Using Data Science to Increase Student Success in Online Education
Concurrent Session 4
Data has been called “the new oil” and online education generates substantial data reserves. We will discuss several applied research projects using data science for student success, including qualitative studies and statistical modeling of social-emotional learning. The results indicate not only what *can* be done, but what *should* be done.
As academic technologies mature and are more deeply integrated into higher education, significant ‘data exhaust’ from student and faculty use of these applications provides a rich set of behavioral data that was not previously available. Over the past decade, data science techniques have been applied to this data and have generated surprising insights. What are some of the most promising results that we’ve seen from this work? What ethical implications are created for institutions, faculty, and practitioners involved in this work?
In this presentation, we will discuss the state of the field from the perspective of a data science leader directing this work and an academic technology leader planning and implementing this work on his campus. We will provide a brief overview of the field then discuss several studies and deployments of learning analytics that we have collaborated on and/or led from our respective roles.
The studies that we will discuss include:
a. Stealth Assessment of Social and Emotional Skills using Learning Analytics – we analyzed data from student use of the Learning Management System (LMS) in a large-enrollment undergraduate chemistry course and examined the relationships among social and emotional (SE) skills, as measured by ACT® Tessera®, course features, and course grades. Our goal was to understand how SE skills would impact students’ online behaviors and course outcomes and to explore a principled design approach for feature extraction using individual activities and sequential data mining techniques. Findings suggest a new source of insights that can be used to create accurate and actionable predictions to improve student outcomes. No bias based on demographic background was observed in the predictive models.
b. Learner Facing Analytics: Analysis of Student Perspectives from Australia. Within the literature there has been a focus on institutional and academic views related to learning analytics or in fact, what academics in the field feel would be useful to students. This study sought to directly include student perspectives in the conversation around analytics. Specifically, the study sought to (a) explore student understandings and concerns in relation to learning analytics; (b) gather student input on the types of learning analytics reports, dashboards and tools that will be most useful in supporting student success; (c) develop a series of principles to guide institutions in the creation of student-facing dashboards; and (d) identify the processes and training required to support students and staff to make sense of the data presented in dashboards and improve student success.