This study investigates the heterogeneity associated with multicultural acceptance among multicultural adolescents in South Korea by integrating machine learning with latent profile analysis (LPA). Utilizing a sample of 1,146 adolescents, the research employed a two-stage analytical framework: first, identifying key predictors through random forest and XGBoost algorithms, and second, delineating latent subgroups via LPA. Machine learning results demonstrated that school life satisfaction, academic satisfaction, and peer relationships were the most robust predictors of the multicultural acceptance, outperforming individual psychological symptoms. Subsequent LPA identified three distinct profiles: (1) individuals with consistently high levels of psychosocial and relational resources; (2) those characterized by uniformly low levels across psychosocial indicators; and (3) a moderately disadvantaged group showing slightly average levels across dimensions. These findings contribute to enhancing multicultural acceptance among multicultural adolescents by facilitating differentiated intervention strategies. These results provide a foundation for fostering successful integration into Korean society by enhancing institutional belonging and social cohesion.
He et al. (Thu,) studied this question.