Communities have employed many strategies for identifying the set of families described above—data matches, vulnerability assessments and/or case conferences are most common.
Cross-System Data Matching
Most public agencies have data collection systems and can identify families who have had contact with child welfare and homeless systems. However, these systems are largely uncoordinated. Cross-system administrative data matches (for example between child welfare agency data systems and Homeless Management Information Systems) may be used to identify families who overlap and have frequent contacts with multiple systems. However, feasibility of a data match depends on the breadth of data collected, quality of the data, and the sophistication of data systems. In some cases, to capture richer data on patterns of service use (e.g., multiple stays in shelter or multiple child were reports, substance abuse, domestic violence, etc.), public agency staff may have to manually go through case records. In the most sophisticated data systems, a human services agency may be able to track families across multiple departments of the agency (e.g., child welfare, homeless, TANF, mental health, etc.). In this scenario, public agencies or departments have Memorandums of Understanding (MOUs) that allow staff from any of the partnering agencies to view all of the data pertaining to a family. This is rare and even with this complex data matching process, additional screening may be required. However, if data and systems permit, this matching approach is particularly effective and efficient for identifying a community’s highest-need/highest-cost families to target for intensive services.
Once the targeted families are identified through the data match, locating and engaging clients can be a challenging process and will most likely require a multi-agency collaborative effort. For example, the list can be shared between a number of collaborating agencies (child welfare agencies, shelters, jails, etc.) that work together to locate, engage and recruit families for participation in the program. When using this method to target specific individuals/families for recruitment into a program, limiting the geographic scope of the match (to city, county, region) may be best to effectively coordinate the location, outreach and recruitment process.
Data-driven Targeting Tools and Vulnerability Assessments
Whether a data match is possible or not, data-driven triage tools or evidence-based vulnerability assessments can help identify and prioritize families for supportive housing. The more sophisticated tools are based on statistical models or predictive algorithms that use information (demographics, homelessness, child welfare involvement) collected at the time of screening to identify families at highest risk for chronic child welfare involvement and long-term homelessness. One such tool is currently being developed by the Urban Institute and CSH using administrative data from five different communities involved in a national demonstration project targeting supportive housing for high-need, homeless families involved with the child welfare system. Vulnerability assessments work in a similar fashion in that they collect information from respondents across a number of different domains (health/mental health, homelessness, child welfare involvement, etc.) to produce an aggregate vulnerability score or rating that is then used to prioritize families for services. CSH along with a group of housing stakeholders in Chicago, Illinois developed a Family Vulnerability Index to prioritize families living in homeless shelters that were at-risk of interaction with, or currently involved in, the child welfare system. The tool was piloted in several different shelters and will soon be applied to Chicago’s centralized supportive housing waitlist. It should be noted that all service providers utilizing instrument-based approaches to targeting families should provide proper training of staff prior to implementation to ensure reliable application of the tool.
Although the literature clearly demonstrates that data-driven approaches are far more effective than case-based “clinical” judgments for accurately predicting outcomes and prioritizing high-need families, these methods may not always be feasible, especially in smaller jurisdictions or under-resourced programs. In such cases, it is also possible to identify families meeting the target criteria through a case conference. Service providers already working together and engaging with common families may be able to sit down together and identify families based on their collective knowledge of the families’ needs. Information can be verified through public agency partners if necessary.