Data Analytics Applications in Education

Jan Vanthienen, Kristof De Witte

• Publisher: Auerbach Publications
• Year: 2017
• ISBN:
  9781351650182
• Edition: 1
• Topic: Business
• Subtopic: Government & Business
• Index: Business

The abundance of data and the rise of new quantitative and statistical techniques have created a promising area: data analytics. This combination of a culture of data-driven decision making and techniques to include domain knowledge allows organizations to exploit big data analytics in their evaluation and decision processes. Also, in education and learning, big data analytics is being used to enhance the learning process, to evaluate efficiency, to improve feedback, and to enrich the learning experience.

As every step a student takes in the online world can be traced, analyzed, and used, there are plenty of opportunities to improve the learning process of students. First, data analytics techniques can be used to enhance the student' s learning process by providing real-time feedback, or by enriching the learning experience. Second, data analytics can be used to support the instructor or teacher. Using data analytics, the instructor can better trace, and take targeted actions to improve, the learning process of the student. Third, there are possibilities in using data analytics to measure the performance of instructors. Finally, for policy makers, it is often unclear how schools use their available resources to "produce" outcomes. By combining structured and unstructured data from various sources, data analytics might provide a solution for governments that aim to monitor the performance of schools more closely.

Data analytics in education should not be the domain of a single discipline. Economists should discuss the possibilities, issues, and normative questions with a multidisciplinary team of pedagogists, philosophers, computer scientists, and sociologists. By bringing together various disciplines, a more comprehensive answer can be formulated to the challenges ahead. This book starts this discussion by highlighting some economic perspectives on the use of data analytics in education. The book begins a rich, multidisciplinary discussion that may make data analytics in education seem as natural as a teacher in front of a classroom.

Chapter 1

Introduction: Big Data Analytics in a Learning Environment

Contents

1.1 Data Analytics

1.2 Data Analytics in Education

1.3 How Has This Become Possible?

1.4 Why Data Analytics Has Become Important

1.5 List of Contributions

1.5.1 Data Analytics to Improve the Learning Process

1.5.2 Data Analytics to Measure Performance

1.5.3 Policy Relevance and the Challenges Ahead

References

The abundance of data and the rise of new quantitative and statistical techniques have created a promising area: data analytics. This combination of a culture of data-driven decision making and techniques to include domain knowledge allows organizations to exploit big data analytics in their evaluation and decision processes. Also, in education and learning, big data analytics is being used to enhance the learning process, to evaluate efficiency, to improve feedback, and to enrich the learning experience. Before discussing some possibilities and issues in the use of learning analytics in education, we define its concept.

1.1 Data Analytics

With data available in large quantities, data analytics refers to a set of techniques and applications to explore, analyze, and visualize data from both internal and external sources. Applications can range from business intelligence (BI), enterprise reporting, and online analytical processing (OLAP) to more advanced forms of analytics, such as descriptive, predictive, and prescriptive analytics. Descriptive analytics identifies relationships in data, often with the intent to categorize data into groups. Predictive analytics exploits patterns found in historical data to make predictions about future behavior. Prescriptive analytics suggests decision options and shows the implications of alternative decisions.

Data can be of many types, and may originate from many sources: internal transaction data, web data, location data, browsing behavior, driving behavior, government data, and so on. With the increased digitization of society, a wealth of data is ready to be explored. The data explosion, however, has created a gap between the volume of data generated and stored, on the one hand, and the understanding and decision making based on these data, on the other hand. Traditional analysis techniques such as query and reporting or spreadsheet analysis are unable to cope with the complexity and size of current data sources.

This is where advanced analytics comes in, generating automatic descriptions of the data in terms of (human-interpretable) patterns, and predictions of unknown or future values of selected variables using, for example, clustering, classification, or regression techniques. In data analytics projects, however, a lot of effort is still devoted to collecting and integrating the available data; data preparation and cleaning; building, testing, and refining models; and finally, communicating results or triggering actions.

Data analytics has huge potential and is changing the world. The technical and managerial issues resulting from the adoption and application of data science in multiple areas are worth exploring (Baesens et al., 2016).

