PART I
Introduction
This section introduces a framework for analyzing questions and issues pertaining to the responsible and purposeful implementation of data mining and data analytics in education. This framework will serve as a unifying structure for the book. To put the book into the appropriate context, this section also addresses the distinctions and similarities between data analytics, data mining, educational data mining, machine learning, and learning analytics, and provides an overview of the historical and theoretical perspectives of data analytics and data mining in education. The section ends with a conceptualization and institutionalization of the notion of responsible learning analytics in education.
Chapter 1 Introduces Khanâs Learning Framework, initially developed in 1997 to address issues pertaining to the successful implementation of e-learning in education and training settings. The E-Learning Framework (2001) has evolved to encompass the design and delivery of effective, efficient, and engaging learning across multiple learning environments and contexts. This chapter introduces the eight dimensions of Khanâs framework and provides a justification for its use as a structure for the organization of the book.
Chapter 2 provides a comprehensive and unified view of data analytics and educational data mining, and how they are used to improve student learning and engagement. The authors discuss the distinctions and similarities between data analytics, data mining, educational data mining, machine learning, and learning analytics. They describe the six functional facets of data analyticsâdescriptive, diagnostic, predictive, prescriptive, visual, and cognitive, and conclude with an overview of machine learning and data mining approaches as a backdrop for educational data mining and learning analytics.
Chapter 3 provides an overview of the historical and theoretical perspectives of data analytics and data mining in education. To capture the effects of data in educational research, key events and works are examined through a timeline that focuses on educational data mining and learning analytics as interconnected research domains. The theoretical discussions within the field are examined, building on the evolution of the field.
Chapter 4 examines the notion of responsible learning analytics. Such an approach requires institutions to take an answerable and response-able approach to using student data. Learning analytics, like all educational technology, is shaped by a range of social, political, economic, and cultural agendas. As such, this chapter argues that ethical and responsible learning analytics is central to a consideration of each of the eight dimensions of Khanâs E-Learning Framework (Khan, 2010). The chapter concludes by suggesting a number of pointers for higher education institutions to ensure the adoption of a responsible and responsive approach to learning analytics.
1
A FRAMEWORK FOR IMPLEMENTING RESPONSIBLE DATA MINING AND ANALYTICS IN EDUCATION
Maria Elena Corbeil, Joseph Rene Corbeil, and Badrul H. Khan
Introduction
According to Kumar (2013), there is a continuous stream of data being collected from the digital activities we engage in every day. This is no different in regard to learning. Data are collected about learnersâ activities and performance from learning management systems used for course delivery, as well as other data collected throughout the educational process, such as demographics and previous academic performance. Kumar adds that today, those data can be valuable resources for identifying patterns and making predictions for improved decision making and providing of learning resources. According to Bainbridge et al. (2015), using simple learning analytics models, educators now have the tools to identify, with up to 80% accuracy, which students are at the greatest risk of failure before classes even begin. However, as we consider the enormous potential of data mining and data analytics in education, we must also recognize a myriad of emerging issues and potential consequences, intentional and unintentional, to implement them responsibly. For example:
⢠Who collects and controls the data?
⢠Is it accessible to all stakeholders?
⢠How are the data being used, and is there a possibility for abuse?
⢠How do we assess data quality?
⢠Who determines which data to trust and use?
⢠What happens when the data analysis yields flawed results?
⢠How do we ensure due process when data-driven errors are uncovered?
⢠What policies are in place to address errors?
⢠Is there a plan for handling data breaches?
This chapter provides an introduction to Khanâs Learning Framework for analyzing educational issues (based on Khanâs E-Learning Framework, 2001) to guide policy makers, administrators, faculty, and IT personnel with a framework for thoughtfully and systematically analyzing issues pertaining to: how data analytics can be used to improve the quality of courses and programs; technological and resource support, including networking capacity and infrastructure; accessibility and use; institutional and management issues related to ownership of data and decisions on how the data and their interpretations may impact students, faculty, and the institution; the quality and reliability of data, as well as accuracy of data-based decisions; and, ethical implications surrounding the collection, distribution, and use of student-generated data.
