Conceptualizing and Creating a Homeless Families Typology

1
Conceptualizing and Creating a Homeless Families
Typology

• Debra J. Rog, Ph.D.
• Westat
• Presentation at the National Conference on Ending
  Family Homelessness, National Alliance to End
  Homelessness
• February 8, 2007

2
Presentation Overview

• What is a typology and why do we need one?
• Description of federally funded project to
  conceptualize typologies of homeless families
• Prevention typology
• Resource allocation
• Description of two efforts to inform typology
  development
• Reanalysis of fragile families dataset
• Study in process of shelter exits in
  Massachusetts
• Final thoughts

3
Definition of a Typology

• What is a typology?
• A classification system that differentiates a
  population into distinct subgroups or subtypes.
• It can be used to
• Describe a population
• Match groups to services and
• Predict service use and response.
• Why create a homeless families typology?
• To effectively target existing services and
• To identify new efforts to both prevent
  homelessness and its reoccurrence and intervene
  with currently homeless families.

4
A Desirable Typology

• Classifies population into subgroups that are
  homogeneous and non-overlapping
• Incorporates both environmental and individual
  factors
• Covers total population
• Is simple to use
• Has practical utility for service providers and
  policy makers

5
ASPE Funded Typology ProjectKey Activities

• Literature review
• Review of existing data and ongoing panel studies
• Identified 15 potential datasets for secondary
  analysis
• Re-analyzed data from Fragile Families Project on
  subgroups of poor families (homeless, doubled-up,
  at-risk)
• Commissioned expert papers
• Expert Panel meeting
• Options for potential research activities
• Final report and debriefing

6
Key Findings from Project

• Need for two homeless families typologies
• Prevention
• Resource allocation
• Staged approach to developing typologies needed
• Initial development guided by existing data
• Elaboration through short-term research options
• Strongest, most lasting development, through
  ongoing national surveys and longitudinal studies

7
Prevention TypologyPurpose

• To rank families according to levels of risk of
  homelessness and probability of a quick exit
• To distinguish families in desperate need from
  those with more moderate needs

8
Prevention TypologyFramework

• Initial 4 Cell Model
• Begin developing based on existing literature and
  enhance with data from one or more study options

Environment
Facilitators
Barriers


Major
Family Needs
Minor
9
Prevention Typology What We Know To Begin

• Key risk factors for homelessness include
• Resources (economic and social)
• Life stage (age having young children)
• Ethnicity
• Mental health and substance use
• Best to target population at risk as families
  request shelter
• Broader targeting, even among poor families, is
  likely to be inefficient and inaccurate

10
Analysis to Inform Prevention TypologyReanalysis
of Fragile Families Dataset

• Fragile Families and Child Well-Being Study
• Five year longitudinal study of new parents and
  children sampled from hospitals
• National sample of marital and non-marital births
• (4,898 families at baseline)
• Two waves of data currently available (1 and 3
  year follow-ups)
• Third wave (year 5) due in 2007
• Offered multi-site sample at high risk of
  homelessness and residential instability
• Opportunity to
• examine incidence of homelessness
• compare homeless to other poor families in range
  of residential arrangements

11
Reanalysis of Fragile Families Data SetSample
Selection

• Re-analysis restricted to sample of very poor
  families
• Total sample of 838 families meeting following
  criteria
• Mother 18 years of age or older
• Household income 50 below poverty level at year
  1

12
Reanalysis of Fragile Families Data
SetDescriptive Analysis

• Constructed 4 residential groups

Yr1 Yr 3 Combined
Residentially Stable 35 42 22 (both times)
At Risk 39 37 41 (either time)
Doubled-up 21 16 28 (either time)
Ever Homeless 6 5 8 (either time)

• Risk of being homeless is low even among
  extremely poor women

13
Reanalysis of Fragile Families Data
SetInferential Analyses

• Analytic Approach
• Logistic regressions performed to determine risk
  and protective factors of experiencing
  homelessness and remaining stable
• 3 models conducted to predict
• Year 1 status
• Year 3 status
• Combined status
• Caveats
• Resulting models have relatively poor fit
• Samples are small
• Homelessness is quite varied
• Models lack contextual variables