1.2 Data Analytics in Education

Rogge et al. (2017) argue that data analytics applications and functionalities provide a broad range of opportunities in the public sector. Their review reveals that governments worldwide have announced plans and road maps to support the development of big data in both the public and private sector. Education economists, in particular, are increasingly using the availability of large datasets (Rogge et al., 2017). As every step a student takes in the online world can be traced, analyzed, and used, there are plenty of opportunities to improve the learning process of students.

First, data analytics techniques can be used to enhance the student’ s learning process by providing real-time feedback, or by enriching the learning experience. The latter might take place in adaptive learning paths that provide a tailored learning environment for students. Thanks to the use of data analytics, the learning environment can better correspond to students’ characteristics in terms of cognitive abilities, earlier acquired knowledge and skills, interests, learning style, motivation or meta cognitive abilities. While similar adaptive and differentiated learning is difficult to realize in a physical classroom, it is relatively easy to realize in the online classroom. We discuss this more extensively in Chapters 2, 4, and 9.

Second, data analytics can be used to support the instructor or teacher. Using data analytics, the instructor can better trace, and take targeted actions to improve, the learning process of the student. By combining comprehensive student data with learning outcomes— in terms of student success, dropout, or cognitive skills— of earlier cohorts of students, the learning outcomes of the evaluated student can be predicted. Forewarned by similar indicators, instructors can pay additional attention to those students who are at risk of lagging behind. Moreover, the instructor can obtain descriptive analytics from the progress that students are making in online courses, or their use of tools in the electronic learning environment (see Chapter 2). In addition, the instructor can use data analytics to detect fraud by students (see Chapter 4) in a cost-effective way. Creating quality indicators (Chapters 7 and 9) for courses is also facilitated by data analytics.

Third, we see possibilities in using data analytics to measure the performance of instructors. Today, it is relatively difficult to compare and assess the performance of instructors. If the performance of instructors is measured by students’ evaluations of teaching (SET), instructors with poor SET scores may argue that they face a more challenging student group, a more demanding topic, or that the students do not take the course seriously. Thanks to the abundance of data, these and similar arguments can be examined, and SET scores can be adjusted accordingly. De Witte and Rogge (2011) provide a model to do so. We will return to this issue in Chapter 5, where we discuss performance at the faculty level.

Finally, for policy makers, it is often unclear how schools use their available resources to “ produce” outcomes. By combining structured and unstructured data from various sources, data analytics might provide a solution for governments that aim to monitor the performance of schools more closely. In Chapter 6, we discuss some techniques to relate resources to outputs (e.g., test scores, graduation scores). While similar techniques have existed for some time, school performance scores can now better capture the observed heterogeneity in school, student, and neighborhood characteristics, thanks to the increasing availability of data. This will facilitate the use of these performance scores for policy purposes.

1.3 How Has This Become Possible?

The upsurge in data analytics is a result of the automatic recording and ready availability of data in electronic form. The main enablers of this evolution are the availability of cheap data gathering, data storage, and computing technology. Information systems store and manage data about student background, registration, program, and performance, and all these data can easily be exchanged, combined, and processed.

Moreover, the introduction of learning management systems and online learning applications allows huge amounts of data about the learning process to be collected in real time and at the source. Analogous to clickstream analysis or customer journey mapping in marketing domains, the availability of data about the learning process allows student behavior throughout the learning experience to be analyzed and described. This amount of education data, combined with student, course, and instructor information, makes it possible to use either traditional reporting techniques or more advanced forms of analytics, such as descriptive and predictive analytics.

1.4 Why Data Analytics Has Become Important

With the possibility of collecting and analyzing educational data comes the potential for enormous benefits through the proper use of data analytics. Policy makers (acting as principals in a classical principal– agent setting) are increasingly aware that, thanks to quantitative and qualitative data, they can better monitor the activities of the organizations they are funding (the agents). In fact, data analytics facilitates data-driven decision making such that, for instance, schools or universities can be better-compensated for their efforts in teaching students from disadvantaged backgrounds.

In addition, the availability of data allows stakeholders to assess the effectiveness (i.e., doing the right things) and efficiency (i.e., doing the thing right) of their interventions (see Chapter 10). While there is constant innovation in education (e.g., teachers who experiment with a different didactical instruction method), most of these interventions are not examined on their efficiency or effectiveness due to the absence of reliable data from before and after the implementation of the intervention. Data analytics can provide a solution.

