Developing New Categories for Homelessness
New typologies to guide early, needs-based interventions
There is a first day of homelessness for everyone who becomes homeless. From this day, their homeless experiences will diverge to where, given enough time, a retrospective look can identify different subgroups based upon the progression of their homelessness, showing common characteristics and experiences associated with these trajectories. We propose to use this retrospective vantage point to update existing typologies of homelessness based on the different life courses of homelessness, and apply what we learn to prospectively guide early, needs-based interventions for those who enter homelessness.
The proposed retrospective analysis will update typologies through identifying clusters of homeless trajectories in which the individuals within each cluster have experiences of homelessness that are as much like each other as possible and as different as possible from the experiences of people in other clusters. After typological clusters are identified, one of the next tasks is to understand the people in each cluster based on attributes such as demographic characteristics, household structure, medical conditions, geography, and public costs. Here we draw on data from over one million Los Angeles County residents who experienced homelessness over a fifteen-year period.
The resulting typologies will be relevant to a prospective, evidence-based intervention framework. The typologies will be applicable to differentiating the levels and types of services needed by different groups within the homeless population to reduce or avoid the economic, social, mental health, and legal wreckage in individuals lives that results from extended durations of homelessness.
This research will update the very influential Kuhn-Culhane typologies developed in 1998. The purpose is to strengthen the connection between homeless classifications and the services needed by individuals in each classification group, as well as to be more adaptable for guiding a wider set of early interventions. Questions that will be answered include:
- What factors most clearly differentiate the outcomes of individuals who experience homelessness?
- How does a longer time window for identifying homeless trajectories and inclusion of carceral and inpatient episodes, and unsheltered homeless stints in addition to shelter stay dynamics alter the Kuhn-Culhane typologies?
- What combinations of attributes and pre-homeless characteristics can be linked to subsequent homeless trajectories to identify pathways to homelessness that regularly precede specific homeless trajectories?
- What are the long-term trajectories of unsheltered homeless individuals?
- Are there early interventions, based upon previous pathways, which can alter subsequent homeless trajectories?
- What subpopulations, based upon this new cluster typology, warrant specific attention? Subpopulations receiving particular focus here will include:
a. Unsheltered individuals.
c. Young adults.
d. Labor force participants.
This project will revisit and build on the typologies of homeless that were developed by Randall Kuhn and Dennis Culhane in 1998, using data about residence in homeless shelters. Those typologies used a temporal framework to group individuals into three categories based on number of shelter days and number of shelter episodes. The three groups are transitionally, episodically and chronically homeless individuals. These typologies have provided a framework for addressing homelessness for two decades. The participation of Dennis Culhane as part of the research team is important for providing analytic and policy continuity between existing homeless typologies and the new typologies that will be developed through this project.
The principal appeal of Kuhn and Culhane’s cluster typology has been the intuitive grouping of homeless individuals (and later families) into three groups based upon number and duration of homeless episodes. The resulting typology were immensely influential on homeless policies, including approaches that intensively target a relatively small group of “chronically” homeless households that consumed a disproportionate share of homeless services. But while this typology provided a framework for addressing the needs of those who had already been homeless, its use of a limited set of criteria to classify individuals becomes disadvantageous when used as a basis for more prospective approaches to address homelessness.
The Economic Roundtable team will identify features that can be used to differentiate the trajectories of individuals who experience homelessness using 15 years of longitudinal linked records for over one million Los Angeles County residents who experienced homelessness. The 84 categories of data in these linked records are listed in the following Data Section. These linked records include twelve-thousand individuals who were identified as unsheltered, making it possible to analyze the multi-year trajectories of these individuals.
Developing Homeless Typologies
Typologies will be developed based on cluster analysis of outcomes of homelessness. We will look at the life course of homeless individuals and create categories for different outcomes. Heavy service users are likely to be one (or more) of the typological groups – possibly an offshoot of the chronic category in the Kuhn-Culhane typologies, but that isn’t the sole purpose of the typologies. It’s also important to understand groups with short experiences of homelessness, as well as other distinct groups.
Jails and hospitals are important care-taking anchors for some individuals, and the primary drivers of public costs. In our predictive modeling for Santa Clara County we’ve seen that these two institutional settings are distinct pathways for different groups of individuals. We will investigate the significance of time spent in these two institutional settings for differentiating homeless outcomes.
Many homeless individuals have employment histories and many are actively seeking employment so as to be able to house themselves through earned income. Employment history will be investigated as a factor that differentiates homeless outcomes.
