help researchers and practitioners navigate in this area. Such a framework can provide clarity, mitigate redundancy, and offer a systematic way to approach and handle the challenges of temporal analytics. This thesis makes several contributions to the expanding field of temporal analytics, and thus, learning analytics. It encompasses two main contributions: a comprehensive mapping study and a systematically developed framework. Additionally, it offers two auxiliary contributions in the form of follow-up case studies. The research starts with a systematic mapping study, categorizing elements of published research to understand current trends that focus on the temporal dimension in educational studies. By employing thematic coding, the study analyzed patterns in temporal research, including the theme of research questions being asked, data collection with different granular levels, analytical techniques employed, and the derived learning insights. Building upon the mapping study, the thesis introduces a novel framework for temporal analytics. This framework aids researchers in data design, feature engineering, and analysis phases of temporal research. It also includes a set of guiding questions for each stage, enhancing the process of conducting temporal studies. To demonstrate the framework's applicability, two case studies were conducted. The first explores the temporal nature of students' self-regulated learning (SRL) behaviors. revealing various temporal patterns in SRL phase transitions. The second case study investigates temporality in students' posting behaviors in online discussions. It underscores the influence of visualization feedback on discussion engagement, particularly in motivating students and fostering behavioral change. Both case studies evaluate and exemplify the framework's effectiveness in providing a richer understanding of learners' temporal behaviors, going beyond traditional correlational or cross-sectional methods. The insights from these studies offer practical implications for designing effective SRL feedback systems and online discussion dashboards, illustrating the framework's adaptability across various learning contexts. Overall, the thesis offers a comprehensive approach to temporal data analysis. Unlike existing models that focus on specific aspects, this framework offers a broader, more inclusive perspective, making it versatile for a wide range of studies in the field. The thesis also addresses the framework's limitations and potential areas for improvement.