Abstract: Background　With the extensive improvement of people's education，domestic cold violence has also increased.Domestic cold violence has severe harm.It not only destroys the mental and psychological health，but also jeopardize the harmony of marriage lives.Meanwhile，it affects the social solidarity.However，there is no effective tools at this moment in China to measure the domestic cold violence.Objective　To develop and psychometrically evaluate a domestic cold violence scale，so as to assess household cold violence.Methods　According to practical definition of domestic cold violence，the initial items pool（20 items）of the scale were compiled by literature review in October 2015.After discussion by expert committee，the pilot test version of the scale remaining 10 items was developed，which was tested among 20 wives of gay men or bisexual men（known as Tongqi in Chinese for short）who were recruited through the online chatting platform “Lang man zhuan shen tongqi group” in December 2015（with a 100.0% response rate）as pre-investigation.All participants filled out the scale within 2 minutes and expressed that the scale was easy to understand.Then a large-scale test was conducted among 195 Tongqis who met inclusion criteria and were recruited from Internet by Respondent-driven sampling（RDS）method with 15 “seed participants” from February 2016 to March 2017.The Demographic Questionnaire，the testing version of the scale（10 items）and subscale of the Revised Conflict Tactics Scales（CTS-2S）were used to collect data.Four weeks later，20 participants were selected to fill out the questionnaire again by simple random sampling method.Correlation coefficient and extreme group method were performed for item analysis，and then valid questionnaires were randomly divided into two groups.One group was tested by Exploratory Factor Analysis（EFA）in order to evaluate the sub-factors of domestic cold violence scale，and the other group was tested by Confirmatory Factor Analysis（CFA）in order to evaluate the structural validity.The validity of the scale was evaluated by the correlation coefficient of scores of the subscale for the abused and the total score of CTS-2S，and the content validity was calculated by using the scores given by experts.Reliability of this scale was evaluated by homogeneity test，split-half reliability and retest reliability.Results　A total of 178 questionnaires（195 in all）in large-scale investigation were valid，with effective recovery rate of 91.3%.A total of 20 questionnaires were sent out in the second survey，and 15 valid questionnaires were collected，with an effective recovery rate of 75.0%.The correlation coefficient between the item and total score of the testing version scale were 0.515 to 0.793（P<0.05）．Scores of all items in high-score group（n=52）were higher than those in low-score group（n=50）（P<0.05）．The KMO of the testing version scale was 0.854，and Bartlett's test showed that χ2=390.689，P<0.001，indicating that CFA could be performed.Two common factors were extracted according to the principle of Eigen value >1.000 and explained 59.913% of the total variance.However，the load values of three items in the common factor 1 were all lower than 0.400，so they were deleted and the CFA was conducted again limited to extract only one common factor.At this point，54.342% of the cumulative variance was explained，and all the item load values were more than 0.400.The result of CFA showed that the fit index χ2/df，RMSEA，GFI，CFI，NFI，IFI，TLI were 1.726，0.091，0.932，0.950，0.892，0.951，0.925，respectively.The score of the subscale of the CTS-2S was positively related with the total score of the domestic cold violence scale（r= 0.431，P<0.001）．The CVI of this scale was 0.97.The Cronbach's α coefficient，split-half reliability and retest reliability of this scale were 0.848，0.820 and 0.714，respectively.Conclusion　The final version of domestic cold violence scale consists of 7 items.This study shows that the reliability and validity of the domestic cold violence scale is acceptable.Therefore，this scale can be considered to assess the situation of individuals who suffer from cold violence in marriage.

Key words: Domestic violence；Homosexuality, male；Reliability；Validity

摘要： 背景　随着人们受教育水平的广泛提高，家庭冷暴力的发生率也逐渐增高。家庭冷暴力危害十分严重，不仅摧残着受害者精神、心理健康，还危害着婚姻家庭的幸福和谐，同时对社会团结造成一定影响。然而，目前我国尚无有效的工具对家庭冷暴力进行测量。目的　编制家庭冷暴力量表，以评估家庭成员中家庭冷暴力情况。方法　2015年10月，根据家庭冷暴力的操作性定义，初步发展了包含20个条目的条目池，经过分析讨论之后，共保留了10个条目，形成预调查家庭冷暴力量表。借助网络公益平台“浪漫转身同妻QQ群”，于2015年12月招募满足纳入标准的20名在婚或离异的同妻为预调查对象，采用预调查家庭冷暴力量表对其进行预调查。共发放问卷20份，回收有效问卷20份，有效回收率为100.0%。调查对象作答时间≤2 min，均表示条目通俗易懂，语言无歧义，从而形成测试版家庭冷暴力量表，仍包含以上10个条目。2016年2月—2017年3月采用同伴驱动的抽样方法，最初在“浪漫转身同妻QQ群”中选定15名同伴“种子”，然后通过其共招募195名同妻为大样本调查对象，采用大样本调查问卷﹝包括一般人口学资料问卷、测试版家庭冷暴力量表和修订版冲突策略量表（CTS-2S）中针对被虐方的分量表﹞对其进行调查。4周后，采用简单随机抽样法抽取20名在婚或离异同妻，对其采用测试版家庭冷暴力量表进行第2次测量。采用相关系数法和极端组法进行条目分析和筛选，采用随机平均分组法将大样本调查回收的有效问卷分为两份，其中一份进行探索性因子分析（形成家庭冷暴力量表），另一份进行验证性因子分析（评价家庭冷暴力量表结构效度）。采用CTS-2S中针对被虐方的分量表得分与家庭冷暴力量表总分的相关系数评价家庭冷暴力量表效标效度，采用专家判断法评价其内容效度，采用同质性信度、分半信度、重测信度评价其信度。结果　大样本调查共发放问卷195份，回收有效问卷178份，有效回收率为91.3%。第2次调查共发放问卷20份，回收有效问卷15份，有效回收率为75.0%。测试版家庭冷暴力量表各条目得分与其总分的相关系数为0.515~0.793（P<0.05）。高分组（n=52）测试版家庭冷暴力量表各条目得分均高于低分组（n=50）（P<0.05）。测试版家庭冷暴力量表的KMO值为0.854，Bartlett's球形检验的χ2=390.689，P<0.001，提示可进行探索性因子分析。根据特征值>1.000的原则共提取2个公因子，累积方差贡献率为59.916%，由于公因子1中3个条目的载荷值较低（<0.400），将其剔除；再次进行探索性因子分析，并限定只生成1个公因子，此时，累积方差贡献率为54.342%，且每个条目的载荷值均>0.400。验证性因子分析结果显示，χ2/df=1.726，RMSEA=0.091，GFI=0.932，CFI=0.950，NFI=0.892，IFI=0.951，TLI=0.925。CTS-2S中针对被虐方的分量表得分与家庭冷暴力量表总分呈正相关（r=0.431，P<0.001）。家庭冷暴力量表的总体内容效度指数（CVI）为0.97。家庭冷暴力量表的Cronbach's α系数、分半信度系数、重测信度系数分别为0.848、0.820、0.714。结论　本研究组编制的家庭冷暴力量表包含7个条目，其信效度良好，可用于评测个体在婚姻中遭受家庭冷暴力的情况。

关键词: 家庭暴力, 同性恋, 男性, 信度, 效度