Improving Our Measure of America: What Census 2000 Can Teach Us in Planning for 2010

摘要

An undertaking as huge and complex as the decennial census requires long lead times to allow for proper development and testing of the overall design and to define and procure systems. Therefore, an agreed-upon design with sufficient funding to support its development must be in place early. The drawn-out dispute over sampling precluded early agreement on Census 2000's design and increased problems in its execution. Our review of the Bureaus readiness to conduct the 2000 decennial revealed that as late as 1995 Census still did not have a design that was sufficiently finalized to undergo full-scale testing. In 1998, as the dress rehearsal approached, many in Congress were opposed to the use of sampling and required the Bureau to test two census designs - one that relied on sampling and one that did not. A final decision on the design did not come until 1999, when the United States Supreme Court ruled that sampling could not be used to apportion the House of Representatives, in effect, telling Census that it must concentrate on the traditional approach of enumerating every household in the nation.