Some communities, especially on the West Coast, have noticed in their data and/or anecdotally have expressed concerns with racial equity and the VI-SPDAT. I am glad they reached out with their concerns as this is good dialogue to have. We should have a tool – and approaches to administering the tool – that are as free from bias and inequity as possible. The two major concerns expressed are, firstly, that non-white people do not score high enough in the VI-SPDAT, and secondly that PSH resources disproportionately go to white people.
These are concerning claims. My go-to in moments like these are to look at the data. For a different project, I just finished compiling data on VI-SPDATs administered from more than a dozen communities from across the country (which includes some west coast communities) representing more than 75,000 VI-SPDAT records. This includes both sheltered and unsheltered. It is a good cross-section of data from urban, suburban and rural communities from all corners of the United States. What does the data show?
I examined the data from all of the VI-SPDAT responses we have from various HMIS and other electronic databases. For the sake of analysis, I categorized people into two groups: Group 1 comprised of people who identify as white, including those that identified as not Hispanic or Latino, and Group 2, using HUD categories of race and ethnicity, which includes people who identify as Black or African-American, Asian, Hispanic or Latino, Native Hawaiian or Other Pacific Islander, and American Indian or Alaska Native. Of the non-white group, African-Americans make up the overwhelming majority.
Data within the sheltered population of adults shows that 75.4% of those surveyed identify as Group 2, with 76.1% of those scoring the highest acuity also being within Group 2. Within the unsheltered adult population, 43% of those surveyed identify as being within Group 2, with 45.6% of those scoring the highest acuity also being within Group 2. Despite what some anecdotes and local data in some cities has suggested, there does not appear to be an issue with the results of higher acuity skewed more towards white people, at least at a national level within the adult sheltered and unsheltered populations.
There is a line of inquiry that this leads to, however, which is why proportions look so different between the sheltered and unsheltered populations – and this is a good discussion to have. For example, is it that Group 2 people are missed in street outreach or is it that people who identify with Group 1 truly make up more than 50% of the unsheltered population? Perhaps communities need to compare the results of the PIT Count data to their VI-SPDAT data to see if the proportions are similar.
When one looks at housing access post shelter, the VI-SPDAT data is illuminating. Remember that 75.4% identified as Group 2. 72.6% of those moving into PSH are Group 2 and 76.4% of those moving into RRH are Group 2. Remember that one of the claims has been that PSH is skewed towards white people. When looking at the national shelter data where VI-SPDAT has been used, that is ever so slightly the case, but not overwhelmingly so. And when it comes to RRH, Group 2 people are ever so slightly more likely to get RRH than Group 1 people. Within Veteran specific programs, 73.6% of people moving using SSVF from shelters are in Group 2. With HUD VASH a different story emerges with 47.3% of people being in Group 2 who move into a HUD VASH program from shelter. A deeper dive is likely required on HUD VASH to see why this is the case.
When one looks at housing access post being unsheltered, the VI-SPDAT data is also illuminating. Remember that 43% of the unsheltered population identified as Group 2. 44.4% of those moving into PSH from the unsheltered population are in the Group 2 category, but only 28.7% of those moving into RRH are Group 2. So, while the PSH access seems proportional, the RRH experience is worth investigating, it would seem. Within Veteran specific programs, 41.1% of those accessing SSVF from being unsheltered are in Group 2, and 42.3% of those using HUD VASH from being unsheltered are in Group 2.
I would encourage all communities to look at their sheltered and unsheltered data through a lens of racial equity and proportional access to services. Perhaps comparing your local results to this national picture would be helpful, but more importantly, at your local level ask yourselves whether the people getting access to the likes of PSH and RRH represent the ethno-racial composition of all people surveyed. If they don’t, you have some investigating to do in order to understand why that is not the case.
As those of you who follow how we create and update the SPDAT Suite of Products know, we take racial equity very seriously. One of the ways we do this is paying for an independent external review of the tools through a lens of an anti-oppression framework. Another way we do this is by ensuring that people with lived experience that participate in the creation and feedback loop on the tool are representative of the population, including race and ethnicity. A third way we do this is by allowing the VI-SPDAT to be translated into different languages. I am not suggesting that any of these are perfect or account for all potential inequitable situations that may arise, but as a white, hetero, English-speaking privileged dude leading this work, I feel it is OrgCode’s duty to be as inclusive and sensitive to issues of inequity as possible. I have blind spots that I need others to shine a light on for me. I may have deeply internalized biases that I am not aware of which requires an external set of eyes to see.
Administration of tools is a key issue that needs to be examined. It is possible that the experience of inequity in some communities is a result of biases on the part of the interviewers. It is possible that those assessing others do not represent or have been insufficiently trained in cultural competency. Or it is possible that the tools are being administered incorrectly. For example, administration of the VI-SPDAT requires a localized introductory script that is followed by all administrators of the tool. It is possible that some of the communities experiencing racial inequity in the results of the tool are not doing this or have inserted language into their script that is problematic.
Some of the communities that have contacted us have suggested we add questions specifically about race and/or add points if someone is non-white. This would be a mistake. Not only would this run into huge issues of Fair Housing, the data as we look at a national level would not warrant this. My discussions with national leaders also suggest this is something that would be frowned upon in Coordinated Entry.
What we can do (and are doing) as part of the update of the VI-SPDAT we are currently working on is to continue to elevate the voice of those in Group 2 in the creation of the questions and the language used in asking the questions. We can ensure that communities that feel there is racial inequity (or where there is actual racial inequity) are invited to participate in the testing of the next version of the VI-SPDAT. We can continue to mine the data to see where biases related to particular questions can be problematic or under-represent the experiences of vulnerability and housing instability of people in Group 2. We can be open to making necessary changes where it is warranted and there is a demonstrated need as shown in the data.
No one is making claims that the VI-SPDAT is perfect. It may not even be the ideal tool for the specific needs of your community. But claims that the VI-SPDAT is skewed towards white people scoring higher or white people getting disproportionate access to PSH are not proven when examining the national data.