The Nature of Homelessness - Counting The Homeless

Methodology

An accurate count of the U.S. homeless population has proved to be a problem for statisticians. The most formidable obstacle is the nature of homelessness itself. Typically, researchers contact people in their homes using in-person or telephone surveys to obtain information regarding income, education levels, household size, ethnicity, and other demographic data. Since homeless people cannot be counted "at home," researchers have been forced to develop new methods for collecting data on these transient groups. Martha Burt explored this issue for the U.S. Department of Housing and Urban Development (HUD) and the U.S. Department of Health and Human Services (DHHS) and published a table of the most common methods of data collection for homeless people. (See Table 1.2.)

If each and every person without a home could be counted it would be the most accurate way to establish the number of homeless people. Such a count is almost impossible. One way to estimate the number of homeless people is to search records at homeless service provider locations. Alternatively, sampling of those records combined with projections, called "probability-based methods," can be used to count the number of homeless. Another way to count the homeless is to count the number of homeless at one particular time in one particular place. This "snapshot" method estimates the number of homeless at any one time. Longitudinal studies are a way to estimate the proportion of people in a population who may become homeless at some point in their lives. These studies follow individuals over a period of time to determine if they become homeless (Anita Drever, "Homeless Count Methodologies: An Annotated Bibliography," Institute for the Study of Homelessness and Poverty, February 1999).

TABLE 1.2
Common methods for collecting planning information SOURCE: Martha R. Burt, "Table 3. Common Methods for Collecting Planning Information," in "Demographics and Geography: Estimating Needs,"Practical Lessons: The 1998 Symposium on Homelessness Research, edited by Linda B. Fosburg and Deborah L. Dennis, U.S. Department of Housing and Urban Development and the U.S. Department of Health and Human Services, August 1999, http://aspe.os.dhhs.gov/progsys/homeless/symposium/1-demograp.htm (accessed February 2, 2005)

COMPARING METHODS AND RESULTS.

As Table 1.2 reveals, methods vary in scope and design. Different designs will produce different results even if the intention is the same—namely to enumerate the homeless population. Table 1.3 shows results of surveys conducted by the Association of Gospel Rescue Missions (AGRM) in 1994, 2003, and 2004. The data presented in Table 1.3 are based on the "snapshot" method—counts of a population at a point in time. AGRM counted all people receiving homeless services during one specific night in each year. Table 1.4 shows results from an Urban Institute study conducted in 1996. Data in Table 1.4 are based on a sample of seventy-six geographical areas selected by the Urban Institute as being representative of all service providers in the United States. The Urban Institute then compared its results by demographic characteristics to the total population as enumerated by the U.S. Census.

The male/female ratios in the AGRM study are quite different from the Urban Institute's study, with AGRM finding that males were more than three-quarters of the homeless (77% in 2004), whereas the Urban Institute's study showed that males were just over two-thirds of the homeless population (68% in 1996). (See Figure 1.1.) Both studies showed that males outnumbered females among the homeless, but the proportions were different. The Census Bureau estimated that in July 2003, women outnumbered men in the U.S. population by a small margin—50.8% of the population were female and 49.2% male.

The Official Count: The U.S. Census Survey

The official U.S. census, which takes place at ten-year intervals, is intended to count everyone in the United States. The results of the census are critical for determining how much federal money goes into different programs and to various regions of the country. Representation of the population in Congress is also based on the census. Since the U.S. Census Bureau counts people in their homes, counting the homeless presents special challenges.

A PROBLEM OF METHODOLOGY.

