Learning Technologies and User Interaction

Diversifying Implementation in Curriculum, Instruction, and Professional Development

Kay K. Seo, Scott Gibbons

• Publisher: Routledge
• Year: 2021
• ISBN:
  9781000441253
• Edition: 1
• Topic: Education
• Subtopic: Inclusive Education

Learning Technologies and User Interaction explores the complex interplay between educational technologies and those who rely on them to construct knowledge and develop skills. As learning and training continue to move onto digital platforms, tools such as artificial intelligence, predictive analytics, video games, virtual reality, and more hold considerable potential to foster advanced forms of synergy across contexts. Showcasing a variety of contributors who are attuned to today's networked technologies, environments, and learning dynamics, this book is ideal for students and scholars of educational technology, instructional design, professional development, and research methods.

UNIT I
Enriching Curriculum

CHAPTER 2
Learning Analytics to Support Student Interaction and Learning Design

Initial Report on a Human-Centered Prototype Design

Priya Sharma, Mahir Akgun, and Qiyuan Li

DOI: 10.4324/9781003089704-4

INTRODUCTION

Interaction is an integral part of learning that occurs via interaction between individuals, artifacts, knowledge, and cultural practices. Interaction can be construed diversely: For example, Merriam-Webster’s 11th Collegiate Dictionary defines interaction as “mutual or reciprocal action or influence” (www.merriam-webster.com/dictionary/interaction), while the Cambridge Dictionary defines interaction as “an occasion when two or more people or things communicate with or react to each other” (https://dictionary.cambridge.org/us/dictionary/english/interaction). Our perspective on interaction more closely maps onto the second definition, and in this chapter, we address interaction from a pedagogical and research perspective.

In face-to-face (f2f) learning contexts, interaction that occurs between students, their peers, and instructors is visible and can be used to inform pedagogy and design of activities. Interaction in online contexts is equally important and has been underscored in a variety of distance and online learning frameworks (Anderson & Garrison, 1998; Moore, 1997). Specifically in online contexts, bolstering interaction between students, peers, instructors, and content is key to reducing transactional distance (Moore, 1989), which is the sense of psychological distance felt by participants in the online space. Higher levels of transactional distance correlate to fewer opportunities to interact and less persistence and learning online (Weidlich & Bastiaens, 2018). In contrast, increased interactions between student-student, student-instructor, and student-content are likely to provide more satisfactory learning experiences (Anderson & Garrison, 1998), although requiring higher time and cost investments.

In this chapter we report on the initial stages of a design research-informed project that mitigates the trade-offs of cost, effort, and interaction by using artificial intelligence, especially machine learning, to support instructors in quickly assessing student interaction in online environments.

Interaction and Learning in Online Contexts

As increasing numbers of students enroll in online courses, with an average 20% growth in enrollment each year (Lederman, 2018), it is imperative to design courses that support student engagement and interaction. The growth in online courses is accompanied by student and faculty preferences for blended learning (Brooks & Pomerantz, 2017), where some portion of the course content is delivered online. The increase in online and blended learning poses dual challenges for instructors and learning designers. First, the design of online experiences should closely attend to processes of learning, including interaction between student and peers, student and instructor, and student and content. Different learning theories emphasize different roles of interaction in the learning process, including the role of interaction in cognitive development (Vygotsky, 1978) and the role of social interaction in constructing shared knowledge (Roschelle & Teasley, 1995). Patterns of interaction between students have also been studied within computer-supported collaborative learning (CSCL) environments. Although the instructor’s role in structuring learning activities is important for supporting interaction and collaboration, it is often overlooked in the literature (Kreijns, Kirschner, & Vermeulen, 2013). Growing evidence suggests that participants are unable to interact in CSCL environments without guidance or support (Kreijns et al., 2013) and that merely assigning students to groups in CSCL environments will not guarantee collaboration or interaction (Lipponen, Rahikainen, Lallimo, & Hakkarainen, 2003). Along with the design of various types of interaction, equally important is designing assessment and rapid feedback to students on their interactions. Assessing quality of engagement is an integral part of assessing the quality of each member’s contribution to the group product, and instructor guidance for supporting students’ participation in peer interaction is very important (Hakkarainen, 1998). However, monitoring individual students’ participation in online group discussions is a challenging and time- consuming task for most instructors. Analytics can be used to ease this burden, and analytic tools can be designed to support instructors in monitoring group discussions and detecting the quality of cognitive engagement demonstrated by group members.

Artificial Intelligence and Learning Analytics

Artificial intelligence (AI) applications in education are multiplying and could grow significantly in the coming years (Zeide, 2019). Machine learning, encompassed within the broader set of AI techniques, is a method of supervised/unsupervised classification and can produce software capable of recognizing patterns, making predictions, and applying newly discovered patterns to novel situations (Popenici & Kerr, 2017). Machine learning has been previously used in education to automate time-consuming tasks such as grading and provision of personalized feedback (Vie, Popineau, Bruillard, & Bourda, 2018) as well as provision of analytical or predictive information about student performance, dropout rate, and sentiment (Hu, Lo, & Shih, 2014; Li, Hoi, Chang, & Jain, 2010; Minaei-Bidgoli, Kortemeyer, & Punch, 2004). In this chapter we examine the design and use of a deep learning classifier that automatically analyzes online student discourse and generates visualizations to represent the data. The visualizations can be represented and manipulated via a learning analytics dashboard.

