Limitations in Methods and Procedures to Include the Homeless
GGD-92-1: Published: Dec 30, 1991. Publicly Released: Dec 30, 1991.
- Full Report:
Pursuant to a legislative requirement, GAO assessed the methodology and procedures the Census Bureau used in counting the number of homeless persons for the 1990 decennial census, focusing on the Bureau's Shelter and Street Night (S-Night) operation, a program designed to include select components of the homeless population in the census.
GAO found that: (1) S-Night was not designed to and did not provide a complete count of the nation's homeless population, but was designed to count selected components of the homeless population; (2) the number of persons who might be considered homeless who were counted through other census efforts could not be fully identified in census data; (3) the Bureau's street count results from S-Night also did not represent a complete count of the homeless street population primarily because the hidden homeless would not be included in the census and the Bureau did not design procedures to guard against counting persons who may be counted during other census operations; (4) a 1989 Bureau test showed that counting the homeless at daytime service locations offered promise as an alternative way to address the limitations of a nighttime street count, but it was too late to conduct additional research and incorporate the method into the 1990 census; (5) operational problems further undermined the accuracy of the S-Night street counts because the quality of address lists varied across the nation and the Bureau experienced implementation problems; (6) the Bureau relied primarily on local governments to identify street locations where the homeless could be found and 36 percent of the local government units responded; and (7) in areas where the Bureau conducted a street count, the accuracy was questionable because about half of the on-site observers reported that they did not see an enumerator at designated sites.