Numerous risk factors have been identified as contributors to the development of later delinquent and criminal behaviors in the U.S., including factors at individual, family, school, peer, and community levels. Not all children with many risk factors, however, follow identical pathways to criminality. Protective factors, which may be present in various domains in a child's life can buffer the effects of risks. Knowledge of risk and protective factors can inform approaches to prevention and intervention programs, where assessed needs of juveniles can be more effectively matched to treatment efforts with families, schools, and communities.
Studies of risk and resilience in P.R. China are to-date non-existent. Due to differences between the U.S. and Chinese legal system as well as social and cultural contexts, it is unknown whether risks and protective factors for developing children and adolescents identified in the U.S. are also significant in China. In 2003, a new, formal community-based program for juvenile delinquents and minor offenders was initiated in 18 provinces in China. The purpose of this study is to examine informal and formal social control regarding risk and protective factors for Chinese juveniles, living in Shanghai, in three conditions: Juveniles in a secured juvenile jail (serious offenders), in a nonresidential community-based program (minor offenders), and juveniles without police records. This study is the first to examine these characteristics among juvenile delinquents in P.R. China.
Methods
Participants were between ages 14 and 18 (n=133) and randomly selected from a juvenile jail and local schools; all participants in one community-based program were included. The Child Behavior Checklist -Youth Self Report (CBCL-YSR) was translated into Chinese and used to show differences in problem behavior among the three groups. Additionally, a list of risk and protective factors created by the author was used as a survey questionnaire. Police records were used to divide the participants into repeat- and non-repeat offenders. Crosstabs were used to show bivariate relationships between single risk factors and delinquency status. Using random forest, a new statistical approach taking the majority of votes from different decision trees, it is expected to have not only a smaller error rate compared with the traditional logistic regression, but also the ranking of the importance of all the significant predictors.
Results
Most of the risk and protective factors identified in the U.S. were also prevalent in this Chinese sample. Random forest model had an error rate of 4.73 comparing with 8.23 in the traditional logistic regression model. A reduced model containing parental divorce, truancy or school dropout, family poverty, endorsement of aggression, and academic performance, which was identified by the random forest, predicted delinquency status with 93.7 percent accuracy. Correlations between individual risk/protective factors and CBCL internalizing and externalizing behavioral problems were all significant, which indicated the consistency of the two instruments.
Conclusion and implications
Most significant risk/protective factors among juveniles identified in U.S. samples were also found to be significant in a Chinese context. The reduced comprehensive model offers a highly accurate prediction of delinquency in China.
