An Exploratory Case Study on the Possibility of Performance-Based Learning Transfer in a Python-Based AI Mathematics General Education Course

This study explores how student’s conceptual understanding (Knowing) may extend toward into practical performance (Doing) in an AI mathematics general education course that incorporates Python programming activities. Grounded in learning transfer theory, the study examines the theoretical significance of programming-mediated AI mathematics education and analyzes patterns of achievement change and structural relationships using midterm and final assessment data, along with weekly programming activity records. The analyses included (1) descriptive statistics, reliability analysis, and item analyses to examine measurement characteristics; (2) comparisons of mid-final changes in achievement; (3) ANCOVA to explore differences in improvement across performance- level groups; and (4) covariate-adjusted regression analyses to investigate the relationships between programming engagement and subsequent achievement while controlling for midterm achievement as a proxy for prior ability.The results indicated a statistically significant increase in performance-based achievement from midterm to final assessment, with a medium effect size. Achievement-level anlayes showed relatively greater improvement among lower and middle achievement groups. However, there differences should be interpreted cautiously in light of potential changes in item difficulty and construct composition across assessments. Regression analysis revealed that programming engagement was not significantly associated with conceptual(theoretical) achievement but showed a positive and statistically significant association with final performance-based achievement after controlling for midterm achievement levels.These findings suggest that programming-based learning may be more closely related to the development of executable, performance-oriented competencies than to direct increases in conceptual understanding. Given that this study is based on a single-course, single-institution quasi-experimental case, the findings should not be interpreted as causal or broadly generalizable. Rather, the study provides exploratory empirical evidence regarding performance-based achievement change and potential for Knowing-Doing transfer in AI mathematics general education contexts, and offer considerations for future instructional and research design.