2006 HUD New England Regional HMIS Conference

1
2006 HUD New England Regional HMIS Conference
 
April 10, 2006 Campus Center University of
Massachusetts Boston

• Sponsored by the U.S. Department of
• Housing and Urban Development

2
Presented by in partnership by
 
The Center for Social Policy atThe University of
Massachusetts Boston,Abt Associates, Inc.,
New England Regional HMIS
3
Understanding the Extent and Nature of
Homelessness Using HMIS Data
 
Tatjana Meschede, Ph.D., Center for Social
Policy, UMass Boston Eric Hirsch, Ph.D.,
Professor of Sociology, Providence College, Rhode
Island

• Sponsored by the U.S. Department of
• Housing and Urban Development

4
Todays Presentation

• What questions can be answered based on HUD
  required HMIS data elements?
• Examples of data analysis processes
• Limitations of HMIS data/caveats
• Describing Homelessness in Rhode Island using
  HMIS data
• Check in with participants

5
HUD Required HMIS Data Elements

• Universal Data Elements
• Program entry/exit
• Gender
• Race/Ethnicity
• Date of birth to calculate age
• Disability status
• Program Level Data Elements
• Employment/Income/Non-Cash Benefits
• Education
• Health/Physical and/or Developmental
  Disability/HIV/AIDS

• Veteran status
• Residence prior to
• program entry
• ZIP of last permanent
• address

• Mental Health/
• Substance Abuse/
• Domestic Violence
• Services Received
• Reason for Leaving and
• Destination

6
Types of Analyses - Definitions

• Descriptive/Exploratory/Explanatory
• Descriptive Presenting characteristics of people
  accessing homeless services in your CoC
• Exploratory Analyze difference of homeless
  sub-populations, for example long-term vs.
  short-term users of services
• Explanatory Predict relationships between
  variables/testing more complex analytic models
• Program Evaluation
• Cross-sectional/Longitudinal
• Cross-sectional Point in time
• Longitudinal Over time

7
Types of Analyses cont.

• Comparison to Other Data Sets
• National Homeless Data (HUD, 1996 AHAR)
• Census Data
• Geographic Information Systems (GIS) Use of
  geographical data for analyses by expert analysts
• Example
• residence before becoming homeless, where served
  and where moved to when exiting homelessness

8
Examples of Research Questions at Different
Levels

• National Annual Homeless Assessment Report
  (AHAR)
• How many people use homeless residential
  services?
• Who uses homeless residential services?
• Where do users of homeless residential services
  receive these services and where did they live
  before?
• What are the patterns of homelessness and of
  homeless residential service use?
• What is the current capacity for housing homeless
  people and how much is utilized?
• State/Local Research Questions
• Same as above, additional questions responding to
  local policies
• Agency/Program Research Questions
• Same as above, additional questions responding to
  program characteristics

9
Example 1 Who uses homeless residential services?

• Goal A general description of people using
  homeless services and exploration of differences
  among sub-groups
• Examples of Subquestions
• What percentage of homeless shelter users are
  disabled? (descriptive/cross-sectional analyses)
• Does the profile of homeless women differ from
  homeless men? (exploratory/cross-sectional
  analyses)
• Is elder homelessness on the rise?
  (descriptive/longitudinal analyses)
• How does the profile or homeless people differ
  from the general population in the same city,
  county or state? (comparison of HMIS data to
  other population data sets)

10
Possible Data Elements to Describe Homeless
Service Users

• Universal data elements
• Gender
• Race/ethnicity
• Date of birth to calculate age
• Disability status
• Veteran status
• Personal identifier
• Household identifier
• Program-level data elements
• Education
• Income

11
Example 1 Data Analysis Steps

• Research Question
• Are homeless women
• more likely to have a
• disability than homeless
• men?
• Step 1
• Raw data are stored in
• spreadsheet

