Skip to content
ParamergeParamerge
Stephen Lieberman
Through Paramerge
Contact

PAN Lab example

VI-SPDAT

The standard nobody validated: a homelessness triage score

The VI-SPDAT questionnaire ranked people experiencing homelessness for housing. It spread to dozens of U.S. states before anyone tested it. Its co-creator, OrgCode, withdrew it.

See more

The Vulnerability Index-Service Prioritization Decision Assistance Tool (VI-SPDAT) is a yes/no questionnaire released in 2013 by OrgCode Consulting and Community Solutions. Homeless-services workers in many U.S. communities asked its questions in an interview and added the answers into a score. Fixed cut-offs then steered each person toward a level of housing help.

How it is used

Coordinated entry is the shared process a community uses to assess people experiencing homelessness and prioritize them for housing. Provider agencies in Continuums of Care, the local networks of homeless-services agencies, gave the VI-SPDAT. The U.S. Department of Housing and Urban Development (HUD) pushed Continuums of Care toward a standardized way to prioritize people. It did not mandate a specific tool.

The VI-SPDAT became the industry-leading coordinated-entry assessment. By 2015 it was in use in at least 39 states, the District of Columbia, and other countries. It was free and open-source, so many communities used it without OrgCode's training.

An intake worker asks the questions and adds up the answers. Higher vulnerability means more 'Yes' answers and a higher score, 0 to 17 for single adults. Roughly 8 and above points toward permanent supportive housing. A score of 4 to 7 points toward short-term rapid re-housing, and 0 to 3 toward minimal services or none.

The score is saved in the person's record in the community's Homeless Management Information System (HMIS). Stored scores are read back to build the community's prioritization list and order referrals.

Who decides

OrgCode designed the tool as a 'Decision Assistance Tool, not a Decision Making Tool.' The score was meant to prompt further assessment for a tier, not to place anyone automatically.

In practice, nothing in coordinated entry required a worker to disagree with the score. In its December 2020 phase-out statement, OrgCode's central concern was that communities relied on the score alone to allocate housing. The sources report no measure of how often workers overrode it.

What the research found

By 2022 at least three studies had found racial disparities in the scores. A 2019 evaluation by C4 Innovations, commissioned by Building Changes, covered four Continuums of Care. They were King and Pierce Counties in Washington, Multnomah County in Oregon, and the Blue Ridge area in Virginia. It found race predicted 11 of the 16 subscales, the groups of questions scored together. People of color received statistically significantly lower prioritization scores. For single adults, being white was a protective factor.

Bitfocus, a company that makes HMIS software, restated the same data in 2021. It reported that white clients were more than 60 percent more likely than clients of color to receive a high prioritization score.

A 2020 study by Courtney Cronley analyzed a large community sample. White women scored consistently higher on vulnerability than Black women and all men. Being white directly and significantly predicted higher scores. The sources say this bias could mask trauma among Black women and delay their housing.

A 2024 study by the Central Valley Health Policy Institute at California State University, Fresno, looked at 1,369 single-adult assessments in the Fresno-Madera Continuum of Care. Clients of color more often answered 'No' to Risk and Wellness items, which the tool scores as less vulnerable. So their scores fell, and more of them were placed in the less intensive rapid re-housing tier. The study's overall tests of race against recommended housing were not statistically significant. So its clear finding is about individual answers, not a proven gap in placements.

The tool was also unreliable. Most people scored higher when it was given a second time, and scores varied by interviewer and site. Its ability to predict housing outcomes was mixed. It predicted outcomes for the youth version (Rice and colleagues, 2018) but not for single adults in one study (Brown and colleagues, 2018). Another multi-community sample found it did predict them (Petry and colleagues).

The withdrawal

OrgCode announced the phase-out in December 2020. It said the tool 'was never designed using a racial or gender equity lens.' It ended all support at the close of 2022 and said it would not build a replacement. It left that role to 'experts in racial and gender equity.'

The tool and local versions of it remained in use in some Continuums of Care into 2025. Communities are building successors. The Fresno-Madera Continuum of Care commissioned a tool called CESMAT in 2024. In Rochester, New York, Partners Ending Homelessness switched to the Homelessness Assessment Tool on June 2, 2025. Everyone already on its prioritization list had to be assessed again with the new tool.

There is no authoritative national count of how many VI-SPDAT assessments were ever given, because the tool was open-source and run locally. For scale, U.S. shelters serve an estimated 1.4 million people a year.

