青少年危险行为的聚类分析

目的 探索存在危险行为青少年群体的分组及其组别特征,为完善干预方案提供依据.方法 利用“我国青少年危险行为的评估、预警与干预示范研究”Ⅲ级干预入组青少年干预前的评估数据,结合两步聚类和快速聚类两种聚类分析方法,对根据危险行为已划分为四组的249名入组青少年进行重新分组,比较新的组间差异.结果 两种聚类分析结果的Kappa一致性检验值为0.861（P＜0.001）,聚类结果稳定.快速聚类分析显示入组青少年分为冲动外显组,自杀高危组,饮食/活动问题组.包括四种危险行为维度均分、中学生心理健康量表总均分、流调中心用抑郁量表总分和焦虑自评量表总分在内的参与聚类的7个变量在三组间差异均有统计学意义,F值为13.874～ 125.753 （P＜ 0.001）.三组性别构成的差异有统计学意义,x2值为59.059（P＜0.001）,冲动外显组男女比例大于其他两组.结论 对于存在危险行为的青少年群体,聚类分析能较好地将其分组,据此制订的干预方案可能更佳.

Cluster analysis of youth with risk behaviors

Objective To explore the grouping of youth with health risk behaviors and the characteristics of each group,and to provide the basis for improving intervention programs.Methods This study used data from ＂Demonstration Research on Assessment,Warning and Intervention of Youth Risk Behaviors in China＂,with 249 adolescents being enrolled in the tertiary intervention part of this research.Two step cluster analysis and quick cluster analysis were used together to regroup these enrolled adolescents who had been grouped into four groups.The differences among new groups were examined.Results Kappa consistency test showed high consistency between two solutions of cluster analysis （Kappa=0.861,P〈0.001）,which certified the stability of clustering result.The enrolled adolescents were clustered into three groups：impulsive explicit group,high-risk suicide group and diet/sport problem group.The differences were significant among three groups in 7 variables for clustering,including the dimension mean scores of four risk behaviors,the total mean scores of Mental Health Scale of Middle School Students,the total scores of CES-D and the total scores of SAS （F=13.874-125.753,P〈0.001）.Sex ratios were significantly different among three groups as well （x2=59.059,P〈0.001）,and male/female ratio of impulsive explicit group was higher than other two groups.Conclusion Cluster analysis can well group youth with health risk behaviors,and intervention programs might be more effective in view of the above grouping.