More Sampling Planned In 2000 Census

The Census Bureau, responding to criticism that it has undercounted minorities in the past, has decided to redesign the 2000 Census.

The change, which is drawing some fire from Capitol Hill, will rely more heavily on statistical sampling to arrive at population numbers for households that do not return census forms and cannot be contacted by census workers.

“We believe that the judicious use of statistical sampling is a prudent, cost-saving and legally permissible means for helping our nation achieve the best census ever,” said Commerce Secretary William Daley in a recent statement to the Senate Governmental Affairs Committee.

The bureau’s decision to use sampling is the latest chapter in a long controversy over how the federal government should conduct the 2000 Census, a massive effort to tabulate information about the nation’s 120 million households. It is expected to cost about $4 billion.

But even as Commerce officials, who oversee the Census Bureau, proceed with their plan, it appears to be headed for a challenge from some in Congress who object to the use of sampling.

Rep. Thomas E. Petri, R-Wis., said he is concerned about the accuracy of sampling, particularly when used to arrive at population figures for small communities.

Unlike the 1990 Census, which attempted to contact every household that did not return a census form, the plan for the 2000 Census is to count 90 percent of households in every census tract, then use statistical sampling to estimate population information for the remaining 10 percent.