Student Research Projects
Building a Better Classroom from the Inside
In the Learning Sciences & Technologies (LST) program, graduate students find their questions, focus, and answers within area schools, teaming with teachers and students in Worcester’s K-12 city classrooms—while developing techniques in WPI labs to test, track, and change student learning patterns. The work may involve data mining, software development, visual tracking, or pattern recognition; all have the goal of improving learning in U.S. schools. Learn about some of our students’ projects, past and present.
Title: Science ASSISTments
Description: We worked to create a tool suite to help educate and assess students in the scientific method, helping them hone skills in hypothesizing, experimenting, and analyzing and communicating data, in the fields of physics and biology. We have successfully created machine-learned detectors for various student behaviors, to allow teachers and future software to assess what areas a student excels at and where they still need support.
Approach: The lab utilized a combination of computer science, learning science, and cognitive psychology. Learning science and cognitive psychology provided the theoretical background in the design. Computer science provided the technical implementation and the algorithms we used to auto-score student work and detect student behavior.
What did you learn that influences you now? I learned the challenges in working with a long-term software product, as well as the human factor in working on a project people are actually expected to use in the classroom. Learn more...
Title: Increasing Parent Engagement in Student Learning Using an Intelligent Tutoring System with Automated Messages
Description: This study explores the ability of an Intelligent Tutoring System (ITS) to increase parental engagement in student learning. A parental notification feature was developed for the web-based ASSISTments ITS, allowing parents to log in and access detailed data about their children’s performance. Middle school parents were then invited to answer a survey assessing how engaged they felt in their children’s education. A randomized controlled experiment was run, during which weekly automated messages were sent to parents regarding their children’s assignments and how the students were performing. A post-study survey found that access to this data caused parents to become more involved in their students’ education. The work also led to increased student performance in the form of higher homework completion rates. Qualitative feedback from parents was very positive.
Approach: This project involved computer science with a little bit of learning sciences thrown in. It consisted of a significant programming portion in the form of adding a parent notification system to ASSISTments. Analyzing student learning data was necessary to demonstrate results.
What did you learn that influences you now? The programming component helped prepare me for the work I am doing now, such as user data aggregation and analysis. More importantly, it taught me how to see my work from the user's perspective and judge it by its effectiveness in providing value (in the case of my project, increased information to parents about their students) rather than just technical merits. Learn more...
Title: Predictive Models of Student Learning
Description: The project, my PhD dissertation, sought to build up modeling capabilities to identify the learning and assessment value of tutor content, interventions, and problem orderings analytically. The work also detects individual student attributes such as prior knowledge and speed of learning. The result is a set of learner analytics technologies that provide significantly more accurate assessment and capabilities for adaptivity.
Approach: These capabilities were built upon a very simple but proven model of student skill acquisition called "Knowledge Tracing," which was influenced heavily from past work in cognitive psychology. Additions to the model were made tractable by advances in computer hardware and development of algorithms for belief propagation.
What did you learn that influences you now? The intensely interdisciplinary effort required to demonstrate progress in educational technology. Learn more...
Title: Math Learning Environment with Game-Like Elements: an incremental approach for enhancing student engagement and learning
Description: Educational games may make learning more enjoyable, but at a potential cost of compromising efficiency by consuming both instructional time and student cognitive resources. Therefore, we created a learning environment with game-like elements—game aspects that are engaging, but may not negatively impact learning effectiveness. We made a tutor incrementally more game-like, then evaluated to determine the effect of game-like elements: benefits such as enhancing engagement and learning, as well as costs such as distraction and working memory overload. We developed four versions of a math tutor with different degrees of game-likeness, such as adding narrative and visual feedback. The four systems were pedagogically equivalent, consisting of 27 main tutor problems with the same hint and bug messages, and mini-tutorial lessons. Based on a study with 252 students, we found more satisfaction with the more game-like tutor. Students also took an 11-item pretest and post-test questionnaire, and those with the most game-like tutor had significant learning gain, but there was no reliable difference between the different versions of the tutor.
Approach: Our goal was to create a new learning environment with an aim to optimize both engagement and learning—one of the burning challenges in the field. We generated our theory and design based on different educational and psychology literature, and study of other learning systems. After we brought our system to the classrooms, we employed different data mining techniques to analyze it for crisper understanding of student interaction and learning.
What did you learn that influences you now? One of the biggest challenges of educational research is validation of results. We went into classrooms and interacted with students, a very indeterminate environment where the data gets noisy. It was thus a real challenge to draw clear conclusions from the experiments. It was a very big lesson for me that, no matter how confident I am in my design and execution, I should be well prepared to deal with the new uncertainties that come in the real implementation of our projects. Learn more...
Title: To Tutor or Not to Tutor: That is the Question
Description: Intelligent tutoring systems often rely on interactive tutored problem solving to help students learn math, requiring them to work through problems step-by-step while the system provides help and feedback. This approach has been shown to be effective in improving student performance in numerous studies. However, tutored problem solving may not be the most effective approach for all students. We wanted to determine whether tutored problem solving was worth the extra time it took, or if students would benefit from practice on more problems in the same amount of time. This study compares tutored problem solving to presenting solutions, while controlling for time. We found that more proficient students clearly benefit more from seeing solutions than from tutored problem solving when we control for time, while less proficient students benefit slightly more from tutored problem solving.Approach: Computer science, cognitive science, math education
What did you learn that influences you now? This research brought us a small step closer to understanding how to optimize learning in an intelligent tutoring system by presenting the most effective and efficient approach to students, determined by their knowledge level and the problem’s difficulty. Highly proficient students should not have to waste time going through long problems step-by-step, causing them to become frustrated or bored. But students who have low proficiency may need to spend the extra time and get the focusing help that step-by-step tutoring provides. Learn more...