Answers are grounded in this paper and cite their sources.
Objectives: With the advances in digital technologies, adolescents are more likely to engage in multiple types of delinquent behaviors; however, few studies have considered various types of delinquencies with a multidimensional approach. Therefore, this study aimed to identify the latent groups based on various delinquencies and examine latent groups’ predictors using machine learning and multidimensional approaches.Methods: Using nationally representative data (N = 8,210; male 52.17%), a latent class analysis was conducted to identify latent groups based on 10 delinquent behaviors self-reported by adolescents. A random forest analysis was then employed to predict latent groups using 53 predictors including demographic (6 items), personal (44 items), and environmental levels (3 items). The Mplus program was used for latent class analysis, and the R version 4.0.5 program was used for random forest analysis.Results: Three latent groups were identified and categorized as “high online delinquency” (3.2%), “drink/media delinquency” (26.8%), and “low delinquency” (70.0%). Random forest analysis found adolescents’ perception of harmful substances was the most important variable predicting the delinquent groups. Partial dependence plots showed the higher the interest in harmful substances, the lower the ability to respond to the harmful substances; the higher the access to harmful environment, the lower the grade; and the lower the autonomy control ability, the more likely the classification as a high delinquency group.Conclusion: This study classified three latent groups while considering the detailed types of delinquencies. The study found the prominent predictors using machine learning and multidimensional approaches to suggest useful delinquency prevention programs.
No takes yet. Share an insight, caveat, or question.
Lee et al. (2022) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: