Tentative pact OK'd on census method Statistical sampling aims at better accuracy in cities


WASHINGTON -- Negotiators from the White House and Speaker Newt Gingrich's office have reached a tentative agreement that would allow the Census Bureau to test its hotly debated new method for estimating the country's population, administration and congressional officials said yesterday.

The deal, the details of which must be fleshed out and sold to Republican and Democratic members of the House, would allow the Census Bureau to test the use of statistical sampling next year, a procedure the bureau would like to use for the next census, in 2000. At the same time, the agreement requires the bureau to conduct a second test next year, this one of a census using traditional methods, but employing more advertising, more mailings of census forms and a higher ratio of head counters going door to door than have been used in the past.

The use of sampling -- estimating a figure based on a count of a smaller representative group -- has been attacked by Republicans who view it as legally suspect and politically harmful. The bureau has said that sampling is the only way to avoid a repetition of the 1990 census, which was the most expensive in history but still managed to miss more than 10 million people, many of them minorities in cities. Six million other people were counted twice or in the wrong place.

The agreement also calls for an independent panel to oversee the 2000 census to assuage Republican concerns that the administration will use sampling figures to manipulate the census numbers for partisan gain. Republicans are worried that they will lose some legislative districts if more minorities, who to a large extent vote Democratic, are counted and districts are redrawn.

In addition, the administration has agreed not to fight a Republican bill that calls for speeding up the Supreme Court review of the constitutionality of sampling.

The fight over its possible use has held up passage of the spending bill that funds the Commerce, State and Justice departments, as well as the federal judiciary.

Under the tentative deal reached at the White House on Thursday night, the Census Bureau would conduct two dress rehearsals next spring of the different methods they want to employ in the 2000 census. In Sacramento, Calif., the bureau will mail out forms and, depending how many are returned from a given neighborhood, send head-counters to the residences of a sample of those who did not respond. From that sample, the bureau will extrapolate the number of and characteristics of those who did not return the forms.

In a rural area in and around Columbia, S.C., the bureau will employ more traditional methods. It will send out several mailings, and then count door to door to try to find those who did not reply. In both tests the bureau will use sampling statistics to evaluate the accuracy of each method.

Administration officials said the two tests would allow them to evaluate the accuracy of plans to use sampling statistics. And it would also allow the bureau to test a souped-up version of its 1990 census, in case the Supreme Court ruled that the use of sampling was unconstitutional.

A Commerce Department official who spoke on condition of anonymity emphasized that the two tests were not a head-to-head competition between the two methods.

Pub Date: 11/01/97

Baltimore Sun Articles
Please note the green-lined linked article text has been applied commercially without any involvement from our newsroom editors, reporters or any other editorial staff.