Skip to main content

Using Data to Understand and End Homelessness

HUD.GOV HUDUser.gov
Periodicals
Evidence Matters Banner Image

Summer 2012   

    HIGHLIGHTS IN THIS ISSUE:

        Tackling Veteran Homelessness With HUDStat
        Using Data to Understand and End Homelessness
        Linking Housing and Health Care Works for Chronically Homeless Persons


Using Data to Understand and End Homelessness

Highlights

      • Measuring the extent of homelessness is essential to combating it, and efforts to count the homeless population have evolved significantly since the early 1980s.
      • A combination of Homeless Management Information Systems, Point-in-Time counts, and Housing Inventory Counts inform policymakers and advocates on demographics, trends, and the availability and usage of services among America’s homeless population.
      • Improved accuracy and detail of homeless data have influenced all aspects of HUD’s policies as well as those of its partner agencies.

Since homelessness emerged as an issue in the United States, a broad group of dedicated individuals and organizations — from advocacy groups and shelters to local, state, and federal government agencies — have fought to help homeless persons find housing and remain stably housed. Data have become a critical component of these efforts. Documenting the number, characteristics, and needs of homeless persons in American communities, as well as the number of people receiving services and the capacity of these services, is essential to identifying the proper strategies to tackle the problem; it’s very difficult to manage what you can’t measure.

Because counting the homeless population is difficult and resource intensive, local governments have had to develop systems that are flexible enough to accommodate differing local circumstances yet also consistent enough to aggregate local data and get a holistic picture at the national level. This article summarizes the evolution in understanding homelessness in this country through data, detailing early attempts at measurement and the current systems used by HUD and its federal and local partners, principally Point-in-Time (PIT) counts, the Housing Inventory Count (HIC), and Homeless Management Information Systems (HMIS). The article also explores the ways that this information has helped policymakers confront homelessness more effectively. When used together, these complementary data collection efforts offer a more in-depth picture of homelessness that enables policymakers to target resources toward effective assistance models and more quickly adapt less effective programs.

Early Efforts To Understand America’s Homeless Problem

As homelessness increased in the 1980s, interest grew in understanding the nature and scope of the problem. Advocates, particularly at the Community for Creative Non-Violence, asserted that the national homeless population totaled two to three million homeless persons. In the absence of other data, these numbers became conventional wisdom.1 To gather more accurate data on homelessness in the United States, federal agencies began to conduct national point-in-time (PIT) studies. These PIT studies were based on the number of homeless persons counted during a specific time period and in specific places and were conducted to enumerate the homeless population.

HUD conducted the first national PIT study from 1983 to 1984.2 The study was limited to a sample of shelters in 60 areas and used statistical methods to derive counts of persons with shelter and those without shelter. Building on HUD’s 1983 sample study, the U.S. Department of Agriculture (USDA) funded a large study in 1987 to derive a national count and learn more about the characteristics of the homeless population. This study involved hundreds of providers in 20 cities and yielded the first nationally representative dataset including demographic information, such as household composition, race, age, and income sources. In 1988, HUD conducted its first shelter inventory to assess the capacity of the shelter system.3

Following the lead of these national efforts, a number of local communities began systematically collecting data on homeless persons as early as 1986. New York City and Philadelphia were pioneers in collecting citywide data. They were among the first cities to have local government-funded homeless shelters that required grant applicants to collect client-level data. Other early municipal or statewide systems included Columbus, Ohio; Phoenix, Arizona; St. Louis, Missouri; and the state of Rhode Island.4

Private researchers drew significant insights from the data that had implications for decisionmakers at all levels. For instance, Dr. Dennis Culhane analyzed New York’s data and found that it cost an average of $40,500 for a single person to live on the streets of New York City during the course of a year.5 This finding helped elected officials, policymakers, program administrators, advocates, and researchers recognize that homelessness was an economic issue as well as a moral one.

The next significant effort to enumerate homelessness at the national level was initiated by the U.S. Census Bureau as part of the 1990 census. The effort, referred to as “S-Night” (the “S” stood for both street and shelter6), did not result in an estimate of the homeless population but introduced the notion of enumerating in every community rather than relying on sampling.7 In that same year, the first longitudinal analysis, tracking changes in homelessness over time, was performed based on a telephone survey which asked respondents if they had ever experienced homelessness and, if so, whether it had been in the past five years.8 The U.S. Interagency Council on Homelessness and its agency members (HUD, the U.S. Department of Health and Human Services, and the U.S. Department of Veterans Affairs) conducted another sample-based PIT count in 1996, similar to the 1987 USDA study, which was used to inform policymakers, especially with regard to the geographic distribution of homelessness.

