Homelessness and Domestic Violence: Examining Patterns of Shelter Use and Barriers to Permanent Housing

Purpose: Domestic violence (DV) is a primary risk factor for women’s homelessness. Homeless DV victims are a vulnerable group with limited options. They often turn to DV shelters and homeless shelters for safety and temporary housing (NNEDV, 2011; Wong, Culhane, & Kuhn, 1997). Although ample research on female homelessness exists, research specifically about DV victims’ patterns of shelter use – both within and across shelter systems – is almost non-existent. The author presents study findings that begin to address this gap. Guided by bounded rationality (March & Simon, 1958) and feminist theory (Reinharz, 1992), study aims were to determine patterns of shelter use among DV victims, assess if differences in individual and shelter-use variables exist among DV victims according to type of shelter type used, and determine if these variables are associated with shelter users’ risk of a repeat stay. 

Methods: A secondary analysis of administrative shelter records from all homeless and DV shelters in the city of Columbus, Ohio between January 2003 and June 2006 was conducted. Homeless shelter data were from the Homeless Management Information System (6,966 women, men, and children in family shelters; 3,217 women in single-women shelters). DV shelter data were abstracted from client intake and exit forms (1,269 women and children). Inclusion criterion was being a female DV victim. Data analysis included descriptive statistics to assess return stays, bivariate statistics (Kruskal-Wallis and Wilcoxon rank tests; Chi-square tests) to compare individual and shelter-use characteristics according to shelter type, and Cox proportional hazards regression to determine risks for repeat shelter stays (using a reference shelter stay of at least 7 days in the year 2004 and followed by 30 days out of shelter).

Results: A total of 1,114 DV victims had at least one stay in a DV, family, or single-woman shelter. Most DV shelter users had only one shelter stay compared to 42% of family-shelter users and 39% of single-woman shelters users. Those who returned often used the same type of shelter; cross-system use occurred among a considerable minority of all shelter users (20%).  As a group, shelter-using DV victims were about 34 years old, more often Black, without income, and in shelter for about one week. However, bivariate comparisons indicated that individual-level (e.g., race, employment) and shelter-use characteristics (e.g., length of and time between stays) differed significantly when shelter users were examined according to the type of shelter they used. Cox regression revealed that having two or more children and receiving income from a source other than employment were significantly associated with a decreased risk of shelter return. 

Conclusion and Implications: Findings suggest that homeless DV victims use all available emergency shelters, are a heterogeneous group, and face considerable barriers. By increasing the capacity and the housing-related assistance of DV shelters, more DV victims could be served within the system developed specifically for them. In addition, on-going federal commitment in the form of financial support to individuals and housing providers can help achieve the structural efforts necessary to prevent homelessness among DV victims.