Arizona Housing Analytics Collaborative

Coordinated Entry (CE) Comparison
Compared the performance of two assessment tools used to designate service prioritization scores for individuals and families seeking homelessness services in Arizona: the VI-SPDAT and the newly designed Maricopa Assessment Prioritization (MAP) Tool.
Description
At the request of Maricopa Association of Governments (MAG) Coordinated Entry Assessment Workgroup, this project focused on evaluating and comparing the performance of two assessment tools: the Vulnerability Index-Service Prioritization Decision Assistance Tool (VI-SPDAT) and the Maricopa Assessment Prioritization (MAP) Tool.
While the VI-SPDAT has been in use throughout the state of Arizona and has been repeatedly identified by providers as a potentially problematic tool in designating service prioritization scores for individuals and families facing homelessness, the MAP Tool has only recently been designed in Maricopa County as an alternative to the VI-SPDAT. The goal of the MAP Tool was to address and reduce the racial inequities exhibited by the prioritization patterns contained in the VI-SPDAT, and to instead prioritize individuals and families equitably across racial groups.
This analytic project had multiple aims:
Evaluate the overall sample of individuals and families who engaged in the coordinated entry assessment process to understand their demographic characteristics (e.g., average age, gender identity, racial and ethnic background, chronic homelessness status, disability status, and veteran status).
Evaluate the demographic characteristics of highly prioritized individuals (i.e., top 5% in priority list) and families (i.e., top 10% in priority list) when using the VI-SPDAT and compare these characteristics with those of the overall sample.
Evaluate the demographic characteristics of highly prioritized individuals (i.e., top 5% in priority list) and families (i.e., top 10% in priority list) when using the MAP Tool and compare these characteristics with those of the overall sample, looking for racial and ethnic inequitable representation.
Analyses have led to several actionable insights:
As suspected by providers across the state, the VI-SPDAT indeed showed inequitable patterns of service prioritization, such that White individuals and families were overrepresented in the high prioritization group, while Black and Latin American individuals and families showed underrepresentation in this high prioritization group.
Anomalies were detected in the automated scoring of the MAP Tool. Individuals were erroneously receiving additional points for items intended for families only. Further, it appeared some items were designated the wrong number of points. These errors in scoring caused individuals to be issued vastly higher points than they should have received, and individuals and families who should have been lower in prioritization were artificially being elevated in the list. These errors also appeared to benefit White individuals and families disproportionately, and to artificially de-prioritize or push Black individuals and families further down the priority list.
The MAP Tool showed racially equitable patterns of service prioritization for individuals, such that individuals who are highly prioritized using the MAP Tool are racially and ethnically similar to the overall sample of individuals coming through coordinated entry.
The MAP Tool showed slightly inequitable patterns of service prioritization for families, such that White families were slightly overrepresented and Latin families were slightly underrepresented in the high priority group. Although these disparities were small, they were consistent over time. When we examined a year’s worth of data, this racial/ethnic pattern remained consistent, showing these disparities would yield additive differences over time.
Close examination of all MAP items and how they are endorsed across racial and ethnic groups led to the identification of one candidate item potentially driving these patterns. The chronic homelessness item—which alone lends almost one third of all possible points—was found to overrepresent White families and vastly underrepresent Latin families. In conversations with the MAG Coordinated Entry Assessment Workgroup, we also established that chronic homelessness was embedded within multiple items in the MAP Tool, and thus point reductions for this item are warranted.
Based on these findings, we issued the following recommendations:
Expedite communication with the vendor for the state of Arizona to update the automated scoring for the MAP Tool, thus allowing the tool to be implemented as intended.
Retain the scoring of the MAP Tool for individuals as currently designed, in order to preserve the prioritization patterns observed.
Update the scoring of the MAP Tool for families and specifically reduce the scoring for the chronic homelessness item, so that families who have been homeless for 365 days or longer will receive 50 points for this item alone, instead of the 100 points that were designated originally.
Collaboration and discussion with the MAG Coordinated Entry Assessment Workgroup for this project is ongoing. Updates will be added to this page when available!