These early studies helped the homeless assistance community make critical strides toward understanding and addressing homelessness. With each new study a national picture began to form. The first study established a baseline number of homeless persons, and subsequent studies have helped bring homeless issues into greater focus, from the demographics of the homeless population to its geographic distribution. However, a few national studies with varying methodologies and purposes spanning a 25-year period were grossly inadequate to understand homelessness and the tools that could best solve it. To more effectively confront homelessness, stakeholders at both the local and national levels needed to have much more reliable data based on regular and consistent local data collection efforts.9

Developing a Vehicle for Regular National Reporting

The new millennium brought a monumental change in HUD’s role in data collection. In 1999, Congress directed HUD to develop a representative sample of jurisdictions to collect unduplicated counts of clients served, demographic information, types of housing received, and outcomes of homelessness projects, such as housing stability.10 In 2001, Congress charged HUD to work with communities to develop unduplicated counts of homeless persons assisted, analyze the patterns of service use by homeless clients, and evaluate the effectiveness of programs locally. To accomplish this expansive mandate, communities needed to collect consistent, longitudinal data through what had become known as Homeless Management Information Systems (HMIS).

From left to right: Carrie Schmidt, Field Office Director in Richmond, Ronnie Legette, CPD Director for Richmond, and Mark Johnston participated in the 2012 PIT count in Richmond, Virginia.
From left to right: Carrie Schmidt, Field Office Director in Richmond, Ronnie Legette, CPD Director for Richmond, and Mark Johnston participated in the 2012 PIT count in Richmond, Virginia.

HMIS is a locally administered electronic system that collects and stores client-level data for those receiving homeless assistance. HUD deployed professionals with HMIS experience to communities to provide extensive technical assistance, including one-on-one assistance and an HMIS implementation guide. HUD also sought input from the early implementing communities, other HMIS professionals, homeless researchers, advocates, providers, and privacy experts to get helpful advice to communities that were slow to implement HMIS. In the course of this undertaking, HUD decided not to develop a software application that all providers would be required to use, instead relying on the marketplace to develop software that would adhere to HMIS standards.

With the congressional mandate to collect and report on the homeless population, HUD created tools and incentives for communities to collect data. Through a coordinated effort between HUD and homeless assistance stakeholders with HMIS expertise, HUD developed HMIS technical, privacy, and security standards as well as a format for an Annual Homeless Assessment Report (AHAR) to be submitted to Congress. HUD also established national standards for the count of homeless persons (through a regular PIT count) and for an annual inventory of homeless beds and units. HUD continues to provide extensive technical assistance on implementing and operating HMIS at national, regional, statewide, and local conferences. Once the standards were issued and the mechanism for technical assistance was in place, HUD began to expect that all recipients of HUD homeless funds participate in HMIS.11

HUD also changed its homeless assistance grants competition to reflect an emphasis on quality data. The Homeless Assistance Grant competition represents the largest single federal resource to combat homelessness. As a part of their annual application for Homeless Assistance Grant funding, communities must conduct a PIT count in their area and report that data in their applications. In addition, communities must report the date they conducted the count, the nature of the count (i.e., sheltered and/or unsheltered), and the methodology for the count. Communities are also required to report annually on their HUD-funded and non-HUD-funded housing inventory targeted for the homeless, referred to as the Housing Inventory Count (HIC).

In addition to reporting data on homeless populations and the housing inventory, HUD added questions to its funding applications regarding the quality of HMIS that communities were using. To help cover the costs associated with implementing and operating HMIS, HUD successfully sought from Congress the ability to allow grantees to use Homeless Assistance Grant funds for this purpose, which has been another key factor in implementing HMIS nationally. As a result of these various efforts, PIT and HMIS participation have increased dramatically.

HUD’s Current Data Collection Efforts

HUD continues to rely on data to learn about and address the homeless crisis in America. The core data sets that HUD uses for its current evaluation are PIT, HIC, and HMIS. Each data set has its own strengths and limitations, and HUD leverages each of these data sets to form a more complete picture of homelessness in the United States.

PIT. HUD requires communities to submit a count of the homeless population in their area as well as information on specific subpopulations, including chronically homeless persons, veterans, and unaccompanied youth. Communities report this information by household type (i.e., households with at least one adult and one child, households without children, and households with only children) and program type (i.e., Emergency Shelter, Transitional Housing, and Permanent Housing). A PIT count is composed of two parts: a sheltered PIT count, which is required every year, and an unsheltered PIT count, which is required at least every other year. Communities submit these data annually through their Continuum of Care (CoC) applications for Homeless Assistance Grants.

