An Examination of Race and Gender in Predicting Housing and Time to Housing for Individuals and Families Experiencing Homelessness

Background and Purpose: In partnership with the City of Knoxville, through an Emergency Shelter Grant COVID-2 grant, Knoxville Homeless Management Information System (KnoxHMIS) coordinated entry system (Coordinated Housing Assessment and Match Plan-CHAMP) analyzed data for evidence of racial parity in the match and referral process. The purpose of this study was threefold: (1) to determine predictors of housing for individuals and families experiencing homelessness, (2) to determine predictors of time to housing, and (3) examine the effect of possible racial and gender disparities on housing outcomes.

Methods: This study utilized one year’s KnoxHMIS CHAMP data (n=1,290) that consisted of individuals and heads of households ranging in age from 18 to 79 (M = 42.68, SD = 13.65). A chi-square goodness-of-fit examined the relationship between race and housed status. A two-way (factorial) analysis of variance (ANOVA) was conducted to determine whether there was an interaction effect between race and gender on the Vulnerability Index - Service Prioritization Decision Assistance Tool (VI-SPDAT) scores. A binary logistic regression was performed to ascertain the impact of other factors that may facilitate or attenuate housing status. A one-way between-subjects ANOVA was conducted to determine if there was a statistically significant difference in the average time (days) to housing by race. A multivariate Cox regression was performed to assess the relative disparate influence of several variables on time to housing.

Results: Analysis of racial disparity suggests that gender – not race - when accounting for head of household status, was a significant factor in being housed. Black female heads of household families and White female heads of household families were significantly more likely to be housed compared to White male heads of household families (reference group). Results also indicated that a statistically significant relationship existed between days to housing and veteran status (p < .001), age (p < .001), being Black female individual (p = .006), and the presence of a prioritization score (p < .001). Those cases assigned a prioritization score were housed 1.5 times faster than cases that were not given a prioritization score. Results indicated that the VI-SPDAT instrument produced higher scores for White cases overall, with Black female and Black male cases (individuals and families) having comparatively lower scores than White females and White males.

Conclusions and Implications: Despite the clear racial bias in VI-SPADAT scores, housing placement and time to housing were not adversely affected by either racial or gender status in this study. These two housing outcomes are a function of multiple factors, some of which were not captured in the CHAMP data. An alternative, unbiased measure of housing vulnerability is needed, as we and other researchers have found racial bias in this triage assessment tool (Cronley, 2020). Future research should look at housing availability and affordability within individual Continuums of Care, exploring racial equity in causes of homelessness, voucher provisions, and housing placement location.