Census Bureau recommends against sampling data

Dealing a serious blow to congressional supporters of statistical sampling, a panel of Census Bureau experts Thursday recommended against using the controversial method for producing redistricting numbers, concluding that while the bureau's sampling operation produced quality work, the panel could not guarantee that it would be more accurate than more conventional methods.

"The [committee] is unable to conclude, based on the information available at this time, that the adjusted census 2000 data is more accurate for redistricting," the panel wrote in its review. "Accordingly [the committee] recommends that the unadjusted census data be released as the Census Bureau's official redistricting data."

Minutes after hearing the news from Commerce Secretary Donald Evans, House Government Reform Census Subcommittee Chairman Dan Miller, R-Fla., a sampling opponent, cheered the news. "It puts the whole issue to rest," Miller said. "Everyone said, `Listen to the experts.' Do not adjust."

Rep. Carolyn Maloney, D-N.Y., has championed sampling in Congress and has strongly advocated allowing Census Bureau professionals, not Evans, to judge the accuracy of sampling as a way to include undercounted populations, most often minority groups.

Maloney responded with a statement expressing her disappointment, but urged Evans to release detailed sampling data in hopes that further statistical work could clear up the problems that led to the recommendation.

"I hope that the census will quickly release the block-level corrected data for others to evaluate and scrutinize and that the [Bush] administration will not stand in the bureau's way," Maloney said.

The 12-member expert committee said the bureau's sampling program estimated more people than separate demographic estimates would have predicted. The panel also cited potential problems with many of the assumptions used to design the formulas for estimating particular population groups.

Evans, who last month issued a controversial rule establishing his authority over redistricting numbers, said in a statement that he plans to decide by March 6 whether to release sampling numbers.

Although he is widely expected to recommend against sampling-- consistent with the prevailing Republican position--Evans said Thursday that he would base his decision on the recommendation and the findings of a group of independent experts.

"My objective is an open and fair process that will generate a decision all Americans can respect," Evans said.

Former Census Bureau Director Kenneth Prewitt, who has endorsed sampling in theory, told CongressDaily that it is possible that another six months of statistical work could clear up some of the questions about sampling, but he endorsed the expert panel's recommendation.

"I think they did a scientifically correct and prudent thing. It's rocket science. It's a very complicated thing," Prewitt said. "They had to ask the question of whether [in using sampling] they would be including as much error as they would be eliminating."