青少年健康相关危险行为问卷的编制及信效度检验

目的:编制青少年健康相关危险行为问卷(AHRBI)并检验其信效度。方法:在文献分析的基础上,并结合前期实证研究结果,确定健康相关危险行为的6个维度。在长沙、苏州、成都和银川四地以班级为单位,共调查普通中学初一至高三的学生5755名(其中男生2997名);另外方便抽取长沙附近3所特训学校学生323名平均年龄(15±2)岁,男生256名]。采用Zuckerman-Kuhlman人格量表(ZKPQ)中的冲动感觉寻求(ISS)和攻击敌意(AH)两个分量表以及Barratt冲动量表(BIS)作为效标。从普通中学学生样本中,分别抽取329人和347人间隔4周和8周后进行重测。结果:通过项目分析确定38个条目,每个分量表5～10个。除健康妥协行为的Cronbachα系数为0.40外,其他5个分量表的Cronbachα系数均在0.75以上。全量表及各分量表间隔4周和8周的重测系数分别在0.55～0.80和0.53～0.76之间。验证性因素分析显示,二阶单因子模型的各拟合指标分别为S-Bχ2=3933.23,df=623,NNFI=0.88,CFI=0.89,RM SEA(90%CI)=0.029(0.029,0.031)。效标关联效度分析显示攻击暴力、破坏纪律和吸烟饮酒得分与效标间的相关系数在0.06～0.43之间,而无保护性行为、自杀自残和健康妥协与效标间的相关系数在-0.06～0.27之间。特训学校学生在所有6个分量表上的得分高于普通中学学生(Cohen's d值在0.67至2.88之间)。结论:本研究编制的青少年健康相关危险行为问卷包含38个条目测查6个方面的危险行为,具有较好的心理测量学特性。

Reliability and validity of the Adolescent Health Related Risky Behavior Inventory

Objective:To develop and validate a comprehensive assessment tool,the Adolescent Health related Risky Behavior Inventory(AHRBI).Methods:Six subscales,including the Suicide and Self-Injury(SS),Health-Compromising Behaviors(HCB),Aggression and Violence(AV),Rule Breaking(RB),Smoking and Drinking(SD) and Unprotected Sex(US),were determined through literature review and the Risky Behavior Questionnaire for Adolescents(RBQ-A) which was previously compiled.Totally 5755 adolescents,grade 7 to grade 12,were recruited from high schools across 4 cities in China.In addition,a target sample,which included 323 adolescents were recruited from three behavior modification schools,was used to evaluate predictive validity of the AHRBI.Meanwhile,Impulsive-Sensation Seeking(ISS) and Aggression-Hostility(AH) subscales of Zuckerman-Kuhlman personality inventory and Barratt Impulsiveness Scale(BIS) were selected as criterion measurement.Psychometric properties including internal consistency(Cronbach’ s alpha),test-retest reliability,convergent validity,and predictive validity were analyzed.Results:Through item-analysis,38 items were selected from 50 initial items which covered six behavior domains(5 to 10 items per domain).Except HCB,the Cronbach’ α coefficients of other five subscales were above 0.75.The test-retest reliability for the whole scale were 0.80 and 0.76;for subscales ranged from 0.55(HCB) to 0.78(SD) and from 0.53(HCB) to 0.76(US) at two time points(4-week and 8-week interval),respectively.The results of confirmatory factor analysis revealed that both first order six factors correlated model and second order model fit the data were acceptable.In order to consistent with our theory,we kept the second order model.The fit indices were S-Bχ2=3933.23,df=623,NNFI=0.88,CFI=0.89,RMSEA(90% CI)=0.029(0.029,0.031).Criterion-related validity analyses indicated that the correlation coefficients between RB,AV and SD subscales and criterions ranged from 0.06 to 0.43;while the correlation coefficients between SS,HCB and US subscales and criterions ranged from-0.06 to 0.27.The target sample scored significantly higher in all six subscales of AHRBI than the common school students(P<0.01;Cohen’ s d 0.67-2.88).Conclusion:These findings provide initial support for the reliability and validity of the Adolescent Health related Risky Behavior Inventory(AHRBI) as an assessment tool for adolescents who involving multiple risk behaviors.