12
Example 1 Data Analysis Steps cont.

• Step 2
• Data are extracted
• and summarized

Female Male Disability No 2
2 Disability Yes 1
1 Null 2 Unknown
2 Total 3 7
13
Example 1 Data Analysis Steps cont.
Table 1 Disability Status of Homeless Women and
Men

• Step 3
• Data are presented
• in a table
• Decision step
• Include/exclude
• missing/unknown
• values

Missing and unknown values are excluded
Missing and unknown values are included
14
Example 1 Data Analysis Steps cont.
Figure 1 Percentage of Homeless Women and Men
with a self-reported disability

• Step 4
• Data are presented
• in a chart

N3
N3
15
Example 2 What are the patterns in using
services?

• Goal To describe/explore service use patterns of
  people experiencing homelessness
• Examples of Subquestions
• What CoC services to people use, and for how
  long?
• Are people receiving services near where they
  lived before becoming homeless?
• What is the extent of seasonal variation in
  service use?
• What proportion of people are long-term service
  users, what proportion of people cycle in and out
  of the homeless service system?

16
Possible Data Elements to Explore Service Use
Patterns

• Universal data elements
• Personal Identifier/Household Identifier
• Program ID (including type of program and
  location of services)
• Prior residence (including time at prior
  residence)
• Last permanent zip code
• Program entry/exit dates
• Program-level data elements
• Services Received
• Reason for Leaving and Destination

17
Example 2 Data Analysis Steps

• Research Question
• Is there a difference
• of time in homeless
• services between
• men and women?
• Step 1
• Raw data are
• stored in
• spreadsheet

18
Example 2 Data Analysis Steps

• Step 2
• Data are
• summarized

19
Example 2 Data Analysis Steps

• Step 3
• Data are
• presented

Time Served Female Male lt1
month 1 (33) 1m lt 3m 1 (50) 3m - lt 6
m 1 (33) 6 m lt 12m 1 (50) 1
year 1 (33) Total 2
3
20
Example 2 Data Analysis Steps
Figure 2 Length of Time Homeless for Women and
Men

• Step 4
• Data are presented
• in a chart

N2
N3
21
Evaluation of Homeless Program Effectiveness

• HMIS data can be used to measure performance by
  single provider, town or city, CoC, or larger
  region
• Examples
• Reduced of stays in shelter
• of discharges to permanent housing
• Access to mainstream benefits
• Access to employment income

22
Evaluation of Homeless Program Effectiveness cont.

• With linkage to other mainstream service
  databases, other outcomes can be assessed
• Examples
• of ER admissions
• of hospitalizations
• of incarcerations

23
Caveats

• Data Quality - No Good Answer Without Quality
  Data!
• Data collection how to ask the questions
• Data entry careful attention to details
• Data checking double checking data for entry
  errors
• Validating data with other data sources
• Consistency among programs contributing data

24
Caveats cont.

• Sample size/Coverage
• Who does your data represent?
• All homeless assistance programs, residential or
  not, in your CoC?
• All agencies with at least one homeless
  assistance program in your CoC?
• All homeless people within your CoC
• Who is left out?
• People who dont use shelters could be missing
• People who use particular types of shelters
  (e.g.. DV 12-hour missions) could be missing
• People who dont want to provide information for
  HMIS data base

25
Linking HMIS Data with Other Data

• What else you might want to find out and how
• Other characteristics beyond standard data
  elements Add fields
• Long-term housing stability outcomes Link to
  other databases or data sharing
• Other data sources
• Surveys, e.g. on program specific issues
• Qualitative studies (focus groups, interviews)
• Other databases (e.g. public benefits, Medicaid,
  etc)

26
Some Outstanding Questions

• How can I compile the burning questions in my
  CoC?
• Bring together different groups of service
  providers at all levels and consumers to
  brainstorm questions in your community
• Who should/can conduct HMIS data analyses?
• Basic analyses can be done by staff using
  spreadsheets or HMIS tools
• Experienced researcher for more complex analyses

