The relationships between community context and entry into a homeless shelter system

ABSTRACT

This study seeks to understand the likelihood of entering the homeless shelter system as reflected in differential levels of community capital, human service support features, and levels of criminal activity. This study used a sample of adults entering the City of Philadelphia-funded Emergency Housing System between February 2008 and January 2010. We linked former addresses to census block groups and constructed a rate of homeless system entrants per 1,000 residents in each block group. After regressing this measure on a series of demographic, socioeconomic, and other salient community characteristics, our preliminary results indicate that basic demographic characteristics of communities and the capacity of drug treatment centers provide predictive value regarding variations in rates of homelessness. We also conclude that both property and violent crime rates at the block group level are shown to be significantly associated with entrance into the homeless shelter system.

Notes

1. Because of limitations in the collection of data on the nonsheltered (street) homeless and marginally housed (e.g., doubling up), this analysis focuses on those people who enter the shelter system.

2. An important limitation of this data stems from the fact that the length-of-stay information for these prior addresses is unknown and therefore we are unable to differentiate between long-term residency at an address and temporary prior accommodations.

3. Many of the invalid addresses included terms such as null, unknown, and streets. Some cases had particular facilities listed as their prior address, such as correctional facilities, which were able to be mapped after replacing the names with the proper street addresses.

4. This hit rate is well above the minimum acceptable hit rate of 85% determined by Ratcliffe (2004), which we believe is appropriate to reference in this context.

5. The repeat users utilized OSH resources between two and seven times in the 2-year period, with the vast majority entering the system twice.

6. Due to the removal of more than half of the original cases provided by OSH for this analysis, a descriptive comparison of demographic characteristics between the cases that were removed and that remain was undertaken. Chi-square tests between the two groups and both gender (male and female) and race (White and non-White) and a t test between the two groups and age indicate no significant difference between the cases removed and those that remain to be analyzed.

7. HUD defines an HMIS as “a local information technology system used to collect client-level data and data on the provision of housing and services to homeless individuals and families and persons at risk of homelessness” (HUD, 2012). HUD compiles the core data from each CoC and publishes the AHAR. In the same guidelines, HUD indicates that “planners and policymakers use aggregate HMIS data to better inform homeless policy and decision making at the federal, state, and local levels.”

8. The exact names and program types were not provided with the data received from the Philadelphia Police Department, other than differentiating between substance abuse sites and halfway houses that are supervised by a court office.

9. Although individuals who resided in shelters and correctional facilities prior to their entrance into the OSH data set were removed from our analysis, we chose to allow the individuals who listed their prior addresses as drug treatment centers and halfway houses to remain in our analysis. First, most of those facilities are not inpatient facilities and, therefore, allow for individuals to live within, and potentially be affected by, the community. Secondly, we believe that their inclusion allows for the accurate measurement of the effects of those types of facilities on entrance into the OSH system.

10. The maximum rate of 61.11 in one block group was nearly 20 points higher than the corresponding rate in the next highest block group. We conducted the subsequent multivariate analyses both with and without this block group to determine whether this outlier would unduly influence the results. We found that they were substantively the same, so we proceeded to report our analyses that include all 1,327 block groups as previously described.

11. To confirm that these items are not autocorrelated, we ran exploratory factor analysis to determine whether any indices comprised of community contextual items could be constructed. The subsequent Cronbach’s alpha for one potential item (this item included the proportions of Black residents, males between ages 25 and 64 not in the labor force, and those living in poverty), however, was rather low and did not indicate an acceptable level of internal reliability for the item. As a result, all items are included separately in the multivariate analyses.

12. A separate issue might arise if one researched the housing history of people who have been residents in one of the shelters in the city, namely, the relative degree to which people have been both in prison and in a shelter over time. We think that this is an interesting research topic but is beyond the scope of our current project.

Additional information

Notes on contributors

David Bartelt

David Bartelt was a professor in the Department of Geography and Urban Studies at Temple University. His research focused on barriers to accessing quality housing created by income differentials, the persistence of race-based community concentration, and the uneven aging of the region’s housing ventures. He has been a co-principal on the Metropolitan Philadelphia Indicators Project (MPIP) and has published works in housing, redlining, the development of American cities, and local economic development. He is a co-author of Philadelphia: Neighborhood, Division and Conflict in a Postindustrial City, as well as an extension of that analysis, Restructuring the Philadelphia Region.

Karin M. Eyrich-Garg

Karin M. Eyrich-Garg is an associate professor in the School of Social Work in the College of Public Health at Temple University. Her research focuses on the technology use; health, mental health, and substance use; service use; social support; engagement; and empowerment of impoverished populations—particularly persons experiencing homelessness. Her work has been supported by NIDA, NIMH, CDC, ACF, the City of Philadelphia, and Temple University.

Brian Lockwood

Brian Lockwood is an associate professor in the Department of Criminal Justice at Monmouth University. His research interests include the spatial analysis of crime, juvenile delinquency, and the effects of facilities on patterns of offending. His recent work has appeared in Justice Quarterly, Social Science Research, and the Journal of Research in Crime & Delinquency.