Learning analytics (LA) include the measurement, collection, and reporting of data to enhance student learning and design of the learning environment (SOLAR, n.d.). LA are well suited to analyze and represent big data generated via online learning; however, the challenge for educators is gathering and interpreting relevant data to form actionable insights into design and learning (Ferguson, 2012). Learning analytics dashboards (LAD) can address this need. Many LADs reported in literature gather and analyze data that are automatically generated by learning management systems. For example, data such as student access of resources, time spent, grades on quizzes, and so on can be used to identify students at risk of failure. In contrast, our prototype dashboard focuses on capturing, analyzing, and visualizing qualitative, discourse-oriented data to provide a more complete picture of student activities integral to interaction and learning to engender impact on user behavior change. Importantly, our project also closely links LAD design to theories of learning and design (Jivet, Scheffel, Specht, & Drachsler, 2018).

Designing Human-Centered Dashboards for Learning

A growing focus in the field of learning analytics (LA) is the use of human-centered design (HCD) to increase alignment between dashboard designs and their contextual use by stakeholders (Ahn, Campos, Hays, & Digiacomo, 2019). This move echoes a larger move by the Human Centered Interaction (HCI) discipline away from a system-centered approach to a human-centered approach that positions stake-holders’ concerns and activities at the forefront of the design process (Bannon, 2011). Human-centered systems are designed and developed with the people who will eventually use the system. The HCD process uses methodologies and techniques that generate an understanding of the critical stakeholders, their needs, their activities, and the context in which those systems will operate (Giacomin, 2014). Learning analytic tools and dashboards have not always followed an HCD process, and the misalignment between their use and intent may result in a distrust of LA tools (De Quincey, Kyriacou, Briggs, & Waller, 2019), which in turn impacts real-world practices significantly. Thus, the success of LA tools depends on whether they were designed by considering users’ needs and contexts (Shum, Ferguson, & Martinez-Maldonado, 2019). Involving stakeholders and users such as instructors, students, and administrators in the design process is important for better understanding how the users work with and act on LA tools in authentic teaching or learning contexts.

PROJECT CONTEXT: DEVELOPMENT OF AN INSTRUCTOR FACING DASHBOARD

In the subsequent sections of this chapter, we present a description of the design and development of an instructor-facing learning analytics dashboard that focuses on automatically analyzing and visualizing the quality and quantity of student interaction in online discussions. Despite an exhaustive web and citation search, we were unable to find a tool with the functionalities we expected, which led to our goal of tool development. Our design of the dashboard is guided by human-centered design guidelines, and in this first phase our design focused primarily on supporting instructor decision making and pedagogical intent. We used a participatory co-design process (Bratteteig, Bødker, Dittrich, Mogensen, & Simonsen, 2013) to refine and redesign the instructor-facing learning analytics dashboard, and we reported the process using a design-research informed approach. Our work began with a focus on solving a practical problem and identifying a suitable intervention; however, our work is also theoretically grounded. As we continued building our intervention, we saw opportunities for theory and design to inform each other, as is the general trend of design research (Easterday, Rees Lewis, & Gerber, 2018). As a first step, we focused on the instructor as stakeholder to design and refine an initial prototype of our learning analytics dashboard.

The context of this work is an undergraduate information sciences and technology course within a large northeastern university in the United States. The course enrolls about 6–8 sections with approximately 50 students per section every fall and spring semester. Sections are offered face-to-face as well as online, and our focus was to support instructors in managing the online sections. Our project team members’ expertise spanned education, learning design, learning analytics, machine learning, and qualitative and quantitative research methodologies.

Initial Needs Analysis

Our first step was to explore the needs and challenges of one of the instructors of the information sciences class mentioned earlier. We conducted several interview-based, need-finding meetings to understand the challenges faced by the instructor in supporting and encouraging students to interact with each other and the course materials. The instructor assigned students to groups of 4–6 at the beginning of the course, and each group was tasked to complete 6 (six) group projects collaboratively during the semester. A sample activity may ask students to identify all the processes and interrelationships involved in the performance of a simple computing activity (such as printing or saving a file) and to create both a written description and a diagram of the process. Specific questions and prompts are also provided to guide the students through the different considerations (e.g., a more direct supportive prompt might read “How does the CPU communicate with the motherboard and the other parts of a system unit?”). Students then discuss their ideas and create the artifact. Through a process of reflection and guided prompting, we arrived at a list of key challenges encountered i...

Table of contents

• Cover
• Half Title
• Title Page
• Copyright Page
• Contents
• Contributor Biographies
• Introduction
• Unit I Enriching Curriculum
• Unit II Diversifying Instruction
• Unit III Revamping Professionalism
• Index