Exploring personal and contextual factors predicting teacher feedback in South Korea: A random forest analysis approach

This study investigated predictors of teachers' enacted feedback frequency, operationalized as two specific feedback behaviors: written feedback on student work and immediate process-level feedback. Using a random forest approach, we analyzed the South Korean cohort of the 2018 Teaching and Learning International Survey. Out of 92 variables encompassing teachers’ personal capacities, instructional practices, classroom ecology, and institutional context, we examined the top 20 predictors distinguishing high- and low-frequency feedback groups. Instructional strategies, self-efficacy, and pedagogical preparedness for diverse settings emerged as the most important predictors, alongside collaborative culture and professional development at the institutional level. Additional interaction analyses identified cognitive activation as central to the strongest interaction patterns, particularly with clarity of instruction and professional collaboration in lessons. These findings refine the perspective that feedback is co-constructed within institutional contexts by specifying which teacher capacities, instructional practices, and institutional conditions are most strongly linked to its enactment in everyday classrooms.