Advocates and OrgCode both say the tool 'revolutionized' homeless services by replacing pure caseworker discretion with a standard process. So its withdrawal reopened a gap. Both a biased tool and unchecked caseworker judgment carry equity risk.

Two loops to watch

The first loop is the record. A score saved in HMIS is read back to build the prioritization list, so a low score keeps shaping referrals after the interview.

The second loop is repeat interviews. Most people score higher when interviewed again, so who gets re-interviewed can shift who comes first.

What the available tools can and cannot address

A failure pathway is a link between two parts of the network, where a mistake made by one part can be passed on to the other. Here a mistake is, for example, a score that understates how vulnerable a person is. A pathway counts as closed once it passes on only a few mistakes. It need not stop them all.

This case has a budget of 9 units. It starts in cascading failure, where mistakes build on one another, and the housing-assessment work falls behind. Five failure pathways are open at the start: Score sets housing tier, Answers scored, Score saved in HMIS, Answers and score entered, and Stored score orders referrals.

Explore (No Targets) sets no targets. There, and under Service Targets Only, no single tool makes the network self-correcting, where it corrects its mistakes rather than building on them. Escalate checks with Mark AI-written records does it, for 4 units. Under Service Targets Only, that pair also meets the service target. That target asks that the score keep clearly helping the housing-assessment work. Many other sets of tools meet it too.

Under Service and Safety Targets and All Governance Targets, the targets can be met. Both levels also ask you to close every failure pathway and keep the housing-assessment work from falling behind. Five sets of tools meet the targets at each of these two levels. Every one includes Escalate checks, Gate record entries, and Mark AI-written records, which together cost 7 units.

Escalate checks closes Score sets housing tier. Gate record entries requires sign-off before a score or answers are saved, so it closes the two pathways into HMIS. No AI writes records on this network, so here Mark AI-written records marks which HMIS entries came from the questionnaire. Workers reading the list can then tell scores apart from other entries. It closes Stored score orders referrals, and it also closes Answers scored.

The other four sets add one more tool: Require sign-off, Keep skills sharp, Assign a challenger, or Review on schedule. The set of three also meets them with Escalate checks at its stronger setting, for 9 units, or with Mark AI-written records at its stronger setting, for 8. The set with Require sign-off also meets them with Mark AI-written records at its stronger setting, for 9. Check with a second model, Review the riskiest first, and Upgrade model are in no set that meets these two levels. Check with a second model starts the validation and equity check, but it closes no pathway.

More is not better here. Every tool at its strongest setting at once closes every failure pathway, but costs 37 units, far over the budget. It also leaves the score no longer clearly helping the housing-assessment work. So it misses the targets at every level that sets them.

Stylized model of a documented deploymentHousing & homelessness services

Open this example in PAN Lab v0.1 to apply pressures and levers and watch what the system does.

What this models

This example runs on the VI-SPDAT-class self-report triage score network: 5 components and 10 pathways between them. Every context in the Lab is a stylized model, never a reconstruction of any actual deployment, and each assumption behind it carries a provenance label.

Evidence base: 5 assumed · 2 published baseline. In the Lab, the shaded evidence band behind each headline readout draws its width from the least-established class below.

Show all 7 assumptions
  • assumed

    This network follows the pattern of self-report housing triage documented in the VI-SPDAT case file. It does not rebuild the actual tool. It shows the tool as communities used it, before OrgCode began phasing it out in December 2020 and ended support at the close of 2022.

  • baseline

    The network assumes no one checked the questionnaire itself as it became the standard in at least 39 states. No validation, reliability, or equity test ran first. OrgCode said the tool was never designed using a racial or gender equity lens.

  • assumed

    The network includes the human override the design intended, but assumes it was not in use. OrgCode called the tool 'Decision Assistance, not Decision Making.' Its central concern at the phase-out was that many communities relied on the score alone.

  • assumed

    The tool reads no records from HMIS or any other agency. Its only input is the interview. So the network has no pathway from the records into the score. Instead, stored scores are read back from HMIS to build the prioritization list and order referrals. Repeat interviews, which tend to raise scores, are a second documented way past assessments shape priority.

  • baseline

    The network places the documented bias where the interview answers enter the score. People of color more often answered 'No' to questions about risks and wellness, which lowers the score. The bias is in how the questions ask and which ones count, not in a learned weight. So upgrading the score's arithmetic does least to reduce the bias.