In 1990 census officials, on what was known as Shelter and Street Night, or TABLE 1.3
Demographic overview of homeless population, 1994-2003 SOURCE: "Data from the 2004 AGRM Fall Snapshot Survey of the Homeless," Association of Gospel Rescue Missions, 2004, http://www.agrm.org/statistics/snap04_data.html (accessed February 2, 2005)

S-Night, counted homeless persons found in shelters, emergency shelters, shelters for abused women, shelters for runaway and neglected youth, low-cost motels, Young Men's Christian Associations (YMCAs) and Young Women's Christian Associations (YWCAs), and in subsidized units at motels. Additionally, they counted people found in the early morning hours sleeping in abandoned buildings, bus and train stations, all-night restaurants, parks, and vacant lots (Diane F. Barrett et al., "The 1990 Census Shelter and Street Night Enumeration," U.S. Census Bureau, 1992). The results of this count were released the following year in the Census Bureau publication, "Count of Persons in Selected Locations Where Homeless Persons Are Found." Homeless advocates criticized the methods and results as inadequate and charged that they provided a low estimate of TABLE 1.4
Basic demographic characteristics of homeless and formerly homeless individuals, 1996
SOURCE: Martha R. Burt et al., "Table 3.1. Basic Demographic Characteristics, by Homeless Status," in Homelessness: Programs and the People they Serve: Findings of the National Survey of Homeless Assistance Providers and Clients, Urban Institute, December 1999, http://www.huduser.org/publications/homeless/homelessness/ch_3b.html#fig3.1 (accessed February 2, 2005)

homeless people in the United States. In response, according to Annetta and Denise Smith in Emergency and Transitional Shelter Population: 2000 (U.S. Census Bureau, October 2001), the Census Bureau emphasized FIGURE 1.1
Homeless population by gender, as reported in two studies, 1996 and 2004
SOURCE: Adapted from Martha R. Burt et al., "Table 3.1. Basic Demographic Characteristics, by Homeless Status," in Homelessness: Programs and the People they Serve: Findings of the National Survey of Homeless Assistance Providers and Clients, Urban Institute, December 1999, http://www.huduser.org/publications/homeless/homelessness/ch_3b.html#fig3.1 (accessed February 2, 2005); and Association of Gospel Rescue Missions, "Data From the 2004 AGRM Fall Snapshot Survey of the Homeless," 2004, http://www.agrm.org/statistics/snap04_data.html (accessed February 2, 2005) that S-Night "should not be used as a count of people experiencing homelessness." S-Night results were not a reflection of the prevalence of homelessness over a given year, but rather a count of homeless persons identified during a single night, a "snapshot," like the census itself.

CENSUS ACCUSED OF UNCONSTITUTIONALITY.

The National Law Center on Homelessness and Poverty alleged that the methodology of the S-Night count was unconstitutional. In 1992 the Law Center, the U.S. Conference of Mayors, the cities of Baltimore and San Francisco, fifteen local homeless organizations, and seven homeless people (the plaintiffs) filed suit in the federal district court in Washington, D.C. They charged the Census Bureau with excluding segments of the homeless population in the 1990 population count by not counting those in hidden areas and by not allocating adequate funds for S-Night.

In its suit, the Law Center cited an internal Census Bureau memorandum that stated, in part, "We know we will miss people by counting the 'open' rather than 'concealed' (two studies showed that about two-thirds of the street population sleep concealed)." Studies funded by the Census Bureau indicated that up to 70% of the homeless street population in Los Angeles were missed, as were 32% in New Orleans, 47% in New York City, and 69% in Phoenix. Advocates were greatly concerned that this underrepresentation would negatively affect the funding of homeless initiatives.

In 1994 the district court dismissed the case, ruling that the plaintiffs' case was without merit. The court ruled that failure to count all the homeless was not a failure to perform a constitutional duty; the Constitution does not give individuals a right to be counted or a right to a perfectly accurate census. The court stated that the "methods used by the Bureau on S-Night were reasonably designed to count as nearly as practicable all those persons residing in the United States and, therefore, easily pass constitutional muster." In 1996 the U.S. Court of Appeals upheld the district court's finding (United States Court of Appeals, for the District of Columbia Circuit, Argued October 6, 1995, Decided August 9, 1996, No. 94-5312, National Law Center on Homelessness and Poverty, et al., Appellants v. Michael Kantor, et al., Appellees, Appeal from the United States District Court for the District of Columbia, 92cv2257).

CENSUS 2000 LIMITS INFORMATION.

