Unlike variable-centered analysis, person-centered analysis facilitates the development of more personalized learning. In this study, we investigated the predictors of academic performance in pre-clinical education using variable-centered analysis and identified academic performance profiles using person-centered analysis (cluster analysis). A total of 321 second-year medical students from our medical English course at Hokkaido University participated in this study between 2019 and 2021. For variable-centered analysis, we assessed the predictors of academic performance (final exam score) using multivariable regression analysis. We performed cluster analysis using Ward’s minimum variance hierarchical clustering method for person-centered analysis. Nine variables, identical to those used in the variable-centered analysis, were selected and standardized for this analysis. Online education, female sex, baseline reading skills, and medical terminology predicted academic performance in the variable-centered analysis; however, cluster analysis identified four subgroups with different academic performances. Cluster 1 had the highest proportion of male and in-person education and the lowest academic performance. Cluster 2, with a high male ratio and low baseline skills, had poor academic performance, whereas most students had online/combined education. Clusters 3 and 4 had the highest online/combined education (Cluster 4: online> combined), and female students with the highest performance. Online education was the main predictor of academic performance using traditional analysis; however, cluster analysis captured a subgroup of students with online education but lacking better academic performance. The current person-centered analysis contributes novel insights into the nuanced interplay of educational formats, gender, and skills, offering implications for optimizing medical education strategies in the future.
Goudarzi et al. (Fri,) studied this question.