  • assumed

    The community prioritization list is shown between the score and the intake workers, because the tool's output is a ranked list. It does not change how mistakes pass.

  • assumed

    The documented racial scoring disparity is recorded in the case file, not computed here. This example shows how mistakes pass among the score, workers, and records. It does not model groups of people or estimate unequal harm to the people experiencing homelessness whom the score ranks for scarce housing.

What this example does not show

Show all 3 limitations
  • Bias enters here the way institutional failures do: in interviews, records, and prioritization lists. This example shows only that. It includes no demographics and estimates no unequal harm to the people experiencing homelessness this score ranks. The documented racial scoring disparity is recorded in the case file: four Continuums of Care (2019), a large community sample (2020), and Fresno-Madera (2024). The newest single-community study's overall tier-assignment tests were not statistically significant.
  • A housing allocation is a one-time rationing decision, not a mistake that passes on. So this example shows the validation and override checks that were not in use, and the biased score kept on the prioritization list. It makes no claim about any one person's outcome.
  • This is a historical example, from 2013 to 2022. OrgCode withdrew the tool, and communities are replacing it one by one. It is kept as an example of a decision-assistance score used as the decision-maker, not as a picture of any community's current system.

Sources and evidence

What this example rests on, claim by claim. Every entry resolves to the same ledger the Evidence Registry publishes.

  • The VI-SPDAT was the dominant U.S. homelessness triage assessment for roughly a decade, adopted in at least 39 states and the District of Columbia by 2015, before its own creators announced its phase-out in December 2020 on equity grounds; a 2019 commissioned racial-equity evaluation across four Continuums of Care found race predicted 11 of 16 subscales and that people of color received statistically significantly lower prioritization scores.

    empirical
    • Advocacy National Alliance to End Homelessness / Homelessness Research Institute (Joy Moses and Ann Oliva), Looking Back at the VI-SPDAT Before Moving Forward (2022) https://endhomelessness.org/wp-content/uploads/2022/08/NextGenTools_VISPDATBrief_08-30-22.pdf
    • Vendor OrgCode Consulting (Iain De Jong), A Message from OrgCode on the VI-SPDAT Moving Forward (2020) https://www.orgcode.com/blog/a-message-from-orgcode-on-the-vi-spdat-moving-forward
    • Academic C4 Innovations (Wilkey, Cannon, Donegan, Yampolskaya), commissioned by Building Changes, Coordinated Entry Systems: Racial Equity Analysis of Assessment Data (2019) https://homelesshub.ca/resource/coordinated-entry-systems-racial-equity-analysis-assessment-data/

  • The VI-SPDAT showed poor test-retest reliability, with most participants scoring higher on re-administration, and poor inter-rater reliability, with scores varying by interviewer and site; its predictive validity for housing outcomes was mixed across studies, positive for the youth version, null for single adults in one study, and positive in another community sample.

    empirical
    • Vendor Bitfocus, Going Beyond the VI-SPDAT: Deficiencies of the VI-SPDAT (2021) https://www.bitfocus.com/blog/deficiencies-of-the-vi-spdat
    • Advocacy National Alliance to End Homelessness / Homelessness Research Institute (Joy Moses and Ann Oliva), Looking Back at the VI-SPDAT Before Moving Forward (2022) https://endhomelessness.org/wp-content/uploads/2022/08/NextGenTools_VISPDATBrief_08-30-22.pdf
    • Academic Shinn and Richard, Allocating Homeless Services After the Withdrawal of the Vulnerability Index-Service Prioritization Decision Assistance Tool (American Journal of Public Health, 112(3):378-382, 2022) https://pmc.ncbi.nlm.nih.gov/articles/PMC8887175/

Where this connects

Institutional pressures in this domain

  • Workload surge — Demand outruns staffing; per-case attention shrinks and review becomes triage.
  • Austerity & recovery incentives — Cost-cutting and overpayment-recovery targets tilt the system toward denial and enforcement errors.
  • Vendor opacity — The deploying institution cannot inspect the model, data, or update pipeline it is accountable for.
  • Data & policy drift — The world, the intake process, and the rules change under a system trained on how things used to be — two mechanisms with different remedies: the statistical properties of what the system processes move (concept drift), or the mixture of inputs arriving in deployment differs from the mixture it was trained on (covariate shift).
  • Compliance over substance — Paper controls (sign-offs, checklists) satisfy audits while the behavior they describe erodes.

All of them in context on the Housing & homelessness services domain page.

Levers available here and the patterns behind them

Documented case histories