Census Bureau facing several court challenges; officials refuse to release sampled data

Wednesday, March 27th 2002, 12:00 am

By: News On 6

WASHINGTON (AP) _ Court challenges to Census 2000 abound nearly two years after the head count. Lawyers for Utah are going to the Supreme Court to try to get back a House seat, while some Democrats are trying to force release of data drawn from statistical sampling.

At the court on Wednesday, Utah lawyers will argue that the state was deprived of a House seat because of a statistical technique they say distorted the census.

Utah is asking the justices to overturn a lower-court ruling that validated the use of this procedure, known as ``imputation.'' This particular Census Bureau device amounts to an estimate, insofar as it permits officials to use secondhand information.

This comes into play when Census officials get incomplete responses to mailed questionnaires and cannot get enumerators onto the premises; Census officials say their enumerators make up to six attempts to get information firsthand before turning to imputation, which permits them to glean information from neighboring households.

A 1999 Supreme Court decision outlawed using statistically sampled data for reapportioning House seats after each census.

Utah claims imputation is similar to sampling, and likewise should not be used. At stake is the House seat that Utah narrowly lost to North Carolina in the reapportionment following the 2000 census. The Bush administration has urged the high court not to get involved.

``This puts a new twist to an old argument,'' said Kimball Brace, president and an analyst with Election Data Services, Inc. in Washington. ``It's an issue that if the Supreme Court doesn't do anything with this time, it's going to come back to them,'' said Brace, who has been analyzing elections since 1976.

In other census cases around the country:

_Two Democratic state senators in Oregon say the Freedom of Information Act allows for the release of statistically sampled data that the Census Bureau has declined to put out. A federal judge in Oregon has ordered that the information be released and the Justice Department has appealed the decision.

_A federal judge in Texas, in a case involving several counties and cities, ordered the release of sampled data based on the plaintiffs' Freedom of Information Act request. Justice officials have not determined if they will appeal.

_A group led by Los Angeles officials is challenging the process by which the bureau and its parent agency, the Commerce Department, decided against releasing adjusted data for purposes of redistricting and distributing federal funds. That case is scheduled to be argued on April 9.

_Democrats on the House Government Reform Committee contend a panel rule gives committee members special access to government records and, as a result, the adjusted data. A federal judge in Los Angeles accepted the Democrats' position, and on Monday he denied the Justice Department's request to reconsider and dismiss the case. Justice officials could not be reached immediately for comment.

In the Utah case, the imputation procedure accounted for less than 1 percent of the total U.S. population of 281 million in 2000. Utah claims that was enough to give North Carolina the House seat it should have gotten.

The Justice Department argues that imputation has long been used by the bureau and differs from sampling. Sampling is a method of collecting data, while imputation is a way to process data that has already been collected, the government argues.

Sampling provides information beyond the actual head count. It is based on a follow-up survey of 314,000 households, designed to measure the accuracy of the actual count. Supporters say this procedure helps protect against an undercount of minorities.

The Census Bureau refuses to release sampled data, contending it would insert more errors into the head count. The bureau maintains the 2000 undercount was less severe than in any of the previous censuses.

Preliminary estimates show the national net undercount for 2000 could have been as low as 1.7 million, compared with 4 million in 1990.