January 21, 2021

Racial and Gender Bias in the VI-SPDAT

I recently published a study in which I analyzed the VI-SPDAT scores for over 1,300 adults experiencing homelessness in a mid-sized city in the southeastern U.S. Given my prior research on homelessness, I expected a strong relationship between trauma and homelessness, and I also expected patterns of advantage and disadvantage to influence that relationship. One question on the VI-SPDAT asks whether a person’s homelessness is due to trauma, so I tested whether this single item predicts total VI-SPDAT scores and whether there are differences in this relationship based on gender (female vs. male), race (Black vs. white), and the intersection of gender and race (female and Black, female and white, male and Black, and male and white).

Why are trauma and vulnerability important in homelessness?

Trauma is endemic in the lives of people experiencing homelessness. When it intersects with gender- and race-based disadvantage, the risks for vulnerability escalate. Women face unique risks for homelessness compared to men, based on interpersonal violence, economic inequalities, unintended pregnancy, and single parenthood. Race is also a factor, and Black people comprise nearly half (40%) of all individuals experiencing homelessness in the United States (U.S.) despite making up just 15% of the general population. It is likely, then, that Black women are disproportionately at risk for trauma and vulnerability on the street.

Homeless service providers try to measure “vulnerability” as a central factor in deciding how to allot their limited housing resources.  A single, unvalidated measure of vulnerability is widely used across the U.S. and in Canada to determine who is most in need: the Vulnerability Index – Service Prioritization and Decision Tool (VI-SPDAT). The tool’s origins are murky. The organization that created the VI-SPDAT, OrgCode Consulting, relied heavily on studies conducted with community samples in one city in the northeastern U.S. These samples were largely male, and data were analyzed for rate of death—and not for vulnerability thresholds. OrgCode used these indicators of mortality to measure what it describes as “vulnerability,” and it describes the VI-SPDAT as the “gold standard of assessment tools.”

The U.S. Department of Housing and Urban Development (HUD) has made funding for homeless services in the U.S. contingent on use of a standardized assessment tool within a coordinated entry system. Several other more robust tools are available, for example, the DESC Vulnerability Assessment Tool, which the University of Washington assessed for reliability and validity. Nevertheless, HUD released a state of the evidence report that highlighted the VI-SPDAT (along with two other tools which are much less common).

Since its public release in 2010, the VI-SPDAT has never been analyzed for standard psychometric properties of validation and reliability. Community-level studies, however, show consistent evidence of racial bias and unreliability in its use. My findings add to this growing evidence.

My Research Findings

• Women were twice as likely as men to report being homeless due to trauma.

• White women and Black women had similarly high odds of being homeless due to trauma.

• White individuals were one and half times as likely as Black individuals to report being homeless due to trauma.

• White women were nearly three times as likely to report being homeless due to trauma compared to Black men.

• Overall, white women had the highest vulnerability scores, followed by white men, Black women, and lastly Black men.

• Trauma and being Black predicted higher vulnerability scores.

• Gender and the interaction of gender and race did not predict vulnerability scores.

 Key Takeaways

There is evidence of racial bias in the VI-SPDAT. Both Black and white women reported similarly high odds of homelessness due to trauma, but white women scored consistently higher on the VI-SPDAT. Closer inspection revealed that white women reported more visits to emergency rooms and formal healthcare use compared to Black women. They were also more likely to report engaging in risky behaviors like survival sex. It is possible that Black women are disproportionately at risk for trauma on the street, but are scoring lower on the VI-SPDAT, because they are less likely to use formal healthcare and reluctant to report illicit behavior. Given historic racism, minority communities’ distrust of health systems, and discriminatory policing practices in the U.S., such an explanation seems logical.

If biases persist within the assessment and coordinated entry process, over time more Black people who are experiencing homelessness will be without housing for longer periods of time compared to their white counterparts.

HUD continues to encourage use of the VI-SPDAT, and it is widely used, despite growing evidence of unreliability and now racial, and potentially gender bias. Communities must reconsider use of the VI-SPDAT for housing decisions. Revising the VI-SPDAT is insufficient. Its primary items are based on mortality risks, not vulnerability, and reflect experiences of people, largely male, living unsheltered in northern climates. Instead, researchers and policy makers should build new vulnerability assessments, perhaps with consideration to geographic, racial, and gender disparities in experiences of homelessness. Finally, greater advocacy from researchers and community advocates is needed to pressure HUD to promote alternatives to the VI-SPDAT and to acknowledge its flaws.


The analysis and interpretations contained in this blog post are those of the individual contributors and do not necessarily represent the views of the Canadian Observatory on Homelessness.

Disclaimer
The analysis and interpretations contained in these blog posts are those of the individual contributors and do not necessarily represent the views of the Canadian Observatory on Homelessness.