Khan, Madiha Azeem; (2024) A learning analytics approach towards monitoring coregulation by human tutors, in a virtual classroom environment. Doctoral thesis (Ph.D), UCL (University College London).

Preview

Text Khan_10196621_Thesis_sigs_removed.pdf Download (29MB) | Preview

Abstract

This thesis explores how learning analytics can be used to monitor the co-regulation of upper primary school-aged learners, by human tutors in a virtual classroom environment. Co-regulation is an important route through which young learners acquire SRL skills. However, limited research has been conducted towards understanding how co-regulation occurs in practice. The proliferation of online learning environments offers new research opportunities that discover favourable and maladaptive co-regulation patterns. Tracking patterns of co-regulation using the data generated from online environments can enable timely interventions to support co-regulation, boosting the development of learner SRL In this thesis, I propose a theoretically grounded approach towards using learning analytics to monitor co-regulation in a Virtual Classroom Environment (VCE). An established model of SRL is contextualised to enable an investigation into how human tutors co-regulate learners in a VCE. A framework of signifiers is designed using a mixed methods approach before being implemented through a combination of learning analytics for datasets comprising high and low-performance tutors. Process mining is implemented to identify holistic and structural features of tutor-learner interactions, while pattern mining and decision trees are implemented to identify statistically significant differences. I experiment with using average progress scores as an alternative criterion for tutor performance to evaluator rankings for student-centred teaching, to explore the differences in the results obtained. My results show that high-performing tutors are more likely to use a diverse and complex range of tutoring practices which extend the learning activities set out in the VCE. Open-ended forms of engagement that encourage learner participation and agency feature prominently. On the other hand, low-performing tutors tend to use a more restricted range of tutoring practices which follow the script of the VCE platform more closely and rely on directive forms of engagement. There are similar differences between the co-regulatory practices of high versus low-performing tutors, regardless of the performance criterion used. However, when average progress scores are used as the criterion for tutor performance, the differences between the co-regulatory practices of high and low-performing tutors are less stark. While co-regulatory practices are a strong predictor of tutor performance when evaluator rankings are used as the performance criterion, they are not able to reliably predict tutor performance to attainment. My work has a range of operational implications, including the development of intelligent approaches towards tutoring training and evaluation, and timely scaffolding of tutor practices to co-regulation.

Download activity - last month

Export as XMLJSONCSV

Download activity - last 12 months

Export as XMLCSVJSON

Downloads by country - last 12 months

Export as JSONCSVXML

Archive Staff Only

View Item