How Much Mastery Is Enough Mastery? The Relationship between Mastery in a Lesson and the Performance on the Subsequent Lesson
Saved in:
| Title: | How Much Mastery Is Enough Mastery? The Relationship between Mastery in a Lesson and the Performance on the Subsequent Lesson |
|---|---|
| Language: | English |
| Authors: | Jiayi Zhang, Kirk Vanacore, Ryan S. Baker, Nabil Ch, Caitlin Mills, Owen Henkel |
| Source: | International Educational Data Mining Society. 2025. |
| Availability: | International Educational Data Mining Society. e-mail: admin@educationaldatamining.org; Web site: https://educationaldatamining.org/conferences/ |
| Peer Reviewed: | Y |
| Page Count: | 7 |
| Publication Date: | 2025 |
| Document Type: | Speeches/Meeting Papers Reports - Research |
| Education Level: | Elementary Education Junior High Schools Middle Schools Secondary Education |
| Descriptors: | Elementary School Students, Junior High School Students, Mastery Learning, Teaching Methods, Competence, Bayesian Statistics, Artificial Intelligence, Mathematics Skills, Educational Technology, Electronic Learning, Skill Development |
| Abstract: | Mastery learning -- requiring students to achieve proficiency in a topic before advancing -- is a well-established and effective teaching method. Digital learning systems support this approach by personalizing content sequences, enabling students to focus on practicing topics they have not yet mastered. To achieve this, digital learning systems use knowledge tracing models, such as Bayesian Knowledge Tracing (BKT), to estimate students' knowledge. The estimation is often converted into a binary indicator reflecting whether mastery has been achieved based on a predefined threshold (e.g. 0.95). Determining optimal thresholds is critical. While prior studies have identified thresholds to prevent over-practice on the same skill, it is equally important to examine how a student's degree of mastery predicts future learning on other skills, where prior mastery may facilitate acquiring new skills. The current study explores this relationship using data from Rori, an online tutoring system for foundational math skills. Using BKT, we categorized students' knowledge estimates at the end of each lesson (lesson N) into eight mastery levels and analyzed how the current mastery level is associated with students' future learning, measured by their performance, early and final knowledge estimates, and learning in the subsequent lesson (lesson N+1). Results indicate that while the widely adopted threshold of 0.95 remains relevant, higher thresholds, such as 0.98, yield additional benefits, including improved performance and learning in subsequent lessons. These findings provide empirical insights for designing adaptive learning technologies that enhance personalization, efficiency, and support for future learning. [For the complete proceedings, see ED675583.] |
| Abstractor: | As Provided |
| Entry Date: | 2025 |
| Accession Number: | ED675652 |
| Database: | ERIC |
| Abstract: | Mastery learning -- requiring students to achieve proficiency in a topic before advancing -- is a well-established and effective teaching method. Digital learning systems support this approach by personalizing content sequences, enabling students to focus on practicing topics they have not yet mastered. To achieve this, digital learning systems use knowledge tracing models, such as Bayesian Knowledge Tracing (BKT), to estimate students' knowledge. The estimation is often converted into a binary indicator reflecting whether mastery has been achieved based on a predefined threshold (e.g. 0.95). Determining optimal thresholds is critical. While prior studies have identified thresholds to prevent over-practice on the same skill, it is equally important to examine how a student's degree of mastery predicts future learning on other skills, where prior mastery may facilitate acquiring new skills. The current study explores this relationship using data from Rori, an online tutoring system for foundational math skills. Using BKT, we categorized students' knowledge estimates at the end of each lesson (lesson N) into eight mastery levels and analyzed how the current mastery level is associated with students' future learning, measured by their performance, early and final knowledge estimates, and learning in the subsequent lesson (lesson N+1). Results indicate that while the widely adopted threshold of 0.95 remains relevant, higher thresholds, such as 0.98, yield additional benefits, including improved performance and learning in subsequent lessons. These findings provide empirical insights for designing adaptive learning technologies that enhance personalization, efficiency, and support for future learning. [For the complete proceedings, see ED675583.] |
|---|
