September 12, 2018
DC Homeless Population: Gender and Race
Background
- In a Coordinated Entry System (CES), people experiencing homelessness are assessed for vulnerability and barriers to housing–and prioritized for housing interventions accordingly.
- There have been questions about whether black people receive a lower prioritization score compared to White people, as measured by the vulnerability assessment tool (VI-SPDAT).
- Process of prioritizing for housing intervention based on this assessment tool could lead White people to be more vulnerable than Black clients.
- In addition, the Community Partnership has repeatedly heard that implementation of the VI-SPDAT or tools that are based off of simlar theories of vulnerability leads to a prioritization of White people towards housing resources–particularly permanent supportive housing.
If this is true, the VI-SPDAT based on similar theories of vulnerability have an embedded bias that leads to disparities in housing prioritization.
VI-SPDAT
Vulnerability Index - Service Prioritization Decision Assistant Tool
- A survey administered both to individuals and families to determine risk and prioritization when providing assistance to homeless and at-risk of homelessness persons.
- A pre-screening, or triage tool, that is designed to be used by all providers with a community to quickly assess the health and social needs of homeless persons and match them with the most appropriate support and housing interventions that are available.
- Designed to determine the presence and acuity of an issue and identify clients to refer for assessment for specific housing interventions (not intended to provide a comprehensive assessment of each person's needs).
- Designed to measure acuity and risks to housing instability.
VI-SPDAT
Domains
| VI-SPDAT Domain | Number of Items | Score | Housing Recommendations |
|---|---|---|---|
| Pre-Survey | 1 | 1 | Score: Recommendations |
| History of Housing and Homelessness | 3 | 2 | 0-3: No housing intervention |
| Risks | 6 | 4 | 4-7: An assessment for Rapid Re-Housing |
| Socialization and Daily Functioning | 5 | 4 | 8+: An assessment for Permanent |
| Wellness | 12 | 6 | Supportive Housing/Housing First |
| Total | 27 | 17 |
VI-SPDAT Uses in U.S.
Housing Type
Proportion of Housing Type
By Gender
| Gender | PSH (N) | TAH (N) | PSH (%) | TAH (%) | Total (%) |
|---|---|---|---|---|---|
| Female | 190 | 235 | 44.7 | 55.3 | 49.2 |
| Male | 296 | 126 | 70.1 | 29.9 | 48.8 |
| Transgender | 13 | 3 | 81.2 | 18.8 | 1.9 |
| Unknown/Refused | 1 | 0 | 100.0 | 0.0 | 0.1 |
| Total | N=500 | N=364 | 100% |
Housing Type
Proportion of Housing Type
By Race
| Race | PSH (N) | TAH (N) | PSH (%) | TAH (%) | Total (%) |
|---|---|---|---|---|---|
| African-American | 354 | 205 | 63.3 | 36.7 | 64.7 |
| White/Caucasian | 26 | 2 | 92.9 | 7.1 | 3.2 |
| Amer.Indian/Alaska Native | 5 | 0 | 100.0 | 0.0 | 0.6 |
| Hispanic/Latino | 5 | 2 | 71.4 | 28.6 | 0.8 |
| Other | 1 | 2 | 33.3 | 66.7 | 0.3 |
| Unknown/Refused | 109 | 153 | 41.6 | 58.4 | 30.3 |
| Total | N=500 | N=364 | 100% |
Housing Type
by Gender and Race in HTH
Test for Potential Bias of VI-SPDAT
DIF (Differential Item Functioning)
- DIF, or item bias, occurs when individuals in a focal group respond differently to a test item than do individuals in a reference group even when comparisons are restricted to individuals with similar overall levels on the trait in question.
- Typically, the total score on the instrument is used to stratify individuals in both focal and reference groups into a number of score-equivalent groups.
- If an item is more or less 'difficult' for one group than another when the comparisons are between subgroups with equal overall scores, then the item is said to function differentially for these groups.
Bias (DIF) Test
Mantel-Haenszel (MH) Odds-Ratio
- Odds Ratio for an individual item at a single ability interval.
\[\alpha = \frac{A_{k}D_{k}}{B_{k}C_{k}}\]
- Common Odds Ratio across all ability intervals k for a specific item.
\[\alpha_{MH} = \frac {\sum_(\frac{A_{k}D_{k}}{N_{k}})} {\sum_(\frac{B_{k}C_{k}}{N_{k}})}\]
Data
Data Description
- Obtained from HMIS (Homeless Management Information System)
- Date Tested between January, 2005 and July, 2018.
- Gender: Female and Male Only.
- Race: White and African-Americans only.
- Total N = 10,031
Preliminary Result from MH Test for Gender
Items Favoring Male Over Female
Preliminary Result from MH Test for Race
Items Favoring White Over Black
Findings
- DIF analysis gives an idea of where potential bias occurs based on group membership (e.g., Gender or Race, etc.).
- DIF analysis of VI-SPDAT showed some items favor one gender group over the other group, and one racial group over the other group.
- Based on bias test of an instrument, we can re-consider how we prioritize for housing interventions.