Sampling Prompts Clash Between GOP, Census BureauThe outcome of Census 2000 continues to be a cause for concern among direct marketers as Republican legislators criticize the Census Bureau's sampling plan to determine the number and characteristics of people not reached by the standard door-to-door methods.
"The Census Bureau has laid a proposal on the table ... [to] use several highly complex … statistical processes as a substitute for the traditional, neighborhood-based head counts that have worked for the past two centuries," said Rep. Dan Miller (R-FL), chairman of the newly formed house subcommittee on the census, in testimony from one of a series of congressional hearings in Washington last month.
Partly because of the disagreement, the Government Accounting Office (GAO), a congressional audit agency that reviews government programs, is alerting the public that preparations are behind schedule, meaning that the census may not be accurate.
"[The GAO] is trying to let people know that there are a number of problems in preparing for the census, and part of the problem is the continuing fight between Congress and the administration over the use of sampling," said TerriAnne Lowenthal of the Coalition to Preserve Census Data, Washington. "Because the design for the whole census is not yet nailed down due to this controversy, it is making things very difficult."
She said Republicans also have criticized the Census Bureau's plans to build the 2000 Census address list.
But James F. Holmes, acting director of U.S. Bureau of the Census, said his agency is committed to achieving its address list goals.
"[As] we found that the U.S. Postal Service's address information was not adequate to achieve this goal [for example, the list did not include housing units], we have changed methodology to include a block-by-block canvasing operation and an additional postal check of the list to help verify the master address file," Holmes testified.
Miller also questioned the validity of a planned dress rehearsal that will examine different collection methodologies for Census 2000. It is scheduled for April 18 in Sacramento, CA, the Columbia, SC, area and the Menominee American Indian Reservation in northeastern Wisconsin. Miller testified that the trial is taking place under conditions significantly different from those for the census.
The Census Bureau expects to have "kinks" during the dress rehearsal, Holmes testified.
"The dress rehearsal will be a success if it provides us with information about what worked well and what didn't and how to fix those things that didn't work well," he said.
At another Congressional hearing last month, Holmes presented his plans and budget for Census 2000 to the House Appropriations Committee and requested $849 million in fiscal year 1999, $459 million more than the previous year.
The bureau's plans include: developing an accurate master file to verify the estimated 94 million addresses that use number and street names as well as locate the estimated 22 million households that do not have a street name; evaluating this year's dress rehearsal; opening local census offices in every Congressional district; purchasing and putting in place an automation and telecommunication infrastructure to support the network of local census facilities; completing software development and testing; and printing more than 100 million questionnaires, notification letters and other Census 2000 forms.
The bureau, however, said it is not ready to use the Internet to collect census data because of security concerns and the possible reduction of response rates.
The hearings did not address the hot topic of whether or not the long-form questionnaire will be used for Census 2000.
But, Lowenthal said: "I have this very deep fear that when the Appropriations Committee starts to unveil its bill in May, there will be some kind of effort to direct the bureau to streamline the whole process, to cut down on costs and perhaps, in their minds, improve public response. That's where the long-form gets caught in the middle."