14
Reanalysis of Fragile Families DatasetPredictors
of Homelessness
YR 1 Model YR 3 Model YR1 or YR3 Model
Demographics
Pregnant YR1
Family Background
Change Partner (YR 1-3)
Live with Mother (YR1)
of Adults in Household (YR3)
Social Support
1,000 loan (YR3)
Social Capital
Mom working (B)
Change Work (YR 1-3)
Income (YR1)
Health, SA, MH
Ever use SA
SA ever interfere
Ever DV
MH Prob
15
Reanalysis of Fragile Families Data
SetPredictors of Homelessness(continued)
YR 1 Model YR 3 Model YR1 or YR3 Model
Neighborhood Housing
Neighborhood safety (1-4, unsafe)
Housing Assistance
TANF
Increases probability of being homeless
decreases probability o f being homeless
plt.05, plt.01, plt.001
16
Reanalysis of Fragile Families Data
SetPredictors of Residential Stability
YR 1 Model YR 3 Model YR1 or YR3 Model
Family Background
Partner
Change Partner
Change live with Mom
adults in household
children (B)
Social support
1000 Loan (YR 1)
Social Capital
Change Mom work (1-3)
Change Partner work (B-1)
Health, SA, MH
Health stability (1excellent 5poor)
Ever use SA
Ever DV
MH Problem
17
Reanalysis of Fragile Families Data
SetPredictors of Residential Stability(continue
d)
YR 1 Model YR 3 Model YR1 or YR3 Model
Neighborhood Housing
Public Housing
Food stamps




18
Reanalysis of Fragile Families Data
SetTentative Findings

• Families experiencing homelessness
• Have lower household incomes
• Are less likely to receive housing assistance but
  more likely to receive TANF
• Are more likely to have experienced domestic
  violence and mental health problems
• Families remaining residentially stable
• Are more likely to live with a partner and have a
  greater of adults living in the household
• Are more likely to have a partner working
• Are less likely to have SA, DV, and MH issues
• Are more likely to have lived in public housing

19
Resource Allocation TypologyPurpose

• To classify families by the factors that
• Block their ability to exit homelessness (e.g.,
  poor credit past justice involvement)
• Challenge their ability to achieve stability and
  self-sufficiency

20
Resource Allocation TypologyFramework

• Create based on 3 types of variables
• Exogenous (housing environment, housing, and
  health and human service access)
• Endogenous (family and individual
  characteristics, including family support needs,
  broad health needs, social needs, childrens
  needs)
• Situational (fit between the families needs and
  accessible resources)
• Use staged approach to building framework

21
Resource Allocation Typology What We Know To
Begin

• Housing subsidies are a key predictor, but there
  are not enough available to meet needs
• In addition
• Not all families may need full subsidy others
  may need more than a subsidy
• Even with subsidies, some families return to
  homelessness
• Therefore, families range considerably in what
  they need to exit homelessness and remain stable

22
Study to Inform Resource Allocation
TypologyMassachusetts Exit Study

• Purposes
• To address knowledge gaps re the shelter exit
  process
• To provide prospective epidemiological study of
  the exit process
• Study Design
• Longitudinal study (12 months) of shelter work in
  Worcester
• Three components
• 1. Analysis of administrative data
• 2. Interviews with 3 samples of families,
  including those who
• Exit shelter within 6 months
• Stay in shelter longer than 6 months
• are eligible for shelter services but are
  diverted
• 3. Interviews with system-level informants
• 18 month study ending in 12-07

23
Massachusetts Exit StudyComponent 2 Exit
Process

• Purpose
• To understand the factors that facilitate and
  block exit
• To understand the residential arrangements after
  shelter
• Design/Sample Selection
• Early exit families (estimate 100 families)
  interview upon exit all families exit in 2006
• Stuck families (estimate 85 families)
• Interview at 6 months and at exit if not exit, 3
  month follow-up
• Diverted families (estimated 50 families)
• Interview at point diverted 3 month following

24
Massachusetts Exit StudyComponent 2 Exit
Process

• Data Collection
• Demographics and background
• Family composition
• Prior homelessness/housing
• Employment, dept, income
• Legal issues
• Services received and shelter experience
• Trauma
• Physical and mental health
• Substance use
• Exit process, problem
• Childrens question
• Resource knowledge and use

25
Massachusetts Exit StudyLikely Implications for
Informing Resource Allocation Typology

• Prospective information on the exit process
• Key predictors of exit with attention to the role
  of
• Services and resources
• Recurring trauma exposure and conflict
• Credit, legal, criminal justice, and other issues
• Mental health, substance abuse, disabilities
• Provide a beginning foundation on the
• Nature of the population
• Dynamics of the shelter system
• Services available, known, and used

26
Conclusion

• Need for two typologies of homeless families
• Voiced by the expert panel
• Illustrated in work on the ground to pilot
  various triage efforts
• Apparent in the range of risk evident in the
  Fragile Families re-analysis
• Staged approach to building typologies
• Provide data to guide current efforts
• Build long-term capacity to refine and direct
  future prevention and resource allocation efforts