As with the Kuhn-Culhane typologies, we will continue to use indicators, including temporal indices, which are externally observable and primary features of homelessness. Because more extensive data is now available, this array of externally observable, primary features will be expanded to include more indices and to encompass a longer time window within the life course of homelessness. The new typologies will utilize externally observable features of time, dwellings and institutions traversed by homeless individuals, including:
- Total duration of homelessness
- Sequence and duration of homeless episodes
- Occupancy pattern of homeless dwellings – unsheltered vs. sheltered
- Amount of institutional care by the jail system
- Amount of institutional care by the health care system
- Amount of labor force participation
Each cluster will be described based on the factors that are relevant for delineating it. Clusters will also be described based on their size relative to other clusters.
Describing the Populations within Typologies
After identifying typologies for the different life courses of homelessness found within the homeless universe, we will describe the personal attributes of people who populate each typology. These descriptions will include the following attributes:
- Age at onset of homelessness
- Age at exit from homelessness
- Accompanied by children
- Accompanied by adult partner
- Medical diagnoses
- Substance abuse
- Public assistance history
- Foster care history
- Geographic subregion of Los Angeles County
- Public costs
Applying Outcome-Based Typologies to the Homeless Population
A follow-on task will be to shift perspectives and describe people in different homeless population groups based upon information available at the point of their initial entry into homelessness, e.g., public assistance receipt, medical diagnoses, or demographic characteristics, based on their distribution among the typological categories.
A better map of homelessness will be valuable for both endgame strategies for housing chronically homeless individuals based on acuity of need, Critical Time Intervention strategies for vulnerable individuals, as well as pre-emptive strategies for identifying new entrants who are likely to stay homeless and targeting them for early intervention.
Outcome-based typologies are valuable for differentiating long-term levels of need, and for identifying and targeting people who already have “qualified” as chronically homeless. A basic challenge is to expand the explanatory power of the typologies beyond a retrospective understanding of homeless outcomes and apply them to prospectively identifying likely candidates for different clusters.
A primary goal for this research is to develop typologies that, while based on outcomes from the life course of homelessness, can be linked to people who are newly homeless or in the middle of their homeless journey. Developing evidence-based typologies that represent actual outcomes will create a foundation for identifying linkages between long-term outcomes and attributes that foreshadow those outcomes.
A large number of people enter homelessness over the course of a year and a small number go on to become chronically homeless. The chronically homeless population that we see has accrued over a number of years. One of the payoffs from the updated homeless typologies will be a clearer understanding of the different trajectories and outcomes for people who experience homelessness.
The proposed analysis of longitudinal data in this project will, among other things, produce a much more detailed and reliable profile of the size and attributes of the homeless population within a longer time window. A better understanding of homeless population dynamics will support better targeting for the crucial task of housing disabled individuals who are persistently homeless as well as targeting strategies for intervening early to assist vulnerable individuals, as well as to support early exits for the subgroup of newly homeless who go on to become persistently homeless.
Ensuring that Typologies are Relevant to Interventions
The typologies that are developed will be relevant to an evidence-based intervention framework. We envision that the framework will use a “progressive engagement” approach to homelessness prevention and intervention. Recognizing that permanent supportive housing is a very scarce and expensive resource, the progressive engagement approach entails treating new entrants into homelessness with a light touch that includes crisis intervention.
For individuals with serious vulnerabilities such as mental illness or long incarceration histories it entails providing Critical Time Intervention support to help them reestablish themselves in stable housing with access to needed supports through a variety of mainstream mechanisms, if possible, or alternatively by providing shallow subsidies.
We assume that individuals will not be queued up for homeless-administered permanent supportive housing until other avenues fail and their homelessness persists, except for those with an already demonstrated pattern of chronic homelessness who are re-entering homelessness.
This framework of progressive engagement includes targeted interventions for new entrants into homelessness who are identified as having a high risk of becoming persistently homeless. An example is providing time-limited wage subsidies, housing, child care, and transportation to help recently unemployed workers who are at risk of persistent homelessness get a foothold in a new job.
An objective is for the typologies to be linkable to different levels of services needed to help individuals exit homelessness. The typologies will also support future research to identify factors that are discernable at early stages of homelessness and indicate a likelihood that individuals will become persistently homeless. This will open additional opportunities for developing screening tools to identify those individuals and provide targeted early interventions to help them become stably housed rather than leaving them on a path to long-term homelessness.
The research team for this project includes:
- Patrick Burns, M.A., Senior Researcher
- Daniel Flaming, Ph.D. President
- Halil Toros, Ph.D., Statistical Analytics Consultant
University of Pennsylvania
- Dennis Culhane, Ph.D., Andrew Stone Professor of Social Policy at the School of Social Policy and Practice
University of Delaware
- Stephen Metraux, Ph.D., Associate Professor and Director of the Center for Community Research and Service at the School of Public Policy and Administration