Children's Exposure to Interparental Violence and Conflict: A Latent Class Analysis

Background: The majority of studies examining how intimate partner violence (IPV) impacts child adjustment have assessed childhood exposure to IPV as a unidimensional construct. However, research suggests that multiple dimensions of children’s exposure to IPV (i.e., severity, type, frequency, and proximity) are important to consider when assessing the potentially deleterious impact of home violence exposure on child outcomes. While children’s exposure to IPV is a multidimensional experience, to our knowledge, no study to date has employed a model-based, person-centered analytic approach to examine whether there are subgroups of children who can be distinguished by distinct patterns of exposure to IPV. The current study utilized latent class analysis (LCA) to identify patterns of children’s exposure to interparental violence and conflict, and explored how individual and contextual factors are associated with those patterns.

Method: Participants included 291 mothers and one of their children (age range 7-12; mean age= 9.07, SD=1.60; 47% female; 78% ethnic/racial minority) who were receiving services from a domestic violence agency. LCA was conducted using 10 items from the Child Exposure to Domestic Violence Scale (CEDV; Edleson et al., 2008), which assess children’s exposure to behaviors such as: mother and partner arguing, partner physically injuring mother, and partner’s use of weapons against mother. Associations of child-, maternal-, and family-level factors with the latent classes were examined using the modified three-step multinomial logistic regression procedure in Mplus 7.3.

Results: Fit indices (i.e., AIC, BIC, adjusted BIC, LMRT, BLRT) resulting from the latent class models containing 1 through 5 classes indicated that the 4-class solution was the optimal model. The classes were labeled as follows:  Low/no IPV Exposure with Verbal Conflict (n= 44; 15%), Emotional/Psychological IPV Exposure (n=93; 32%), Physical and Psychological IPV Exposure (n= 124; 43%), and Severe Multi-type IPV Exposure (n=30; 10%). Results of the multinomial logistic regression identified several individual and contextual factors that are associated with class membership. Of note, Latino children, when compared to their non-Latino White peers, were 3.72 times more likely to be in the Physical and Psychological IPV exposure class than the Low/No Exposure with Verbal Conflict class (p=.03). Moreover, as the severity of maternal psychological abuse increased, the odds of a child being in the Emotional/Psychological (OR= 1.08, p<.01), Physical and Psychological (OR= 1.08, p<.01), and Severe Multi-type (OR= 1.07, p<.01) IPV exposure classes increased when each class was compared to the Low/no IPV exposure class.

Conclusions: Results of this study suggest that there is heterogeneity of IPV experiences among children recruited from community-based IPV services. The plurality of children in the sample (43%) were characterized by exposure to both physical and psychological IPV, and Latino children were more likely to be dually exposed to these dimensions. In addition, our findings suggest that the severity of psychological abuse experienced by women is important to consider when assessing children’s risk for exposure to severe acts of interparental violence. We discuss the implications of our findings for future research and child-centered, trauma-informed interventions.