Adaptive Learning Analytics: Understanding Student Learning Behavior and Predicting Academic Success
2019-08-06T00:00:00Z (GMT) by
A literature search demonstrates a strong interest in adaptive learning technology (ALT), from notable authors, foundations, professional institutions, commercial enterprises, and government agencies (Healy 2015, Levy 2015, Johnson and Samora 2016, Mavroudi, Giannakos, and Krogstie 2018). As with any potentially effective teaching tool, ALT needs to be designed and implemented with the knowledge, skills and aptitudes of its recipients in mind. The purpose of this study was to examine the impact of instructional technology on student learning, using an adaptive learning system called Realizeit. In this study, the Realizeit Adaptive Learning Platform is referred to interchangeably as the Adaptive Platform or RALP. In addition to comparing student learning performance under traditional (face-to-face) and adaptive-mediated instruction, I applied Learning Analytics (LA) methodology to provide an in-depth analysis of individual student profiles, addressing different demographic, and academic and RALP-related variables to identify behavioral patterns during student learning. This study demonstrates how LA can serve as a methodological foundation for studying the impact of instructional technology on learning, as well as providing actionable recommendations based on the linear regression prediction models that resulted from the data analysis. Track dynamic data from the adaptive platform, combined with static (demographic) and semi-static (prior academic) data from the Student Information System, constituted the main data source for LA in this study. In addition to studying data generated from the adaptive platform, self-reported data from students, using a modified validated survey, provided valuable perspectives regarding students’ perspectives about their learning experiences with the adaptive platform. The results of this study provide empirical evidence that an adaptive learning intervention can have a significant impact on student learning performance. The use of digital traces shed light on other important aspects of the learning process, such as effective learning strategies, amount and type of effort, and engagement processes and time-management skills. Although study variables that reflected prior academic performance (student cumulative grade point average and pre-test score) were used as control variables, past education success was found to be the most significant predictor of success. Student self-reported data, especially, demonstrated the need to carefully consider the type, amount, and quality of education resources available to students to ensure that they are relevant to their perceived learning needs. Students’ responses also indicated that certain aspects of the adaptive platform, with regards to its design and technical issues, need further improvement. I conclude this study with actionable recommendations intended for course stakeholders, based on interpretation of the predictive model outputs coupled, with analyses of students’ responses.