Learning Analytics Explained

Niall Sclater

• Publisher: Routledge
• Year: 2017
• ISBN:
  9781317394556
• Edition: 1
• Topic: Education
• Subtopic: Education General

Learning Analytics Explained draws extensively from case studies and interviews with experts in order to discuss emerging applications of the new field of learning analytics. Educational institutions increasingly collect data on students and their learning experiences, a practice that helps enhance courses, identify learners who require support, and provide a more personalized learning experience. There is, however, a corresponding need for guidance on how to carry out institutional projects, intervene effectively with students, and assess legal and ethical issues. This book provides that guidance while also covering the evolving technical architectures, standards, and products within the field.

Part I

Background

1 The Evolution of a New Field

Online and mobile technologies are facilitating the accumulation of vast amounts of data across business, industry, government and other areas of human endeavour. We have become dependent on the internet as a source of information for most of what we need to know, for access to entertainment, for communication with others, for purchases and banking and for carrying out our work. As our lives become increasingly intertwined with technology, we are generating huge quantities of ‘digital exhaust’ through the use of internet-connected extensions of ourselves: devices such as smartphones, tablets, laptops, e-book readers and fitness trackers.

The use of analytics to process and help interpret this data is enabling organisations to develop better insight into people’s activities and to optimise organisational processes and outputs. Business intelligence, as this area is often known, is now essential for the survival and expansion of commercial organisations and an increasingly essential tool in other sectors. Insurance providers, for instance, offer discounts to customers prepared to install devices in their cars that monitor the safety of their driving. Meanwhile, in medicine, analytics identify the spread of disease across populations and help agencies to target interventions more rapidly.

In universities and colleges, data has until recently often been stored on paper in filing cabinets and in a wide variety of different formats,¹ sometimes residing on spreadsheets on a teacher’s or administrator’s machine. However, large, more easily accessible datasets increasingly exist about learners, their learning activities and the environments in which they study. Educators are beginning to understand how to exploit this to help solve some of the challenges faced by their institutions. Analytics about learning promises to enhance many aspects of the educational experience through the use of data about students and their learning contexts.

This chapter outlines how the field of learning analytics is evolving: what the drivers are for institutions and which existing fields of enquiry are contributing to it. I discuss some of the definitions for learning analytics and to what extent it can be differentiated from other closely related fields. Finally, I look at what the main current focusses of investigation are and how the worldwide community of researchers, practitioners and vendors is (at least sometimes) working together to build the new domain of learning analytics.

The Opportunities Presented by Big Data and Analytics

Access to the internet has become an important aspect of education – an essential part of it if learners are to become effective citizens and employees in a world organised digitally. Institutions offer many courses in blended or fully online modes and it is increasingly unlikely that a university student will complete a degree programme without carrying out some studies using web-based tools. The internet has encouraged the growth of for-profit online colleges in the US,² and massive open online courses (MOOCs) provide online education to millions more learners globally.

As students navigate online learning systems, they leave ‘digital footprints’, or traces of their activities. Clearly, the more learning that takes place online, the more data is likely to be accumulated. Meanwhile, other aspects of student activity are being recorded continually, such as their presence on campus, their attendance at lectures, their use of library facilities and their submission of assignments. These multiple sources of information enable the development of a much richer view of student behaviour than has been possible before, helping institutions to identify opportunities to improve courses, to personalise the study experience for individual learners and to assist those identified by the systems as academically at risk.

Data-Informed Decision Making

The growth of big data – datasets that are beyond the ability of traditional software to capture, store, manage and analyse³ – is driving the development of the tools and methods of learning analytics.⁴ While there are problems with the reliability of some of this data, it can generally be collected cheaply⁵ or is already being gathered by the systems that the students are using. Analysing learners’ experiences of education has traditionally been carried out using questionnaires and interviews with limited numbers of students. Using the data accumulated by students during their normal study activities is less intrusive than these research methods and provides a more continuous, uninterrupted and complete picture of study activities.⁶

Traditionally, most of the decisions made in educational institutions by faculty, staff and administrators have been based on intuition, anecdotes or presumptions. Formulating a hypothesis and then attempting to prove or disprove it can be a time-consuming process, limited by the quality of the research question. Empirical educational research, until recently, has been designed to answer a single question: which of two approaches works better?⁷ Decisions on how to enhance education are likely to have better results if they are founded on data, facts and statistical analysis.⁸, ⁹ This is not to suggest that human qualities such as experience, expertise and judgement should be entirely replaced but they should be supplemented by analytical techniques where appropriate.¹⁰