Many communities develop their sheltered count from their HMIS data. However, when the HMIS data are insufficient, due to lack of coverage across the community of providers or other reasons, communities generally supplement the data based on surveys. The surveys vary in complexity from mere observations of the surveyor to in-depth, interview-based surveys. HUD does not prescribe the survey method to use but does provide guidance on survey techniques in its publication, A Guide to Counting Sheltered Homeless People.

The unsheltered count is more complicated and costly to conduct than the sheltered count, and HUD is more strict about the acceptable methodologies for performing these counts. Because unsheltered persons are not generally recorded in HMIS, communities have much more planning to do. HUD’s A Guide to Counting Unsheltered Homeless People outlines the three basic approaches that HUD accepts for conducting an unsheltered count. First, many communities conduct street counts, in which community volunteers visit the streets and locations where they expect to find homeless individuals and count them based on observation over a very specific period (usually between dusk and dawn on a single night). This method is relatively easy to organize, train volunteers to conduct, and aggregate. Although simple, this method of counting invariably misses some people, and little information is gained beyond the total number of unsheltered persons.

The second approach combines the street count with an interview. With this approach, count participants are trained to either interview every single person they encounter who appears to be unsheltered, or interview every nth person to create a simple random sample. The sample-with-interview approach yields a much richer level of data to the community but tends to be more complicated to staff, conduct, and unduplicate. The third method for counting the homeless population is a service-based count in which the community counts people as they receive homeless services during the specific count period. Communities using the service-based approach will often plan a specific event that is likely to attract homeless persons such as a special breakfast or health-care option. Although this method requires the community to carefully determine who has already been counted, it tends to reach a particular homeless population that chooses to use the supportive services available, including soup kitchens, drop-in centers, and street outreach teams, but would be difficult to count through other methods because of where they choose to sleep.

To determine the most appropriate methodology to use, communities need to evaluate, among other things, their climate, size, and availability of resources. The number of participants in the count and the size of the area often drive the method that is chosen. However, several communities use a combination of these methodologies.

In addition to homeless population data, HUD requires communities to submit subpopulation data on chronically homeless individuals and families, veterans, severely mentally ill individuals, chronic substance abusers, persons with HIV/AIDS, victims of domestic violence, and unaccompanied children (under 18). When the subpopulation data are incomplete, communities use sampling and extrapolation methods to derive their counts.

Barbara Poppe, Director of USICH, and HUD Secretary Shaun Donovan participated in a 2011 PIT count in Washington, DC.
Barbara Poppe, Director of USICH, and HUD Secretary Shaun Donovan participated in a 2011 PIT count in Washington, DC.

HIC. HUD requires communities to collect HIC data, which is an annual inventory of the beds, units, and programs designated to serve the area’s homeless population. These data are also submitted annually, in conjunction with the PIT population and subpopulation data. HUD requests that the data be reported based on household types served in the inventory (i.e., households with at least one adult and one child, households without children, and households with only children). The HIC data are often pulled directly from the community’s HMIS. When the HMIS data are incomplete, communities contact the missing providers to determine the nature of their homeless assistance inventory.

HMIS. An HMIS is an electronic data collection system that stores longitudinal client-level information about those who access the homeless services system through a CoC program.12 Because HUD does not create or own HMIS software, HUD does not directly receive client-level information. To ensure consistency and data quality, HUD publishes its HMIS Data Standards as well as other notices and guidance. Communities use HMIS to track homeless individuals as they access services in the community, and they are able to develop a rich data set on homeless persons, from their demographic data to the services they receive to where they go after exiting a program.

Communities aggregate their HMIS data and submit it to HUD through various mechanisms, including their Homeless Assistance Grant applications and their Annual Performance Reports for their HUD-funded projects. HUD also receives HMIS data through its AHAR process, in which it collects unduplicated annual HMIS data at the community level to evaluate its coverage and completeness. HUD uses aggregated HMIS data from communities that have sufficient coverage and completeness to determine national estimates on the nation’s sheltered homeless population.

Each of these three major data sources plays a unique role in informing HUD and the public about the nation’s homelessness. PIT data provide a snapshot in time of the homeless population. Although PIT data are limited to household population, program types, and subpopulation data, they are the only means HUD has of determining the unsheltered population, and they allow communities to report data on providers that are not participating in HMIS. HIC data are HUD’s primary means of gauging the nature and extent of resources that are dedicated to homeless persons across the country, whether funded by HUD or not. HMIS data allow a more holistic understanding of the homeless clients served by participating providers and offer an understanding of data on an annual rather than a point-in-time basis.

