Pursuant to a congressional request, GAO compared the content of the 1980 and 1990 census questionnaires. GAO found that the 1990 questionnaire was similar to the 1980 questionnaire for both 100-percent and sample data, but that the Census Bureau: (1) included congregate housing and housing quality topics in the 1990 questionnaire; (2) dropped topics relating to housing unit identification; and (3) requested telephone numbers with 100-percent data for possible questionnaire follow-up.

Pursuant to a congressional request, GAO reviewed the results of the Census Bureau's 1980 post-census improvement programs, which helped reduce errors in census counts. GAO found that the Bureau: (1) used 14 coverage improvement programs for the 1980 census, three of which involved address list operations performed before the census; (2) conducted 11 post-census programs, including a comparison of persons counted in the census to those included on records independent of the census; (3) spent $73 million on the programs, which resulted in an increased population count of about 2.6 million persons and the assignment of 1 million persons to different geographic locations; (4) decided to delete the Nonhousehold Sources Program from the 1990 census because of its low coverage and high costs; and (5) determined that the programs also resulted in double counting 218,000 persons.