Predicting positive youth development among Chinese adolescents: A machine learning approach using multiwave longitudinal data.

This study used machine learning to analyze four waves of data from over 5,000 Chinese adolescents aged 9-19 to identify what predicts positive youth development. The researchers found that school psychological climate, depression, and parental loneliness were the strongest predictors. The importance of these factors varied by age and by the student's circumstances: school climate mattered most for younger students, while parental loneliness was more important for older students, and different pathways emerged for migrant, left-behind, and rural children.

Key points

• School psychological climate, depression, and parental loneliness were identified as the top three predictors of positive youth development using machine learning analysis
• School climate was the dominant factor for primary and middle school students, but parental loneliness became more important for high school students
• Three distinct developmental pathways were identified: an aspirational pathway for migrant children, a relational pathway for left-behind and urban children, and a clinical pathway for rural ordinary children
• The machine learning model (CatBoost) achieved an R² of 0.816 in predicting positive youth development
• Results support the need for differentiated, context-sensitive intervention strategies tailored to diverse Chinese adolescent populations

Limitations

This study identifies predictors of positive youth development but does not evaluate a specific intervention; it cannot tell us what actually works to improve outcomes or how to implement recommendations. The findings are specific to the Chinese context and may not apply to other populations.