摘要
目的:建立儿童期虐待问卷量表的中文版,并分析其信度和效度。方法:于2003-10/2004-12随机抽取河南省新乡市一所乡村中学和广东省惠州市两所普通高中16个班级共819名学生作为调查对象。由班主任组织采用儿童期虐待问卷进行团体测验。儿童期虐待问卷共有28个条目,分为5个分量表:情感虐待,躯体虐待,性虐待,情感忽视和躯体忽视,每个分量表含5个条目。每个条目采用5级评分(1,2,3,4,5,分别表示从不,偶尔,有时,经常,总是);其中第2,5,7,13,19,26和28需反向计分,每个虐待分量表评分5～25分,总分25～125分。设3个条目作为效度评价。测试中使用统一的指导语,为保证回答的真实性,无记名,问卷当场回收。2个月后随机抽取其中2个班级93名调查对象进行第2次评定,以评价量表的重测信度。信度检验包括①内在一致性:计算Cronbachα系数,分量表相关,计算各分量表条目之间的Person相关。②重测信度系数:计算93名学生前后各分量表得分的Person相关。效度检验:计算每个条目与量表总分的Spearman相关系数,以检验每个条目的效度。并进行验证性因素分析,以考验数据与5个分量表模型的拟合程度,检验量表的构想效度。采用SPSS11.5统计软件对数据进行分析。应用Amos4.0软件进行验证性分析。结果:发放问卷819份,剔除25份数据不全问卷,收回合格问卷794份,有效率96.9%。①信度检验:儿童期虐待问卷中文版量表总的Cronbachα系数为0.77;各分量表的Cronbachα系数为0.41～0.68。量表条目间Person相关系数0.01～0.45;条目间平均相关系数为0.12。各分量表的条目间平均相关系数0.14～0.30。2个月重测信度为0.75,各分量表的重测信度0.27～0.73。②效度检验:内容效度:量表分量表与总量表的相关系数0.45～0.76,均有极显著相关(P<0.01)。结构效度:各分量表与总量表的相关均超过各分量表与其他分量表之间的相关。分量表之间的相关系数0.04～0.45,多数有极显著相关。③验证性因素分析的指标:验证性因子分析的标准路径显示每个项目在相应因子上的负荷为:情感虐待0.39～0.59;躯体虐待0.27～0.74;性虐待0.51～0.61;情感忽视0.15～0.61;躯体忽视0.30～0.43。复相关系数0.03～0.56。各种拟合指数χ2/df(3.64)、距离指数近似均方根误差RMSEA(0.06);增值指数犤IFI(0.80),NNFI(0.91),CFI(0.80)犦;简约指数犤PGFI(0.74)犦。结论:①儿童期虐待问卷中文版具有较好的信度、效度,除躯体忽视分量表外具有较好的内部一致性,内容效度显示分量表与总量表均有极显著相关,结构效度显示各个分量表相对较独立。②各分量表相互之间既相互独立又相互联系地反映了儿童期虐待概念的各个侧面,说明该量表是一种较好的儿童期虐待测评工具,但其中躯体忽视分量表需要进一步修订。
AIM： To create Chinese version of Childhood Trauma Questionnaire-28 item Short Form （CTQ-SF） and analyze its reliability and validity.METHODS： Totally 819 students from 16 classes were collected randomly from a countryside middle school of Xinxiang, Henan Province and two common senior high schools of Huizhou, Guangdong Province from October 2003 to December 2004. The teachers in charge of the class organized the community test with the Chinese version of CTQ-SF. CTQ-SF included 28 items and divided into 5 subscales, named emotional abuse,physical abuse, sexual abuse, emotional neglect and physical neglect. Five grades were adopted in each item, named 1 score： never; 2 score：occasionally; 3 score： sometimes; 4 score： often; 5 score： always. Among those, the counter-scoring was required on No. 2. 5. 7. 13. 19. 26 and 28.Each sub, ale was varied from 5 to 25 scores and the total score was in the range from 25 to 125 scores. Another 3 items that was taken as validity evaluation had not been analyzed yet. In order to ensure the reality of the answer, students did not filled up their names and answer the questions with the unified instructed words. The questionnaires were collected on the spot when accomplished. Ninety-three of them were retested 2 months later in order to assess the test-retest reliability. Determination of reliability： ①Internal consistency： To calculate Cronbach a coefficient, correlation of subscales, and correlation of Person in each subscales ② Coefficient of the test-retest reliability： to calculate Person correlation of the results in every subscale in twice measurements of 93 students. Determination of validity：to calculate Spearman correlation coefficient between every item and total score of scale so as to verify the validity of each item. Using confirmatory factor analysis, we checked the fit degree between the data and the patterns of 5 subscales and content validity of the scale. SPSS 11.5 statistical software was used for analysis on the data. Amos4.0 software was used for analysis of the confirmatory factor analysis.RESULTS： Totally 819 pieces of questionnaire were distributed on the spot and 794 pieces of questionnaire with integral and regular answers were collected, excluded 25 incomplete questionnaire. Utility rate was 96.9%. ① Reliability check： The total Cronbach＇s alpha coefficient of the CTQ-SF was 0.77. The Cronbach＇s alpha coefficient of each subscales was varied from 0.41-0.68. The correlation Person coefficients of each suhscales ranged from 0.01-0.45. The average correlation coefficients was 0.12, and average coefficient of each subscales ranged from 0,14 to 0.30,The test-retest reliability coefficient was 0.75, and coefficient of each subscales ranged from 0.27-0.73, ② Validity cheek： The correlation coefficients of the five subscales score with the total scale score ranged from 0.45 to 0.76. It had significant correlation （P 〈 0.01 ）. Validity of formation： The correlation between general scales and subscales was more than the correlation between each subscale. The correlation coefficients among the five factors ranged from 0.04 to 0.45. Most of them had significant correlation. ③ Marker of confirmatory factor analysis： Standard way of confirmatory factor analysis displayed the load of every item on corresponding factor： emotional abuse ranged from 0.39-0.59; physical abuse ranged from 0.27-0.74; sexual abuse ranged from 0.51-0.61;emotional neglect ranged from 0.15-0.61 and physical neglect ranged from 0.30-0.43. The squared multiple correlations ranged from 0.03 to 0.56. The fit indices which included ,x^2/df （3.64）,RMSEA （0.06）,IFI （0.80）,GFI（0.91）,CFI （0.80）,PGFI （0.74） and PCFI （0.71） all met the criteria standards for adequacy of fit.CONCLUSION： ① The Chinese version of CTQ-SF presents good reliability and validity, which indicated that it provides better internal property but physical neglect. Content validity shows that each subscale is significantly correlated with the total scales. Structural validity shows that each subscale is relatively dependence, ②The mutual dependence and mutual relation among subscales show the content and structure of the Chinese version of CTQ-SF. The Chinese version of CTQ-SF is a good psychometric instrument for the evaluation of Chinese childhood abuse. But the physical neglect requires a further improvement.
出处
《中国临床康复》
CSCD
北大核心
2005年第20期105-107,共3页
Chinese Journal of Clinical Rehabilitation
基金
美国纽约中华医学基金会资助项目(01-749)~~
