摘要
目的:检验中文版长处与困难问卷信度系数(α系数、重测信度系数和分半信度)的稳定性,以及探讨α系数的影响因素。方法:根据预先设定的文献纳入标准,本研究共纳入46篇论文的60个样本(N=114576),采用信度概化元分析技术对中文版长处与困难问卷的信度系数进行分析。结果:(1)三种信度的异质性显著,随机效应模型估计的α均值分别为:困难分量表0.749、情绪症状分量表0.686、品行问题分量表0.593、多动与注意不能分量表0.741、同伴关系分量表0.397和亲社会行为分量表0.743。(2)通过混合效应模型的调节分析发现:出版年份、使用版本、文献来源、出版物类型这4类研究特征对各分量表分数的α系数有不同程度的调节作用。在样本特征方面,东部沿海地区和儿童群体α系数都较低。结论:品行问题和同伴关系分量表可靠性低,题项异质程度大,建议修订题项。同时应注意样本特征和研究特征对α系数的影响,特别是用于东部沿海地区及儿童样本时需要谨慎。
Objective: To test the stability of the reliability coefficient of the Chinese version of Strengths and Difficulties Questionnaire(SDQ), and to explore the influencing factors of the coefficient a. Methods: A total of 46 studies, 60 samples(N =114576) met the selection criteria. Reliability generalization meta-analysis was used to analyze reliability coefficients of the SDQ. Results: ①Three types of reliability coefficients of the SDQ total scores and subscales scores were significantly heterogeneous. Based on random-effect model, the estimated coefficients of alpha were 0.696(total scale), 0.749(difficulties subscale), 0.686(emotional symptoms subscale), 0.593(conduct problems subscale), 0.741(hyperactivity/inattention subscale), 0.397(peer relationship problems subscale) and 0.743(prosocial behavior subscale).(2)According to mixedeffect model, publication year, SDQ version, article resource, publication type, sample type, sampling area were found to moderate the coefficient alpha of each subscales on different levels. In terms of sample characteristics, coefficient a of SDQ scores in samples from the eastern coastal areas and children’s types was low. Conclusion: The peer relationship problems subscale and conduct problems subscale demonstrate an unacceptable coefficient a. We suggest to revise the items. The influence of sample characteristics on coefficient a should be noticed when using SDQ in the future studies.
作者
许文兵
王孟成
邓嘉欣
刘惠华
曾红
杨文登
XU Wen-bing;WANG Meng-cheng;DENG Jia-xin;LIU Hui-hua;ZENG Hong;YANG Wen-deng(Department of Psychology,Guangzhou University,Guangzhou 510006,China;The Center for Psychometric and Latent Variable Modeling,Guangzhou 510006,China;Guangdong Minor Mental Health Education Cognitive Neuroscience Key Laboratory,Guangzhou 510006,China)
出处
《中国临床心理学杂志》
CSSCI
CSCD
北大核心
2019年第1期67-72,共6页
Chinese Journal of Clinical Psychology
基金
国家自然科学基金(31400904)
广州大学青年拔尖人才培养项目(BJ201715)的资助
关键词
长处和困难问卷
信度
信度概化
元分析
SDQ
Reliability
Reliability generalization
Meta-analysis
