The Census Bureau already is offering a spirited and detailed defense of its plans for the census three years from now and says it expects to be sued if it does what it wants to do.
Bureau director Martha Farnsworth Riche and several groups that are vitally interested in the accuracy of the census are urging support for statistical sampling, which the bureau plans to use for the first time to complete and correct its count of the population.
The bureau also plans to continue using a controversial long-form questionnaire to document details of daily life.
In using the traditional door-to-door counting system, the 1990 Census is widely acknowledged to have missed many people, particularly minorities in cities.
At a briefing Friday designed to gain support from Congress and the public for its plans for 2000, Riche said that without using a system of sampling the population, "we will be left with a census less accurate than it should be and one that costs more than it needs to."
As for the long questionnaire, Riche said that she hoped lawmakers would resist political pressure to do away with it. The bureau has said that it will be shorter than the 1990 version, which ran to some 20 pages, and that one household in six will get it in addition to the standard short form. It will ask questions about income, how people in a household get to work and other habits of Americans' lives.
The Census Bureau has until April 1 to submit its proposed questions for the long form to Congress. The census director said research has shown that the questions about household income are the most sensitive.
Riche's remarks were directed not just at Congress, several of whose committees in the House and Senate have jurisdiction over the census, but in a broader sense at residents, whose cooperation the bureau will need in three years if the survey is to succeed.
The plan to use statistical sampling, the continuation of the long form and the question of money _ the bureau expects the 2000 Census to cost some $4-billion _ are sensitive issues at a time when Congress is torn by a debate over how to best balance the budget.
Some members of Congress have suggested that the long-form questionnaire should be discarded as too cumbersome and, at $300-million, too costly. But the information it gathers is widely used, helping to determine where stores are built, where roads should be widened, where banks should install automated teller machines.
The issue of statistical sampling has economic, political and constitutional implications. The bureau hopes to survey at least 90 percent of the households in each census tract and then use statistical sampling to estimate the number of people it missed.
Bureau officials have said that a census tract (typically several blocks in a city) is small enough that with statistical sampling there should be no severe undercounting.
The number of people living in a given area determines how much federal assistance goes to a local government and, in some instances, how much state aid, which is why officials in big cities and representatives of minority groups support statistical sampling.
But so do real estate agents, home builders, planning agencies and mortgage bankers. An official with Housing Statistics Users Group, an umbrella organization for those interests, joined Riche to support sampling.
The official, David Crowe, said the varied people who belong to his organization need to have the right numbers when making decisions that affect lives.