A Process-Oriented Approach to Assessing High School Students’ Mathematical Problem-Solving Competence: Insights from Multidimensional Eye-Tracking Analysis

The assessment of mathematical competence, particularly in real-world problem-solving contexts, has become increasingly crucial in high school educational evaluation. While traditional methods have shifted towards emphasizing problem-solving skills, they remain predominantly outcome-oriented, often failing to adequately capture the nuanced cognitive processes underlying students’ problem-solving behaviors. To address this gap, this study introduces a process-oriented assessment method leveraging eye-tracking technology. Fifty-three university students (primarily first- and second-year undergraduates) in China were recruited to solve six context-based mathematical problems of varying difficulty levels while wearing portable eye-tracking glasses, allowing for natural problem-solving behaviors in a paper-and-pencil test format. The study established a multidimensional model of eye movement features to evaluate problem-solving processes. Using China’s National College Entrance Examination (CNCEE) mathematics scores as the dependent variable, a Partial Least Squares Regression (PLSR) analysis achieved its best predictive performance (prediction R2 of 0.271) based on multidimensional eye movement features when solving the most difficult problem. The first visual intake duration on problem-reading areas and key information regions emerged as significant contributors of the students’ CNCEE scores. These findings substantiate the potential of eye-tracking technology as a valuable tool for educational assessment, offering insights into the assessment of students’ mathematical competence and supporting the development of more comprehensive learning diagnosis and intervention strategies.