Application of Latent Class Analysis to Predict Children’s Internalizing Problems

Abstract

Objectives: The purpose of this study was to identify the number of latent classes among elementary school children who internalize problems, and to find out whether internalizing problems of each latent class can be predicted by predictors. Methods: Data were collected from 1,957 children who participated in the Korean Children and Youth Panel Survey 2010 at four different time points (grade 2, grade 3, grade 4, and grade 6). Data were analyzed using latent class analysis and multinomial logistic regression. Results: First, analyses revealed four latent classes of children's internalizing problems: low, increasing, decreasing, and high. Second, children with high levels of self-esteem, ego-resilience, peer attachment, peer relationship, and parental monitoring are more likely to be in the low group as compared to the high group. On the other hand, children with high levels of teacher-children relationship are more likely to be in the low group as compared to the increasing group. Third, children with higher levels of abuse experiences are more likely to be in the high group than in the low group. With regards to boys and girl's internalizing problems, girls are more likely to be in the high group as compared to the low group. Conclusion: The results suggested that internalizing problems of the extracted latent classes were differently predicted by gender, self-esteem, ego-resilience, peer attachment, peer relationship, teacher-child relationship, parental monitoring, and parental abuse.