Unadjusted data recommended for redistricting

Census inconsistencies raise the possibility of `unidentified error'

March 02, 2001|By Cox News Service

WASHINGTON - The Census Bureau recommended yesterday that the government issue unadjusted results from the 2000 Census as the basis for states to redraw political boundaries, despite evidence that about 3.3 million Americans were missed in the count.

The decision by a panel of the agency's senior staff and statistical experts eased the political heat on the Bush administration, which has been under pressure from Democrats and civil rights groups to use adjusted data to compensate for people - predominantly minorities, children and the poor - who were not counted in the national tally.

The agency's experts were "unable, based on data and other information currently available, to conclude that the adjusted data are more accurate for use in redistricting," acting Census Bureau director William Barron said in a letter to Commerce Secretary Don Evans.

The final decision to adjust or not rests with Evans, a political adviser to President Bush. Evans said he will decide by Tuesday, adding that the Census Bureau recommendation will be "a critical element of my deliberations."

Later this month, the Census Bureau is scheduled to begin releasing state-by-state population figures for the purposes of redistricting and as the basis for about $200 billion a year in federal spending. By law, those census numbers must be provided to the states by April 1.

The issue of an adjusted census, a process also known as sampling, has spawned an intense political battle because of the enormous stakes.

Democrats dubbed it "the civil rights issue of the decade," saying that scientific sampling is the only way to guarantee that all Americans are fairly represented in the final census numbers.

Reacting to yesterday's announcement, congressional Democrats and civil rights groups urged Evans to give the Census Bureau experts more time to refine their data in hopes that a valid set of adjusted numbers could be furnished.

"The bad news is millions of Americans had the clock run out on them," said Rep. Carolyn Maloney, a New York Democrat and member of the House subcommittee on the census.

Republicans contend that adjusted data is merely a means of boosting the census numbers - and political clout in urban Democratic strongholds - with people who may or may not be there. The Constitution calls only for an "actual enumeration," they say.

The Census Bureau's internal committee has been meeting since September to review the results of the national count and compare it in a complex quality-control review against the "accuracy and coverage evaluation," a sample of 314,000 households nationwide. By comparing the census with a sampling of the population, the demographers attempted to discern whether adjusted numbers would offer a better picture of America.

The experts were pressed for time as state legislatures across the country prepare to redraw congressional and legislative districts ahead of the 2002 elections. But the panel ran into a tough problem it couldn't solve - inconsistencies in the unadjusted figures and the sampling data that may have been used to adjust the census results.

Barron said yesterday that the inconsistencies "raised the possibility of an unidentified error" in methods used either in the census or the follow-up sampling.

However, Barron said that the panel's review showed that the 2000 Census was perhaps the best in U.S. history and resulted in a significant reduction in the historic undercounts of blacks, Hispanics, Asian-Americans and American Indians.

The Census Bureau found that the national undercount in 2000 was 1.18 percent, or about 3.3 million of America total population in 2000, 281 million. The net undercount in 1990 was 1.6 percent, or about 4 million people of the total at the time.

Despite improvements, the chances of being missed were still greatly higher for minorities than for whites.

In 1990, 4.57 percent of blacks were missed compared to an undercount of 2.17 percent in the 2000 Census. Among Hispanics in 1990, about 5 percent were not counted compared to 2.85 percent in 2000.

Whites, as in the past, had the lowest undercount rate among the racial categories. In 1990, 0.68 percent were missed compared to 0.67 percent in 2000.

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.