Abstract

This study examines how undergraduate students' written reflections function as learner-generated pedagogical evidence in LOC-R-based Lesson Study to improve Population Geography instruction. A qualitative classroom action research design was integrated with two Lesson Study cycles involving 28 students, one model lecturer, and seven observers. The corpus comprised 56 written reflections, supported by consolidated observer records and written see-stage minutes. Data were examined through qualitative content analysis and cross-cycle evidence-chain analysis. Cycle 1 reflections disclosed positive engagement alongside fatigue, uneven participation, and difficulty with map-based country classification. These findings informed clearer instructions, stronger visual support, adjusted pacing, structured group roles, and more focused participation in Cycle 2. However, Cycle 2 also introduced different sustainability content, activities, and sharper reflection prompts. Consequently, SDG-oriented statements were classified as prompt-dependent awareness or sustainability valuation rather than evidence of individual development in critical socio-spatial reasoning. No retained reflection met the minimum threshold for relational socio-spatial explanation or critical questioning. The study therefore distinguishes pedagogical improvement from changed content exposure and elicitation conditions. Its principal contribution is a transparent, auditable mechanism linking learner-reported experience, observational context, collective see-stage interpretation, and documented lesson-revision decisions, while demonstrating the importance of maintaining cautious inferential boundaries in non-equivalent classroom cycles.