抄録
This study examines how undergraduates engage with data science when it is framed through social inclusion. A short general-education lecture at Rissho University enrolled 168 students (Sports program: 89; Regular program: 79). Using a mixed-methods design that combines quantitative analysis of six Likert-type items and qualitative natural language processing of open-ended reflections, we describe post-lecture patterns rather than changes from before to after the session. Quantitatively, comprehension items tended to be higher among Regular students, while interest and self-relevance were strong across both groups. K-means clustering (k = 3) yielded interpretable profiles: C0 High-Understanding, C1 Moderate, and C2 Low-Confidence/High-Interest, indicating that interest can precede confidence. Qualitatively, responses shifted from abstract notions of technology or data use toward concrete social domains, including disability and welfare, poverty alleviation, decision transparency, and community issues, organized into the themes Data Utilization, Social Issues, and Disability and Welfare. Taken together, the results support context-led, problem-first pedagogy that pairs on-ramp scaffolds for beginners with brief applied outputs and equity-aware assessment. Such design choices can lower psychological and curricular barriers without recreating STEM-style gatekeeping, extending data literacy to learners beyond traditional STEM tracks.
https://ror.org/0496p0503 