The Bureau of the Census undertook a special operation, called Service-Based Enumeration (SBE), for the 2000 census. From March 27 through March 29, 2000, census workers focused solely on counting the homeless population at the locations where they were most likely to be found. For the SBE, the Census Bureau released the following schedule:

• Monday, March 27, 2000—Emergency and transitional shelters, hotels, motels, or other facilities. Enumerators will leave blank questionnaires for residents who usually stay at the shelter, but who are away at the time of the enumeration.
• Tuesday, March 28, 2000—Soup kitchens, regularly scheduled mobile food vans.
• Wednesday, March 29, 2000, from 4 A.M. to 7 A.M. only—Outdoor locations. Census workers will complete the census forms for each person at an outdoor location.

The SBE methods were considered an improvement over the methods used in the 1990 census survey. Homeless citizens and advocates alike expected to see an increase in the number of homeless persons reported by the Census Bureau in the 2000 census as compared with the count reported for the 1990 census. Expectations that the higher population counts would translate into higher funding levels for services to the homeless were also raised.

An Associated Press story dated June 27, 2001, reported that the U.S. Census Bureau would not be TABLE 1.5
City data on homelessness, 2003-04
SOURCE: Eugene T. Lowe, et al., "City Data on Homelessness," in Hunger and Homelessness Survey: A Status Report on Hunger and Homelessness in America's Cities, A 27–City Survey, December 2004, The U.S. Conference of Mayors, December 2004, http://www.usmayors.org/uscm/hungersurvey/2004/onlinereport/HungerAndHomelessnessReport2004.pdf (accessed February 2, 2005)

releasing a specific homeless count because of the liability issues raised after the 1990 census. The Census Bureau stated that it would have only one category showing the number of persons tabulated at "emergency and transitional shelters." The people who, in 2000, were counted at domestic-violence shelters, family crisis centers, soup kitchens, mobile food vans, and targeted non-sheltered outdoor locations (i.e. street people, car dwellers, etc.) during the March 2000 SBE night were to be included in the category of "other non-institutional group quarters population." This category was overly inclusive; it included, for instance, students living in college dormitories. The homeless portion of the category could not be extracted.

Rather than release counts of all homeless people, the Census Bureau published Emergency and Transitional Shelter Population: 2000, a special report on people sleeping in shelters. Census Bureau officials said the homeless people they did find during the exhaustive, three-day SBE count were included in total population figures for states, counties, and municipalities. Researchers voiced concern that the numbers teased from these data sets would be flawed.

People involved in the receipt or delivery of services to the homeless were worried that their programs would suffer from the lack of SBE night information. A detailed homeless count was thought to be essential for city officials and advocacy groups to plan budgets for shelters and other homeless outreach programs. Results from the U.S. Conference of Mayors 2004 study illustrated the negative impact that inadequate information and funding can have on the delivery of human services. (See Table 1.5.) For example, the needs of 54% of homeless people for shelter could not be met in Los Angeles due to lack of resources. Homeless program funding for most cities was already strained. Two-thirds of cities surveyed in 2004 showed increased requests for emergency shelter services.

Only Estimates Are Available

The actual number of homeless people is unknown. The Urban Institute estimated that 3.5 million people were homeless at some point of time during the year 1996 (America's Homeless II—Populations and Services, Urban Institute, February 1, 2000). The 2000 Census counted 170,706 individuals in emergency and transitional TABLE 1.6
Population in emergency and transitional shelters, 1990 and 2000
SOURCE: Annetta C. Smith and Denise I. Smith, "Table 1. Population in Emergency and Transitional Shelters for the United States, Regions, States, and Puerto Rico: 1990 and 2000," in Emergency and Transitional Shelter Population: 2000, Census 2000 Special Reports, CENSR/01-2, U.S. Census Bureau, October 2001, http://www.census.gov/prod/2001pubs/censr01-2.pdf (accessed February 12, 2005).

shelters, down from 178,638 individuals in 1990. (See Table 1.6). The Census Bureau expressly stated that this number was not a total count of the homeless. The U.S. Department of Housing and Urban Development estimated in 2005 that 150,000 people were chronically homeless—homeless for a year or more—and stated that this population was only about 10% of all homeless individuals, putting the homeless population at 1.5 million people (News Release, U.S. Department of Housing and Urban Development, HUD No. 05-007, January 25, 2005).