|
How Census 2000 will be Conducted Using Modern Science
April 1, 2000 was Census Day, but in reality the census started long before that day, and major
census taking operations will continue for a year after the date. The 2000 Census will count
approximately 275 million people residing in 120 million housing units across the nation on a
single day -- April 1, 2000. To do this, over 500,000 people had to be hired for over 860,000
positions. To fill this number of positions, close to 2.5 million people had to be tested and
interviewed. The vast majority of these positions were needed for nonresponse follow-up
operations, which began in mid-April. All 520 Local Census Offices (LCOs) opened and were
operational on January 1, 2000. The will begin closing after the completion of nonresponse
follow-up and coverage improvement operations in late July.
Modern census-taking is based on a mailout/mailback strategy - the questionnaire is
mailed to every household in the country and hopefully completed and mailed back to the Census
Bureau. In 2000, about 66 percent of the households mailed back their forms, reversing a
decades-long decline in participation with the census. To accomplish this massive mailing, the
Census Bureau creates a Master Address File (MAF) that identifies all living quarters and locates
them in its geographic database, called TIGER. Creation of the MAF is a multi-year process and
involves a number of operations and the active partnership of state, local and tribal governments.
The Bureau created its initial address list by combining the 1990 Census Address Control File
with the US Postal Service Delivery Sequence File. The Bureau then checked this list with a
block canvass operation in which listers physically inspected each street, road, and structure
around the nation. The Bureau then allowed state, local and tribal governments the opportunity
to review and update the address list. Finally, the Bureau conducted a New Construction
Program, beginning in January, 2000, to add any addresses created since the local review.
Census 2000 also included, for the first time, an integrated communications and marketing
program to increase awareness of the Decennial census and boost response rates. The three
phases of the marketing program were (1) Prior to Census Day, build awareness of the Census
and how it will benefit communities, (2) During the mailout/mailback period, motivate people to
return their questionnaire promptly, and (3) During the enumerator follow-up period, encourage
cooperation with census-takers. There were several major activities involved in the marketing
program. Paid Advertising was used for the first time in a $167 million campaign. Partnerships
between the Bureau and other federal agencies, state and local governments, community-based
organizations and businesses increased awareness of the census. Methods to improve mail
response included a letter alerting people to the coming census questionnaire and a reminder
postcard urging their response. Promotion and Special Events included parades, athletic events,
public service announcements, and a fleet of twelve road tour vehicles to build awareness and
support for the census.
The Census Bureau also opened a national network of temporary field offices from which
employees collected and processed data for the 2000 Census. This office structure includes 12
Regional Census Centers (RCCs, open since March 1998), 520 Local Census Offices (LCOs,
open since January 2000), 3 temporary Data Capture Centers, and one permanent national
Processing Center located in Jeffersonville, MO. To conduct the 2000 Census, the Bureau will
recruited about 2.5 million applicants for 860,000 positions. The actual counting of people for
the 2000 Census began on January 18, 2000 when, due to climatic conditions, enumerators began
counting the residents of remote areas of Alaska. In early March other major field operations
began. In extremely remote areas of the country, enumerators visited households and completed
questionnaires with the residents in an operation called List/Enumerate. Also in early March, in
many rural areas, Census Bureau employees physically delivered a mailback questionnaire to
about 20 million households in an operation call Update/Leave.
On March 13, 2000, major enumeration activities for the 2000 Census began with the
mailing of questionnaires to about 100 million households across the country. The U.S. Post
Office delivered these forms to every "city-style" housing unit with a street name and house
number. The Bureau provided assistance in a number of ways to persons who needed help in
completing their questionnaire. Telephone Assistance Centers operated a toll-free system in
English, Spanish and several other languages. Questionnaire Assistance Centers were located in
convenient locations, in many cases with bilingual staff. Questionnaire Assistance Guides were
available in over 50 languages. In addition, respondents were able to access an Internet web site
to receive assistance and, in some cases, submit their responses. Questionnaires were also
available at Be Counted sites for those who believed they did not receive one in the mail. The
Bureau also designed a number of different operations to enumerate individuals who do not live in
traditional housing units. Beginning on March 27 and continuing until April 11, the Census
Bureau published the mail response rates for all 39,000 state, local, and tribal jurisdictions.
Starting on April 27 and continuing until July 7, the Census Bureau conducted the most
labor intensive and highest visibility operation - Nonresponse Follow-up (NRFU). All 42 million
households that did not mail back a questionnaire were visited by an enumerator to obtain a
completed questionnaire. During this period, Census operations were at their peak, with
approximately 500,000 people in the field.
Upon completion of NRFU operations, the Census began an operation called Coverage
Improvement Followup (CIFU). During CIFU, enumerators attempt to contact housing units
from three categories. First, enumerators will contact housing units added to the address file
during the update/leave operation. During update/leave Census Bureau employees update the
address list and leave mail-back forms in mainly rural areas. About 700,000 previously
unidentified housing units were added during this operations. Second, housing units added by the
new construction program are visited. This program is intended to identify housing units which
are constructed right up to April 1, 2000, and added about 600,000 addresses. The third and
biggest category of addresses to be visited during CIFU are units coded as vacant during NRFU.
Eight to ten million addresses coded as vacant will be visited one final time to ensure that no one
lives at that address. The Bureau estimates that it will need approximately 156,000 enumerators
for this operation. These operations produce as complete a count as possible using only old
methods and after being tabulated, will be used to provide uncorrected apportionment counts of
state population that are delivered to the President and Congress on December 31, 2000.
After NRFU, the Accuracy and Coverage Evaluation (ACE) began. This is the major field
operation using modern statistical methods designed to provide data to correct historic errors in
the census count, particularly the racial differential. The ACE is essentially a second census of
314,000 households and began on June 19, on a flow basis as NRFU ended. The Census Bureau
deployed more than 10,000 specially trained census takers to conduct the Accuracy and Coverage
Evaluation across the country. Using techniques tested and employed by the Bureau since the
1940's, a more exact population count of each neighborhood, locality, and state was determined
through the use of scientific methods.
The major stages of the ACE operation include: independent address listing, housing unit
matching, person interviewing, person matching, estimation, and final housing unit matching. The
independent listing and the housing unit matching stages are complete.
The ACE population numbers are now being compared to the less accurate direct
population counts to develop a second set of Census 2000 population data, combining scientific
methods and the direct count figures, that should greatly reduce the undercount. This corrected
data, down to the block-level, will be made available to the states for redistricting by March 31,
2001.
|
|
|
