The PAA Committee on Population Statistics (Linda Gage, Chair) has been following the Census 2000 planning process since 1990. The committee met in November with census officials, congressional staffers, and other key players. There is still much uncertainty over the 2000 census, but the suspense cannot last much longer. By April 1997, the Census Bureau must submit to Congress the list of data items it proposes to collect in the 2000 census, and by April 1998, the bureau must submit proposed wording for questions. The 1998 "dress rehearsal" is not a design test, but an operational dry run. In short, we are rapidly approaching the point where everything must be ready to go. The Census Bureau has been working hard to meet deadlines, but congressional relations have become a critical factor. Congress made it clear from the outset that it expected an improved population count and reduced costs. The Bureau responded with a serious planning effort and a plan for a census which would improve the population count, contain census costs, and even provide for the nation's data needs. With just three years to go, Congress is balking at major parts of the census design, and we still face the threat of a 2000 census that provides very little data. The most contentious issue now is sampling and estimation. The 2000 design includes measures to increase mailback response, but with public cooperation eroding, these measures might not maintain response at 1990 levels. Nonresponse will be significant in 2000, and how it is handled will have a major impact on costs.
Resistance to sampling was anticipated, but it is now real and intense. Congressional critics grant the validity of sampling for survey purposes, but denounce its use in the basic census count as a subjective exercise in creating phantom people. On the surface the dispute is perplexing. Congress called for a redesigned census that could improve the count and save money. The Census Bureau responded with a sampling proposal, but now Congress appears ready to reject the most critical innovation, while still expecting count improvement and cost savings. The situation is better understood in the context of a changing congressional landscape. The 1994 elections transferred census oversight to committees with little interest in the census. There is concern that congressional input may default to the appropriations committees, where the census is pitted against popular causes, such as "cops on the beat," in the competition for truly limited funds. A Congress that sees content reduction as an easy ticket to count improvement and cost reduction may be little moved by arguments for sampling and estimation. That the census will collect only information required by law for federal purposes matters little. The long form is dismissed as corporate welfare, and arguments stressing the need for data tend to sink without a ripple. Thus, with funding scarce and Congress preoccupied with other priorities, census sampling is on the ropes, and the threat to the long form appears as serious as ever.
Following the census has become an exercise in following Congress as much as the Census Bureau. Given the timetable, the 105th Congress can play a makeorbreak role, and its subcommittee assignments are at least as relevant as census test results. Even following such developments does not make one confident in predicting ultimate outcomes. However, with budgets severely limited, and Congress at odds with the Bureau over major issues, it is clear that many insidethebeltway sorts of things will have to go right if we are to see a 2000 Census that delivers the content and quality we have come to expect.