As education moves increasingly online, teachers may lack the visual clues that helped them to identify students who were insufficiently challenged, bored, confused or who were failing to attend:¹¹ the use and interpretation of learning activity data thus becomes key for those teaching online. Data can help highlight issues in ways that were not previously possible and its use can encourage a philosophy of continuous improvement.¹² Patterns and trends can be identified and the merits of different options can be weighed up.¹³

Cooper suggests that analytics can help us to answer questions of information and fact, such as ‘What happened?’ or ‘What is happening now?’ and ‘Where are trends leading?’ He differentiates these from questions of understanding and insight such as ‘How and why did something happen?’ or ‘What should we do next?’ and ‘What is likely to happen?’¹⁴ Colleges and universities that fail to ask such questions and to learn from the increasingly valuable data being accumulated about their learners risk being left behind by more innovative institutions, which offer better and more personalised education to their students.

Pressures on Institutions

The need for more effective decision making at all levels of institutions is given impetus by a wide range of pressures affecting education, particularly greater student numbers. In an era of accountability and liability, there is a requirement for better measurement and quantification of many educational processes.¹⁵, ¹⁶ Institutional budgets are more and more stretched and there is a need for evidence-based prioritisation of spending. Meanwhile, students who pay for their education will increasingly and justifiably expect to be able to see evidence that their fees are being spent appropriately.

Simultaneously, there is increased government scrutiny of issues such as retention and the equality of educational provision for minority groups. In the US, high student attrition levels are a focus of the Department of Education, and a number of for-profit institutions have been targeted for poor retention rates and for receiving financial aid for students who subsequently withdraw.¹⁷ In the UK, the government-imposed Teaching Excellence Framework requires universities to measure aspects of their teaching provision in order to ensure quality, value for money and, ultimately, the employability of graduates.¹⁸ The concerns of Western governments that their economies are falling behind the rising nations of Asia is captured in an Australian review of higher education, which identifies the link between educational attainment and economic productivity. The document also sets targets for improving retention and completion rates for indigenous students and those from low socio-economic backgrounds.¹⁹

Poor retention rates affect institutions in numerous ways, as well as broader society and countries’ economic progress. The financial impact resulting from the loss of income from student fees, in particular, can be significant. The cost of recruiting the students may have been wasted and revenues from catering outlets and student accommodation can also be affected.²⁰ The personal impact on a student who has dropped out can, of course, be dramatic as well. Apart from feelings of failure and the loss of self-esteem, job prospects may be adversely impacted and debt may have been accrued that is never paid off. There are strong correlations between academic achievement and higher income levels.²¹ Graduates tend to have higher status jobs, be healthier and live longer. They may also be more likely to rate their former institutions highly, leading to improved reputation and enhanced recruitment possibilities.

Influences

Key contributing disciplines to learning analytics are computer science, education and statistics.²² Dawson and colleagues analysed contributions to the International Conference on Learning Analytics and Knowledge (LAK), for example, and discovered that approximately 51% of papers were from computer scientists whereas 40% of authors had a background in education.²³ Theory and methodologies are drawn from disciplines as varied as psychology, philosophy, sociology, linguistics, information science, learning sciences and artificial intelligence.²⁴ Learning analytics draws on web analytics to help make sense of the data created in log files of users’ access to websites.²⁵ Other methods deployed include social network analysis, predictive modelling and natural language processing – all fields of enquiry in their own right. Demonstrating the connections between learning analytics and other disciplines, for example, was an international conference in the well-established field of computer supported collaborative learning, which had nine papers, three posters and an invited session with ‘learning analytics’ in their titles.²⁶

Clow argues that this eclectic approach facilitates rapid development of the field and the ability to build on established work. However, it means that learning analytics currently lacks its own coherent epistemology and established approaches.²⁷ Contributors to the key conferences in learning analytics and associated areas are, though, increasingly using theories from learning sciences to guide their selection of methods.²⁸

A Confusing Mix of Disciplines and Terminology

Given the recent emergence of learning analytics as a field and its multidisciplinary origins, it is hardly surprising that there is confusion about what exactly it is and how it is differentiated from related areas. Analytics can refer to:

• a specific topic, such as health analytics;

• the aim of the activity, for instance predictive analytics;

• the data source, for example Google Analytics.²⁹

Analytics itself has been defined as:

the analysis of data, typically large sets of business data, by the use of mathematics, statistics and computer software.³⁰

Campbell et al. sugge...

Table of contents

• Cover
• Half Title
• Title Page
• Copyright Page
• Dedication
• Table of Contents
• Preface
• Acknowledgements
• List of Abbreviations
• Introduction
• PART I: Background
• PART II: Applications
• PART III: Logistics
• PART IV: Technologies
• PART V: Deployment
• PART VI: Future Directions
• Index