The Child Exposure to Domestic Violence Scale: A Rasch Analysis of the Home Exposure Subscale in a Sample of 7- to 12-Year-Old Children

Background: Research on children’s exposure to intimate partner violence (IPV) continues to be limited by a paucity of empirical knowledge of the forms of IPV-related violence to which children are exposed, the frequency and proximity of their exposure, and children’s involvement in IPV-related events (Edleson et al., 2007, 2008). Among the limited number of children’s self-report measures of exposure to IPV, the Child Exposure to Domestic Violence (CEDV) scale was developed purposefully by Edleson and colleagues (2007, 2008) to examine the multidimensional ways children experience IPV. To date, all published studies reporting on the CEDV have provided psychometric parameters estimated using classical test theory-based methodology and many have reported on small samples with insufficient power for rigorous testing of reliability and validity. 

Current Study: In the current study, we report findings from a psychometric analysis of the Home Exposure Subscale of the CEDV, extending what is known about the psychometric properties of the CEDV utilizing item response theory.

Method: Data were collected as part of a larger study on children’s exposure to animal cruelty in families affected by IPV. Participants (N=245) ranged in age from 7 to 12 years (M=9.25; 57% racial/ethnic minority; 54% male). Rasch analysis was conducted using the software, Winsteps (Linacre, 2014). Scale dimensionality, rating scale use, reliability, person and item fit, invariance, and targeting of the measure were assessed. 

Results: After removal of ten misfitting persons, a unidimensional structure was found to adequately fit the data. Rasch principal components analysis of residuals indicated a single latent dimension among the 10 items by Linacre’s (2010) standards. Approximately 63.9% of the variance was explained by the first contrast; the eigenvalue of the first contrast was 1.7. Overall item and person fit statistics further confirmed this structure (outfit and infit mean squares were >.5 and <1.5). Scale use was appropriate with Rasch-Andrich thresholds that increased with category values and no evidence of disordering. Examination of item-person maps indicated the sample was also well-targeted with a person mean of -1.49 and item logit positions ranging from -3 to 3. Person separation was 2.11 with a person reliability of .82, and a Cronbach’s alpha of .85. Our results suggest the measure exhibits invariance across genders; however, evidence of differential item functioning was found across Spanish and English translations for item 8 (“How often has mom’s partner done something to hurt her body?”). 

Discussion: Findings indicate that the CEDV Home Exposure Subscale is an appropriate unidimensional measure of children’s exposure to IPV in the home that can be reliably administered in the context of large-scale community research. Results of the Rasch analysis suggest the measure is well-suited for children ages 7 to 12; however, our findings identify multiple ways the measure could be improved for children in this age range. Most notably, our findings suggest the Spanish translation of the measure needs to be verified and modified for item 8 in a culturally considerate context if the translated version is to be implemented with Spanish-speaking children in future research.