PURPOSE: Little is known about how combinations of assets and vulnerabilities (e.g., mental health, physical functioning, and social support/strain) influence battered women's help-seeking, escape, or protective acting, and less is known about how to target interventions to combinations of co-occurring problems that may substantially affect battered women's capacity to self-protectively cope and seek aid. Using person-centered analyses, the aims of this investigation were 1) to examine the heterogeneity of battered women's biopsychosocial assets and vulnerabilities, testing for coherent subgroups on the basis of these concomitant factors, 2) to test for differences among these groups as to sociodemographic characteristics of the women as well as their help-seeking, escape, and protective actions subsequent to a serious incident of domestic violence, and 3) to consider the implications of these findings for partner violence interventions.

METHODS: This investigation analyzed an urban sample of battered women (n= 448) following an incident of partner violence that triggered legal involvement. Stratified random sampling was used to recruit participants, and data were collected through survey questionnaires, using several well-evaluated measures. Latent class analyses (using M+, version 3) were used to determine multivariate participant profiles from measures of depression, physical health functioning, injury, both positive and negative social support measures, and the Women's Experience of Battering scale. Classes established through these multivariate combinations were then tested as to differences in sociodemographics, help-seeking, and escape/protective actions.

RESULTS: Multiple latent class solutions were iteratively examined. The fit statistics showed that the five-class model best fit the data (BIC= 13888.67)(Muthen & Muthen, 1998). Preliminary MANOVA analyses wee used as omnibus tests of differences to guard against Type I error. Differences among the classes on help-seeking scales (F=3.77, p< .001) and on escape/protective actions (e.g., leaving the relationship, no contact/protection orders) both at time of the reporting incident (F=4.67, p< .001) and 3 months subsequent to incident (F=5.47, p< .001) all achieved significance. Follow up post-hoc tests on the individual variables also achieved significance. However, none of the sociodemographics (race, income, education, employment) achieved significance, tested on the basis of chi-square tests (X2=16.19, 17.51, 16.72, 17.33, respectively, n.s.).

IMPLICATIONS FOR PRACTICE: Offering a unique analytic perspective, person-centered analyses provide tools to distinguish meaningful subgroups and enable the translation of descriptive profiles into targeted interventions. Findings indicate that social workers should use targeted screening, assessment and intervention strategies that match battered women's profiles of combined needs and strengths. Although sociodemographic factors play critical roles in women's exposure to and protection against domestic violence, these findings indicate profiles of vulnerability on social and psychological dimensions that apparently cut across social stratification. Implications for partner violence interventions that promote women's safety by matching interventions to combinations of needs and strengths are discussed.

Muthen, L. K., & Muthen, B. O. (1998). Mplus user's guide. Los Angeles, CA: Muthen & Muthen.