U.S. Census Bureau and homeless advocates figures on homelessness are as wildly different as a penthouse in New York's Trump Tower is from the Rescue Mission at Los Angeles' Fifth and Wall Streets.
The gap, of about 629,000 people, highlights Census limitations in taking a single-day snapshot of the shelter population. But a few states are collecting more accurate figures by developing computerized information systems to track their homeless.
A recent congressional directive requires communities receiving federal homeless assistance aid to start making unduplicated counts of people served in homeless programs by 2004. Most counts now are conducted at the city level, but Wisconsin and Massachusetts have begun to make computerized statewide counts. Rhode Island, Delaware, and Iowa are moving toward similar systems.
Localities in Illinois, Florida, Vermont, Oklahoma, Maryland, California, North Carolina, South Carolina, Washington, Michigan, Tennessee, Ohio, Texas, and Alaska are also developing computerized tracking systems.
In Massachusetts, 60 percent of shelters provide homeless population counts for the statewide data-collection system.
The system "can reduce the suffering for people who are seeking help when they're homeless," said Donna Friedman, director of the McCormack Institute's Center for Social Policy, University of Massachusetts, which houses the system. "Most homeless people actually want to have a conversation with somebody who can help them."
Computerized information systems are more reliable for estimating the homeless population over the course of a year than point-in-time surveys, Friedman said.
But the expense and difficulty of creating an easy-to-use system that respects confidentiality probably has kept more states from tracking their homeless, said Julie Hovden, community services specialist for the Wisconsin division of housing and intergovernmental relations.
About 40 percent of Wisconsin's shelters provide data for a statewide, computerized system in a three-year project that cost about $850,000.
No uniformity exists on how states number their homeless. Counting is made more difficult because homelessness is often a temporary circumstance, and some homeless people tough it out on the streets without entering shelters where they could be counted.
"You want to make sure that you're not counting the same person two or three times which can happen as people try to protect their identity when they're in a homeless situation," said Robert Kaylor, resident manager of a single-room occupancy shelter for Boston's Haley House. Kaylor was homeless himself during 1994. "If you're homeless, it's rather embarrassing. You don't want people to know your situation."
Tracking homelessness can help assess the outcome of state programs such as welfare, foster care, and state mental health programs, said Dennis Culhane, professor at the University of Pennsylvania's School of Social Work.
"States could take the lead in trying to help jurisdictions implement information systems, providing funding for that kind of effort, and I think they should require it," said Culhane. "Absent a requirement, you find that people will drag their feet and avoid implementing it."
Culhane's research concluded that it costs nearly the same amount of taxpayer money to treat mentally ill homeless people on the streets as it does to give them permanent housing.
Census figures released Oct. 30 showed that New York, Delaware, Washington, Alaska, and Oregon had the highest percentages of their residents sleeping in homeless shelters on March 27, 2000, the day the count was made. Arkansas had the lowest percentage of its people in shelters.
While Census-takers found just 170,706 people in the nation's shelters, an Urban Institute study estimated last year that at least 800,000 people are in shelters on any given night.
In 1990, the Census also faced heavy criticism so this year, Census officials decided not to release a complete, separate homeless count. Despite spending $10 million to count the homeless last year in soup kitchens and other places, the Census Bureau released figures on only those people found in homeless shelters, transitional housing for people without homes, and motels used as temporary housing. The Census numbers don't reflect shelter usage over time, the souls turned away on March 27, or those sleeping on the street.
The Census' "shelter coverage was not in any way close to complete. It's so low as to be ridiculous," said Martha Burt, director of the social services research program at the Urban Institute. "States could keep track of homeless numbers better in general, but if you'd want to be able to compare them, they'd all have to agree, and would you like to bet on that?"