Summary: In response to congressional requests, GAO examined the Census Bureau's 1990 decennial census planning effort, focusing on the 1990 short-form questionnaire, since it is a major factor affecting response rates, quality of response, respondent burden, and data processing requirements.
Because the Bureau incurred the majority of 1980 census costs in the collection, preparation, and processing of data, its efforts to reduce the 1990 questionnaire both in size and questions could be cost-beneficial. GAO believes that: (1) the Bureau's short-form questionnaire should be limited to the basic population and housing questions in order to obtain an accurate population count; (2) if the Bureau collected those data only on the long form, about 85 million households would be required to answer eight fewer questions; (3) the housing questions increase the complexity of the questionnaire and tend to discourage responses; and (4) a shorter form may provide the Bureau with cost-saving options for its data automation decisions. GAO found that: (1) the need for a housing data questionnaire from 100 percent of the households appears questionable; (2) the Bureau did not fully substantiate and evaluate user requirements for the decennial data questionnaire before it placed questions on the form; (3) some data users have used sample data questionnaires even though 100-percent data questionnaires were available; (4) some data users had requested 100 percent housing data questionnaires for geographical levels for which data were also estimated from sample questionnaires; and (5) the Bureau did not weigh other data users' needs against the costs of data collection.