A Framework for Analyzing Educational Issues
When analyzing the myriad of complex issues related to the responsible implementation of data mining and analytics in education, it is helpful to use a systematic method, such as the one provided by Khanâs Learning Framework, as illustrated in Figure 1.1 below. The dimensions of the framework address by category, the major issues raised in the execution of a new initiative or program.
Dr. Badrul H. Khan, educator, researcher, and pioneer of web-based instruction, first developed the E-Learning Framework in 1997 (Khan, n.d.). âKhanâs framework remains a valuable tool for evaluating an organizationâs educational technology readiness and opportunities for growth. It helps stakeholders think through every phase of a new initiative to ensure that desired learning outcomes are achievedâ (Khan, Corbeil, & Corbeil, 2016, âA Framework for Analyzing Educational Issuesâ, para. 2). As a result, since its publication over 20 years ago, the framework has been used around the world to guide the purposeful design, development, and delivery of highly effective online learning solutions in a variety of fields, including education and business.
One of the benefits of the Learning Framework is that it is suitable for both large-scale and small-scale implementation of responsible data mining and analytics in education. As Macfadyen (2017, p. 31) noted, âWhile many schools are currently gripped by analytics panic . . .,â the implementation of a big data plan may not be feasible for all institutions. She recommends,
(p. 31)
Khanâs Learning Framework provides a structure and guiding methodology for stakeholders at all levels of education, whether large or small, experienced in data mining or not, to meaningfully utilize data and learning analytics for making informed education and training decisions.
The Eight Dimensions of Khanâs Learning Framework
The following sections describe the eight dimensions of Khanâs Learning Framework.
Pedagogical
The pedagogical dimension addresses issues pertaining to how instructional content is designed, delivered, and implemented, with a strong emphasis on the identification of learnersâ needs and how the learning objectives will be achieved. This dimension addresses the delivery method for course activities and the appropriateness of the learning environment for achieving the goals of its intended audience. It also concerns issues pertaining to the role of data analytics and data mining in learning, their possibilities and limits, and how educators can mine legal and meaningful data sources to make informed instructional decisions.
In Chapter 6 of this volume (pp. 101â118), Moore points out that stakeholders at all levels of education can benefit from the data that are already collected and analyzed to help inform decisions regarding the highly complex aspects of student learning, educational programming, and assessment, to name a few. Tufan and Yildirim (Chapter 3, this volume, pp. 43â62) agree, adding that learning analytics, supported by current data mining and instructional technologies, also allow both educators and students to create, and participate in, more interactive, personalized, and adaptive learning environments.
In Chapter 5 of this volume (pp. 83â100), Voithofer and Golan offer numerous examples of this. For instance, they highlight the ways in which the data available from the learning management system, such as the time of day, frequency, and duration of students log ins, can be harnessed to identify patterns that can inform, as recommended by the pedagogical dimension, how the learning goals are being achieved by the intended audience.
Similarly, Harasim (this volume) provides an example of how learning analytics can be applied by instructors in class. In Chapter 7 (pp. 119â137), she delineates how, framed within the Collaborativist Theory of Learning, data collected from discourse analyses of online discussion forums can be used to make the learning process evident to both students and instructors, helping to inform how instructional content is designed, delivered, and implemented, one of the most important aspects of the pedagogical dimension.
Technological
The technological dimension is concerned with the learning environment, its creation, and the tools required to deliver the learning program. This dimension also addresses hardware and software requirements, as well as infrastructure planning. Technical requirements, such as server capacity, bandwidth, security, backups, and other infrastructure issues are also addressed. This is a key consideration in regard to data mining and analytics, as the technological dimension also addresses networking infrastructure issues relating to data volume and transmission.
The technological dimension guides administrators, IT professionals, and educators at all levels of education regarding the technical tools and considerations required for the safe and responsible implementation and maintenance of data mining and learning analytics.
Along those lines, in Chapter 2 (this volume, pp. 16â42), Gudivada, Rao, and Ding observe that rece...