The Demographics of Homelessness - How Numbers Are Used

The ordinary citizen, hearing of the homeless, envisions people, including children, who live on the street permanently and sleep in cardboard boxes under bridges or in cars. There are, of course, people in this category, but they are the minority among the homeless. HUD has labeled such people the chronically homeless and estimates their number at around 150,000, close to the number of people counted by the Census Bureau as inhabiting emergency and transitional shelters in the 2000 census count (170,706 individuals). Most of the homeless are not chronically homeless but are temporarily without a residence. After some period of homelessness, they find permanent shelter or move in with relatives; people who have moved in with family as well as people who are TABLE 2.1
Number likely to be homeless at least once during the year, 1996
SOURCE: "Number Likely to be Homeless at Least Once in a Given Year," in America's Homeless II: Populations and Services, Urban Institute, February 2000, http://www.urban.org/UploadedPDF/900344_AmericasHomelessll.pdf (accessed February 10, 2005)

	New homeless spells begun in last week	Average week estimate	Annual projection
	A	B	C
February 1996	52,000	842,000	3.5 million
October 1996	36,900	444,000	2.3 million
Note: The projection is developed by taking column A times 51 weeks and adding the result to column B. Column B represents the estimated constant population of homeless in any one week. The assumption is that a population of the size shown in column A is continuously passing into and also out of homeless status throughout the year. Data for February were based on the estimates of homeless program employees, data for October on interviews with the homeless.

"doubled up" are also counted as homeless by some programs and homeless advocates.

A more accurate definition of the homeless population is the group of people who are, on any day, without proper shelter. When agencies or the media cite numbers in the 600,000-800,000 range, they mean the size of the homeless population at any one point in time. Individuals are continuously joining this population while others are leaving it. If all people who are homeless at some point during a given year were counted, the number would reach between 2.5 and 3.5 million individuals, as indicated by HUD in Evaluation of Continuums of Care for Homeless People.

The manner in which the annual projections for 1996 were derived is shown in Table 2.1. The data for October, projected from counts of homeless services seekers, show that an estimated 36,900 individuals began spells of homelessness during the week surveyed, while the total number of people in the homeless population in any one week was estimated to be 444,000. The annual projection assumed that each week, 36,900 became homeless and an equal number passed out of the homeless status. Multiplying 36,900 by the fifty-one weeks remaining in the year, and then adding that total to the average homeless population in a week, produced the 2.3 million count of people who were homeless at least once in 1996. The number does not mean that there were 2.3 million homeless during the entire span of 1996.

Counting Children

Sometimes stories in the media cite 600,000 homeless and one million homeless children (see for instance "Are Shelters the Answer to Family Homelessness," USA Today, January 1, 2003). Such statements double count the homeless by using two different sources of incompatible data. Under the McKinney-Vento Homeless Assistance Act, the U.S. Department of Education (USDE) is required to file a report on homeless children served by the act. USDE obtains the data from school districts; school districts use different methods of estimation. In its 2000 report to Congress (Education for Homeless Children and Youth Program, Washington, DC), USDE estimated that 930,032 children experienced homelessness at some point during the year. This number was much higher than the number of children who were homeless on a particular night during the year.