HUD’s Data Produces an In-Depth Picture of Homelessness

Data collection efforts have advanced considerably in the past few decades and have opened up new opportunities and insight into homelessness in America. Having regular, accurate data locally and nationally is key to solving homelessness. Initial studies largely provided basic information about the homeless population and demographic composition. For instance, the 1987 USDA survey found that only 10 percent of homeless adults were in households with children, and 84 percent of these households were female headed. The 90 percent of households that had no children were overwhelmingly headed by single men.13 Data collection methods have evolved beyond mere enumeration to allow a more robust understanding of the nature of homelessness and effective interventions.

At a local level, elected officials, government agencies, nonprofit service providers, advocates, and the public can use the data reported in PIT counts, HIC, and HMIS to more effectively engage in solving homelessness if they understand the scope of the problem. Communities are using the PIT count to determine the extent of homelessness in their area and then comparing that with HIC data to determine the resources available. These communities then use HMIS data to determine whether the resources they have are effectively meeting the needs of their homeless populations. Communities are reviewing HMIS data measurements, such as length of stay, to determine the best-performing projects. This review is leading communities to provide assistance to low-performing projects or even consider defunding them in favor of projects that are more efficient.

The depth and frequency of reporting has also been a critical factor in national decisionmaking. Knowing how many persons are chronically homeless, how many are veterans, and how many are families with children enables HUD to more strategically work with communities. For instance, when HUD saw an increase in family homelessness in 2009 and 2010, especially in less urban areas, the agency was able to target more CoC resources to this needy population. In part because of increased funding for family projects and communities’ use of the Homelessness Prevention and Rapid Re-Housing Program (HPRP) to serve families, family homelessness had declined by two percent by 2011.14

HUD recognizes the importance of letting all stakeholders review and comment on the homeless picture as it is depicted by HUD-collected data. Each year, HUD makes its HIC and PIT data publicly available and reports these data, as well as HMIS data, to Congress in its Annual Homelessness Assessment Report (AHAR). In the 2010 AHAR, HUD reported that in the last 10 days of January nearly 650,000 homeless persons were on the streets and in emergency shelters and transitional housing, and that over the course of the year approximately 1.59 million people spent at least one night in an emergency shelter or transitional housing program. These two figures demonstrate the tremendous churning in the homeless population. A closer examination reveals that homeless individuals in emergency shelters tended to stay for short periods of time, a finding that has been consistent year after year. The most recent report on 12-month sheltered data found that about one-third (34%) stayed a week or less in emergency shelter during a 12-month period, and 61 percent stayed less than a month.15

Knowing the data about homeless persons’ length of stay in emergency shelters has allowed policymakers to recognize that many — in fact most — homeless individuals do not need a permanent housing subsidy and supports to exit homelessness. Rather, a short-term intervention such as rapid re-housing is an effective and more efficient form of assistance for most homeless persons. Whereas nearly two-thirds of homeless persons who enter emergency shelters are homeless only for a month or less during the year, only 6 percent are homeless for more than 6 months during the 12-month period; these long-term homeless persons will typically need a more robust intervention, such as permanent housing with supportive services, to successfully exit homelessness and remain stably housed.

Other federal partners are using the data to make decisions and are encouraging their partners to use HMIS and similar databases. In 2010, the Obama administration, through the U.S. Interagency Council on Homelessness, published Opening Doors: Federal Strategic Plan to Prevent and End Homelessness, the first comprehensive plan to end homelessness nationwide. In the Federal Strategic Plan, the Obama administration set goals to end chronic and veteran homelessness by 2015 and family, youth, and child homelessness by 2020. HUD’s data have been crucial in setting these targets, tracking progress toward accomplishing those goals, and determining which partners and interventions are most effective in reaching these targets. (See “Tackling Veteran Homelessness With HUDStat,” p. 1).