27
Limitations of HMIS data

• Hopefully information gathered and analyzed in
  HMIS will help to reduce the extent of
  homelessness/end homelessness
• However, homelessness is the outcome of failure
  of many service systems in conjunction with
  market failures (housing and labor)

28
What can HMIS data contribute to ending
homelessness?

• Help run the services more efficiently
• Document Service Gaps and Apply for Funding
• Example Documented increase in elder
  homelessness helped secure funding for a staff
  position in Boston addressing the needs of elder
  homeless people
• Inform Policies
• Example Why were homeless families at Scattered
  Sites staying longer than families in Congregate
  Living?
• Families at Scattered Sites were more likely to
  have two parents
• On average, families at Scattered Sites were
  larger and had older children

29
Conclusion and Implications

• HMIS data have great potential to inform policy
  and planning at program, CoC, or statewide levels
• Requires full participation of programs, and
  consistent data collection across programs
• It takes time to gather sufficient data for
  analyses and comparisons across different groups

30
For More Information

• Tatjana Meschede
• Senior Research Associate
• Center for Social Policy, UMass Boston
• Tatjana.meschede_at_umb.edu
• [phone]
• http//www.mccormack.umb.edu/csp/

31
Homelessness in Rhode Island- 1989 to 2005
32
Rhode Island Shelter Statistics- 2004-2005

• Unduplicated Clients 6,408
• Total Bed Nights 217,871
• Average Daily 217,871/365 597
• Census
• Average Length 217,871/6,408 34 nights
• of Stay
• Turnover 6,408/597 11

33
Shelter Use Rates per 1,000

• Rhode Island Resident 6.1
• Child Under 5 10.1
• Non-Hispanic Whites 3.7
• Blacks 21.6
• Hispanics 12.1
• Native-Americans 14.7
• Asian-Americans 1.7

34
Types of Homelessness
35
Characteristics by Homeless Type

• From Mental Health
• Shelter/Street Problems
• First Time 28 27
• Once or Twice 31 33
• Chronic 44 49
• Long-Term 53 36

36
Family Status and Gender Differences

• Hispanic Incarceration Alcohol
• Single Men 16 22 21
• Single Women 14 12 21
• Family Men 34 5 3
• Family Women 37 2 2

37
Rise in Income Inequality in Rhode Island

• In 1997 s
• Richest 1/5 of Families
• Late 70s 92,130
• Late 90s 151,190 up 64
• Poorest 1/5 of Families
• Late 70s 11,900
• Late 90s 9,910 down 17

38
Affordable Housing Crisis in Rhode Island- 2004

• Average homeowner costs 1,469/month
• Income needed 58,476 at 30 of income
• Average rents 1,121/month
• Income needed 44,840 at 30 of income
• Actual median rental household income 31,489

39
Current Government Housing Programs are
Inadequate

• Federal housing support in 2004 dollars
• 83 billion in 1978 29 billion in 2003
• Low and moderate income housing units
• 34,600
• Households needing subsidy- making lt
• 45,000/year 192,325

40
How to End Homelessness in Rhode Island

• Housing Production
• Rhode Islands Neighborhood Opportunities
  Program
• 21.5 million invested
• 545 units created
• 201 units of subsidized family housing
• 141 units of permanent supportive housing
• 183 million leveraged
• 5,400 local jobs created

41
National Housing Trust Fund

• Will Produce 1.5 million Low-Income Housing Units
  in 10 Years
• 75 for Households lt30 of Area Median
• Rhode Islands Share 5,000 Units, 60,000 jobs
  created, 1.6 billion in wages
• Units to be Affordable for the Useful Life of the
  Property
• Uses Excess FHA and Ginnie Mae Revenues- No
  Increase in Federal Deficit

42
For More Information

• Eric Hirsch
• Professor of Sociology
• Providence College, Providence, RI
• ehirsch_at_providence.edu
• 401-865-2510