Supporting long-term student learning in math and science
We're creating a smart study tool that helps K-12 students truly remember what they learn in science and math classes, not just for a test, but for the long haul. It's like having a personalized tutor that knows exactly when to remind students of what they've learned. What makes our tool special is that it smartly adjusts to any amount of time a student has for practice, unlike others that can overwhelm them. This means less stress for teachers and students, and ultimately, more kids feeling confident and excited about STEM careers, especially as we help them catch up after the pandemic.
This project primarily impacted students in grades 6-12 across Texas, Minnesota, and New York and has the potential for broader application. While the direct implementation so far occurred in Houston ISD, Mansfield ISD, El Paso Leadership Academy, a Minneapolis independent school, and the Ascend and DREAM charter networks in New York City, the methods and findings are relevant and scalable to educators nationwide.
Our research has directly involved 1,445 K–12 students to date. This work has significantly advanced our understanding of how cognitive science interventions can be effectively integrated into educational technology.
Our primary output is the ongoing development and refinement of Podsie, an educational platform designed to improve long-term learning. Specifically, we've developed:
- A "Question Variation" feature within Podsie, which presents different versions of questions during review to foster conceptual understanding.
- A new adaptive algorithm (currently in development) for spaced retrieval practice. This algorithm is designed to be highly flexible, accounting for students’ historical performance, repetition frequency, and spacing intervals. It aims to provide a more personalized and time-efficient learning experience than existing methods.
Our work directly addresses several critical challenges in educational technology and learning:
- Enhancing Conceptual Understanding: Teachers reported a need for students to develop deep conceptual understanding, not just rote memorization. We are addressing this by investigating whether reviewing concepts not just facts at spaced intervals (spaced retrieval) truly enhances concept learning, and at what level (question or topic).
- Optimizing Spaced Retrieval for Concepts: Our research found that learning of concepts can be improved by spacing individual questions in varied types (e.g., multiple choice, short answer, etc.), rather than by spacing entire topics using different individual questions. This provides a crucial insight for effective design of educational technology.
- Overcoming Time Constraints and Overload in Classrooms: A major problem with existing spaced retrieval algorithms (like SuperMemo2, currently in Podsie) is their lack of flexibility. Pilot teachers reported that these algorithms often scheduled more review items than students could complete within limited classroom time, leading to student overload and reduced effectiveness. Our new adaptive algorithm is specifically designed to solve this by personalizing review schedules to fully utilize available time, allowing students to practice continuously and effectively within any given time window, thereby offering the personalization needed for effective STEM learning environments.
This project contributes to the knowledge base in STEM education by advancing our understanding of how spaced retrieval practice can be effectively applied to support long-term learning and conceptual understanding in K–12 classrooms. Our findings offer novel, evidence-based insights into the precise conditions (when and how) retrieval practice and item variability optimize STEM learning. Beyond new knowledge, we're developing an adaptive algorithm that uniquely tailors practice to each student based on their learning history and available time, making the learning experience more personal and efficient. This innovative learning tool is designed not just to improve memory but to help students build a strong, lasting foundation in STEM, thereby increasing engagement, confidence, and potentially leading to higher retention and success in future STEM pathways and careers. By effectively translating cutting-edge cognitive science research into easy-to-use classroom technology, we are empowering teachers to implement powerful, evidence-based learning strategies in their everyday practice.
No, the work would not have been possible without EDU funding. The support enabled us to conduct first of its kind large-scale, multi-year research with over 1,400 K–12 students in just two studies so far (more to come!), develop and test adaptive learning technologies, and collaborate with multiple schools and teachers. The funding was essential for building the infrastructure needed to implement and evaluate spaced retrieval practice in real-world STEM classrooms and for generating findings that can inform both educational practice and policy.
