From learning to e-learning: mining educational data. A novel, data-driven approach to evaluate individual differences in students’ interaction with learning technology
In recent years, learning technology has become a very important addition to the toolkit of instructors at any level of education and training. Not only offered as a substitute in distance education, but often complementing traditional delivery methods, e-learning is considered an important component of modern pedagogy. Particularly in the last decade, learning technology has seen a very rapid growth following the large-scale development and deployment of e-learning financed by both Governments and commercial enterprises. These turned e-learning into one of the most profitable sectors of the new century, especially in recession times when education and retraining have become even more important and a need to maximise resources is forced by the need for savings. Interestingly, however, evaluation of e-learning has been primarily based on the consideration of users’ satisfaction and usability metrics (i.e. system engineering perspective) or on the outcomes of learning (i.e. gains in grades/task performance). Both of these are too narrow to provide a reliable effect of the real impact of learning technology on the learning processes and lead to inconsistent findings. The key purpose of this thesis is to propose a novel, data-driven framework and methodology to understand the effect of e-learning by evaluating the utility and effectiveness of e-learning systems in the context of higher education, and specifically, in the teaching of psychology courses. The concept of learning is limited to its relevance for students’ learning in courses taught using a mixture of traditional methods and online tools tailored to enhance teaching. The scope of elearning is intended in a blended method of delivery of teaching. A large sample of over 2000 students taking psychology courses in year 1 and year 2 was considered over a span of 5 five years, also providing the scope for the analysis of some longitudinal sub-samples. The analysis is accomplished using a psychologically grounded approach to evaluation, partially informed by a cognitive/ behavioural perspective (online usage) and a differential perspective (measures of cognitive and learning styles). Relations between behaviours, styles and academic performance are also considered, giving an insight and a direct comparison with existing literature. The methodology adopted draws heavily from data mining techniques to provide a rich characterisation of students/users in this particular context from the combination of three types of metrics: cognitive and learning styles, online usage and academic performance. Four different instruments are used to characterise styles: ASSIST (Approaches to learning, Entwistle), CSI (Cognitive Styles Inventory, Allinson & Hayes), TSI (Thinking Styles Inventory and the mental self-government theory, Sternberg) and VICS-WA (Verbal/Imager and Wholistc/Analytic Cognitive style, Riding, Peterson) which were intentionally selected to provide a varied set of tools. Online usage, spanning over the entire academic year for each student, is analysed applying web usage mining (WUM) techniques and is observed through different layers of interpretation accounting for behaviours from the single clicks to a student’s intentions in a single session. Academic performance was collated from the students’ records giving an insight in the end-of-year grades, but also into specific coursework submissions during the whole academic year allowing for a temporal matching of online use and assessment. The varied metrics used and data mining techniques applied provide a novel evaluation framework based on a rich profile of the learner, which in turn offers a valuable alternative to regression methods as a mean to interpret relations between metrics. Patterns emerging from styles and the way online material is used over time, proved to be valuable in discriminating differences in academic performance and useful in this context to identify significant group differences in both usage and academic performance. As a result, the understanding of the relations between e-learning usage, styles and academic performance has important practical implications to enhance students’ learning experience, in the automation of learning systems and to inform policymakers of the effects of learning technology has from a user and learner-centred approach to learning and studying. The success of the application of data mining methods offers an excellent starting point to explore further a data-driven approach to evaluation, support informed design processes of e-learning and to deliver suitable interventions to ensure better learning outcomes and provide an efficient system for institutions and organization to maximise the impact of learning technology for teaching and training.
Showing items related by title, author, creator and subject.
Matcha, Wannisa (The University of Edinburgh, 2020-07-28)Learning has expanded beyond formal education; yet, students continue to face the challenge of how to effectively direct their learning. Among the processes of learning, the selection and application of learning tactics ...
Campbell, Katherine (The University of Edinburgh, 1999)This thesis examines how adults learn to play Scots fiddle. It focuses on a group of adults who are members of an intermediate-standard fiddle class in the city of Edinburgh, Scotland. The class is part of the wider ...
Sentential Cross-Situational Learning: The Combined Efforts Of Statistical And Syntactic Cues In Word Learning Panamsky, Lilia (The University of Edinburgh, 2011-11-23)Recently, cross-situational word learning (XSL) has gained attention as a viable mechanism for learning the meaning of words. XSL refers to the tracking of wordreferent pairs over a series of different exposures (Siskind, ...