Educational Systems for Maximizing Learning Online and in the Classroom

Thesis (Ph.D.)--University of Washington, 2016-06

Over the past few decades, technology has evolved to a point where there is great potential for educational system to transform how we teach and learn. This has led to many exciting advances in educational technology. Platforms such as Khan Academy and Coursera provide millions of students with free access to online educational content. Intelligent tutoring systems personalize instruction and give intelligent feedback in response to any student mistake. Educational games such as Dragon Box and Refraction have a seemingly endless capacity to engage students in learning activities. However despite their potential, these systems have not fundamentally changed the state of education in the United States. Recent data shows that only one in four high school graduates are prepared for college, and student test scores have remained stagnant since the 1970s. While many factors influence the effectiveness of educational technology, one central problem is that computer scientists are trying to reinvent the wheel, rather than learning from prior research in education and the learning sciences. In this dissertation, I combine techniques from computer science and the learning sciences to develop novel technical systems grounded in learning theory that tackle core challenges in education. My collaborators and I apply this approach to address challenges in three key areas. First, we rethink the design of motivational incentives for educational games by considering psychology research on the importance of student mindsets. Next, we develop new technical approaches to facilitate the automatic generation of empirically validated instructional scaffolding. Finally, we study the behavioral impact of a novel formative assessment system to better understand the integration of technology into real-world classrooms. Taken together, this research demonstrates the value of leveraging findings from the learning sciences literature in the design of new educational technologies. The systems presented in this dissertation can have an immediate impact on learning: they have been used by over 100,000 students online, adopted by companies, and used in classrooms throughout the Seattle area. Most importantly, this research takes an important step towards the development of personalized data-driven systems that meet the unique needs of each student, teacher, and classroom.