Using Homeless Management Information System Data to Identify Client-Level Factors Associated with the Geographic Mobility of Homeless Service Users

Background and Purpose: Past research on the geographic mobility of people experiencing homelessness has primarily focused on describing who moves, how often, and for what reasons. These studies predominantly use self-report survey data regarding distance traveled or the frequency of inter-city moves. Past studies have not used administrative data to examine the mobility of people experiencing homelessness. Homeless Management Information System data, which records service admissions in communities receiving federal homelessness funding, offers the opportunity to ask mobility-related questions that address entire service systems, counties, and regions. It is especially useful for homelessness research because homeless individuals lack a fixed, permanent address and therefore connections to shelter and housing programs serve as location indicators.

Methods: Homeless Management Information System data from 17 counties in New York State was used to examine the relationship between client-level characteristics and geographic mobility. Mobility was operationalized as accessing emergency shelter, rapid rehousing, transitional housing, or supportive housing outside the county in which the person was most recently housed—indicating a cross-county move. The final sample included 3658 adults who accessed at least one of these four services in 2017. The sample was 67% male, 38% Black/African American, 12% Hispanic/Latino, and the average age was 41. Twenty percent of service users were accessing a service outside the county in which they were last housed. Chi-square tests of independence were used to identify variables to include in a multivariate model. A series of exploratory logistic regressions were then specified to develop a final model predicting mobility.

Results: The results of the final model showed that the odds of accessing a service outside the county of last residence were lower for service users who were older (OR=.99), Black/African American (OR=.79), Hispanic/Latino (OR=.73), receiving public assistance (OR=.43), and those who had a substance use problem (OR=.75). In contrast, veterans were more than twice as likely (OR=2.52) to be accessing a service outside the county in which they were last housed. These results were presented to homeless service providers in the study region and used to inform data-driven discussions and system planning decisions aimed at improving service access and engagement.

Conclusions and Implications: Findings are consistent with previous research indicating that age, race, and substance use are predictive of mobility for people experiencing homelessness. The impacts of veteran status and receiving public assistance are novel findings and suggest new areas of study and policy analysis. Examining geographic mobility in the context of service use raises questions about the mechanisms of mobility. For example, do individuals who are older, Hispanic/Latino, Black/African American, substance users, or public assistance beneficiaries have less desire or means to make cross-county moves? Alternatively, do these factors function as barriers to mobility or barriers to accessing services in other counties. Further, what drives cross-county mobility for homeless veterans? Possible answers to these questions are discussed. And, suggestions are offered for using these findings, and Homeless Management Information System data more broadly, to effect system change to improve service access.