Confirmatory Factor Analysis of the WHO Violence Against Women Instrument in Pregnant Women: Results from the BRISA Prenatal Cohort

Background
Screening for violence during pregnancy is one of the strategies for the prevention of abuse against women. Since violence is difficult to measure, it is necessary to validate questionnaires that can provide a good measure of the phenomenon. The present study analyzed the psychometric properties of the World Health Organization Violence Against Women (WHO VAW) instrument for the measurement of violence against pregnant women.
Methods
Data from the Brazilian Ribeirão Preto and São Luís birth cohort studies (BRISA) were used. The sample consisted of 1,446 pregnant women from São Luís and 1,378 from Ribeirão Preto, interviewed in 2010 and 2011. Thirteen variables were selected from a self-applied questionnaire. Confirmatory factor analysis was used to investigate whether violence is a uni-or-multidimensional construct consisting of psychological, physical and sexual dimensions. The mean-and-variance-adjusted weighted least squares estimator was used. Models were fitted separately for each city and a third model combining data from the two settings was also tested. Models suggested from modification indices were tested to determine whether changes in the WHO VAW model would produce a better fit.
Results
The unidimensional model did not show good fit (Root mean square error of approximation [RMSEA]  = 0.060, p<0.001 for the combined model). The multidimensional WHO VAW model showed good fit (RMSEA = 0.036, p = 0.999 for the combined model) and standardized factor loadings higher than 0.70, except for the sexual dimension for SL (0.65). The models suggested by the modification indices with cross loadings measuring simultaneously physical and psychological violence showed a significantly better fit compared to the original WHO model (p<0.001 for the difference between the model chi-squares).
Conclusions
Violence is a multidimensional second-order construct consisting of psychological, physical and sexual dimensions. The WHO VAW model and the modified models are suitable for measuring violence against pregnant women.

背景：孕期暴力筛查是预防针对妇女虐待的策略之一。由于暴力行为难以量化，亟需验证可有效衡量该现象的问卷。本研究针对世界卫生组织针对妇女暴力问卷（World Health Organization Violence Against Women, WHO VAW），分析其用于测量孕期妇女所遭受暴力的心理测量学特性。
方法：本研究采用巴西里贝朗普雷图与圣路易斯出生队列研究（Brazilian Ribeirão Preto and São Luís birth cohort studies, BRISA）的数据。研究样本包含来自圣路易斯的1446名孕妇、以及来自里贝朗普雷图的1378名孕妇，受访时间分别为2010年与2011年。从自填式问卷中选取13个变量，采用验证性因子分析，探究暴力属于单维构念还是由心理、躯体与性暴力维度构成的多维构念。分析采用均值与方差校正加权最小二乘估计量，分别针对两座城市拟合模型，并测试整合两地数据的第三种模型。同时基于修正指数提出模型变体，以检验针对WHO VAW原模型的调整是否可提升模型拟合效果。
结果：单维模型未展现出良好拟合效果（整合模型的近似误差均方根（Root Mean Square Error of Approximation, RMSEA）=0.060，p<0.001）。多维版WHO VAW模型则展现出良好拟合效果（整合模型RMSEA=0.036，p=0.999），且标准化因子载荷均高于0.70，仅圣路易斯样本的性暴力维度载荷为0.65除外。基于修正指数提出的、包含同时测量躯体与心理暴力交叉载荷的模型变体，相较原WHO模型拟合效果显著提升（模型卡方值差异p<0.001）。
结论：暴力属于由心理、躯体与性暴力维度构成的多维二阶构念。WHO VAW原模型与经调整的模型变体均适用于测量孕期妇女所遭受的暴力。