Other agencies recognize the value of these data sources in achieving their objectives. In addition to jointly enumerating veteran homelessness with HUD each year through the PIT count, the U.S. Department of Veterans Affairs is beginning to adopt HMIS. For instance, the new Supportive Services for Veteran Families program requires grantees to participate in HMIS. Similarly, Projects for Assistance in Transition From Homelessness (PATH), the U.S. Department of Health and Human Services’ formula grant program, is implementing HMIS for its grantees; PATH serves individuals with severe mental illness who are homeless or at risk of homelessness. As is the case with PATH, HMIS is useful not only for people who are homeless but also those who are at risk of homelessness. Congress directed that grantees of the $1.5 billion HPRP program, which primarily served persons at risk of homelessness, must participate in HMIS. As of March 31, 2012, HPRP has served more than 1.3 million clients. By including those at risk of homelessness who then received prevention assistance in HMIS counts, communities can learn whether these persons become homeless over time.

Looking to the Future

Although HUD has made great strides in its data collection efforts, there is more to learn and do. The recently enacted Homeless Emergency Assistance and Rapid Transition to Housing (HEARTH) Act is pushing HUD to improve data collection at the community level. The HEARTH Act requires more critical analysis of recidivism and the nature of those experiencing homelessness for the first time. HUD will continue to encourage communities to analyze projects based on performance. Although a number of communities are using their data to evaluate performance and make critical decisions, HUD desires to instill that approach in all of its providers, resulting in effective projects that meet the needs of each community. HUD will continue to improve its data collection process to help the agency and its partners prevent and end homelessness in the United States.

 

Related Information:

HMIS data



  1. Anna Kondratas. 1991. “Estimates and Public Policy: the Politics of Numbers,” Housing Policy Debate 2:3, 631–3.
  2. U.S. Department of Housing and Urban Development. 1984. “Report to the Secretary on the Homeless and Emergency Shelters,” Washington, DC: Office of Policy Development and Research.
  3. U.S. Department of Housing and Urban Development. 1989. “A Report on the 1988 National Survey of Shelters for the Homeless,” Washington, DC: Office of Policy Development and Research.
  4. Stephen R. Poulin, Stephen Metraux, and Dennis P. Culhane. 2008. “The History and Future of Homeless Management Information Systems,” in Homelessness in America, Robert Hartmann McNamara, ed. Westport, CT: Praeger Publishers, 172–3; Dennis P. Culhane and Stephen Metraux. 1997. “Where to from Here? A Policy Research Agenda Based on the Analysis of Administrative Data,” in Understanding Homelessness: New Policy and Research Perspectives, Dennis P. Culhane and Steven P. Hornburg, eds. Washington, DC: Fannie Mae, 341.
  5. Dennis P. Culhane. 2008. “The Cost of Homelessness: A Perspective from the United States,” European Journal of Homelessness 2:1, 102–3.
  6. U.S. Census Bureau. “History: 1990 Overview.” (www.census.gov/history/www/through_the_decades/overview/1990.html). Accessed 11 June 2012.
  7. Diane F. Barrett, Irwin Anolik, and Florence H. Abramson. 1992. “The 1990 Census Shelter and Street Night Enumeration,” Washington, DC: United States Census Bureau.
  8. Bruce G. Link, Ezra Susser, Ann Stueve, Jo Phelan, Robert E. Moore, and Elmer Struening. 1994. “Lifetime and Five-Year Prevalence of Homelessness in the United States,” American Journal of Public Health 84:12, 1907–12.
  9. Culhane and Metraux, 343.
  10. Poulin, Metraux, and Culhane, 171.
  11. “Homeless Management Information Systems (HMIS); Data and Technical Standards Final Notice.” 2004. Federal Register 69. (www.gpo.gov/fdsys/pkg/FR-2004-07-30/html/04-17097.htm). Accessed 11 June 2012.
  12. U.S. Department of Housing and Urban Development. 2010. “Homeless Management Information System (HMIS) Data Standards, Revised Notice,” Washington, DC: Office of Policy Development and Research.
  13. Martha R. Burt and Barbara E. Cohen. 1988. “Feeding the Homeless: Does the Prepared Meals Provision Help? Report to Congress on the Prepared Meal Provision,” Vol. 1, Washington, DC: U.S. Department of Agriculture, 45.
  14. U.S. Department of Housing and Urban Development. 2011. “The 2011 Point-in-Time Estimates of Homelessness: Supplement to the Annual Homeless Assessment Report,” Washington, DC: Office of Policy Development and Research, 3.
  15. U.S. Department of Housing and Urban Development. 2011. “The 2010 Annual Homeless Assessment Report to Congress (2010 AHAR),” Washington, DC: Office of Policy Development and Research, 24.

Evidence Matters Home           Previous Article           Next Article



The contents of this article are the views of the author(s) and do not necessarily reflect the views or policies of the U.S. Department of Housing and Urban Development or the U.S. Government.