At the same time, we need to be careful with the abundance of data. While combining data from various sources might create privacy issues, dealing with the overwhelming amount of data can also be an intricate issue (Chapter 9). It also forces us to think about such normative questions as whether we should store the data that are gathered during the student learning process. If we answer this question in the affirmative, to evaluate the effectiveness and efficiency of educational innovations, the question of how long we should store the data automatically arises.

Similar questions show that data analytics in education should not be the domain of a single discipline. Economists should discuss the possibilities, issues, and normative questions with a multidisciplinary team of pedagogists, philosophers, computer scientists, and sociologists. By bringing together various disciplines, a more comprehensive answer can be formulated to the challenges ahead.

This book provides a start to this discussion by highlighting some economic perspectives on the use of data analytics in education. We hope that the book marks the start of an interesting and multidisciplinary discussion such that, in the medium term, data analytics in education will seem as natural as a teacher in front of a classroom.

Kristof De Witte and Jan Vanthienen

1.5 List of Contributions

This book on data analytics in education is structured in three distinct parts. The first section, consisting of three chapters, discusses the use of data analytics in to improve the student learning process. The second section, with four chapters, details the use of data analytics to measure the performance of faculty, schools, and students. In the third section, two chapters are devoted to the policy relevance of data analytics and the challenges ahead.

1.5.1 Part I: Data Analytics to Improve the Learning Process

Part I of the book begins with a chapter by Johannes De Smedt, Seppe vanden Broucke, Jan Vanthienen, and Kristof De Witte. The chapter focuses on supporting the automated feedback learning environment through process mining. It discusses some new ways to process student data, for example, by social network analysis. As similar data are shown to predict student performance, these can be used by instructors to obtain insights into student’ s behavior and to act accordingly in real time.

Chapter 3, by Wouter Schelfhout, provides a model, based on learning communities, as a platform for growing data use. Research indicates that data use by schools and teachers is not widespread, and where it does occur, is often superficial. In this chapter, we argue that schools and teachers are not open to data use because the essential conditions for integrating it in daily practice are not met in many schools. There is a profound lack of an effective professional development policy, which should start with the core processes and concerns of schools and teachers. Equally important is the frequently observed absence of shared instructional leadership as a basis for shaping this policy. Developing different forms of learning communities— in focused interactions— will provide a platform for addressing these challenges and needs, while at the same time promoting a gradual increase in the integrated use of data. Learning how to gather specific process data on teaching practices must form part of educators’ professional development cycles to reach these goals. This will form a basis to give meaning to school internal output data and to school external data sources. Cooperation between schools and with external stakeholders such as education networks, governmental education departments, and school inspectors will be needed to support this endeavor. As part of this contribution, a holistic model of “ data for development” will be defined.

Chapter 4, by Silvester Draaijer and Chris van Klaveren, discusses the impact of fraudulent behavior and the use of learning analytics applications. Online quizzes are frequently used to prepare students for summative achievement tests. To encourage student participation, extra credits can be awarded to students who pass these quizzes. While anecdotal evidence indicates that offering quizzes carrying extra credit can result in fraudulent behavior in which students cheat to inflate their scores, there is as yet no empirical evidence investigating the extent of score inflation among, and its impact upon, cohorts of students. In this chapter, the impact of fraudulent behavior of first-year Dutch law students on weekly online quiz scores is studied. Exogenous variation in feedback to students was used to identify the impact. This exogenous variation was generated by abruptly, and without prior notice, ceasing to provide direct feedback to students on online quizzes. The main finding of the study was that the average quiz scores dropped by 1.5 points (on a scale of 0– 10) immediately after the unanticipated feedback change. This result, first, supports the anecdotal evidence that online quizzes may not be a valid representation of student knowledge due to fraudulent student behavior. Second, and more importantly for this volume, fraudulent behavior may cause online quiz data to undermine the effectiveness of learning analytics applications.

1.5.2 Part II: Data Analytics to Measure Performance

Part II of the book focuses on the use of...

Table of contents

• Cover
• Half-Title
• Title
• Copyright
• Contents
• Editors
• Contributors
• 1 Introduction: Big Data Analytics in a Learning Environment
• SECTION I DATA ANALYTI CS TO IMPROVE THE LEARNIN G PROCESS
• SECTION II DATA ANALYTI CS TO MEASURE PERFORMANCE
• SECTION III POLICY RELEVANCE AND THE CHALLENGES AHEAD
• Index