6 Jul
2001
6 Jul
'01
2:51 p.m.
From: Census2000 <Census2000(a)ccmc.org>
Congress Reviews American Community Survey Plans;
Census Bureau to Release First National ACS Test Data
Plus: Updates on Funding, Lawsuits, and Monitoring Board
Activities; Mayors, State Officials Press for Release of
Adjusted Numbers; and more.
The House of Representatives Subcommittee on the Census held its second
hearing on the Census Bureau's proposed American Community Survey (ACS)
on June 13. Acting Census Bureau Director William Barron Jr. told panel
members that the ACS "is the cornerstone of the government's efforts to
keep pace with the country's ever-increasing demands for timely and
relevant household data." The proposed survey, which would collect data
from three million households a year when fully implemented in 2003, is
one of three key components of the Bureau's strategy for planning a 2010
census that does not include the traditional long form. (Note: The
Census 2000 Initiative is preparing Fact Sheets on the ACS history,
methodology and operations, and data products, as well as on key issues
associated with development and implementation of the program. Look for
these new materials on our web site starting July 16, at
www.census2000.org <http://www.census2000.org>.)
Three demographers who have closely followed development of the ACS
expressed strong support for the concept of producing annual estimates
of important demographic and socio-economic characteristics, but warned
against
abandoning the census long form before a thorough evaluation of the ACS
program is complete. Dr. Paul Voss, Professor of Rural Sociology at the
University of Wisconsin-Madison and Director of the Wisconsin Applied
Population Laboratory, noted that the ACS evolved from a recognition
following the 1990 census that "old data... cannot possibly serve
decade-long programmatic needs in an era of fast-paced demographic
change and devolution of authority." He commended the Census Bureau for
modifying basic design elements of the ACS plan in response to feedback
from local data users, particularly those from rural areas and small
communities. The current ACS plan would produce annually updated
estimates for small areas and small population groups, starting in 2006
and 2008, based on three or five-year averages. The multi-year
averaging is necessary to ensure accumulation of sufficient sample size
to approximate the reliability of census long form data gathered from 17
percent of U.S. households.
Dr. Voss said his greatest concern is the sample size, over time, for
small areas, which he said must be sustained to ensure high quality data
for these communities. While the Census 2000 long form was sent to an
average of one in six households, the sampling rate in the most sparsely
populated areas was as high as 50 percent. The proposed ACS sample
size, Dr. Voss cautioned, is "already beginning to fall short of th[e]
goal" of yielding data as reliable as census long form data for rural
and small areas. Insufficient funding, he said, "could well place the
quality of small-area data from the ACS outside the range of
acceptability."
Linda Gage, who served as the California Governor's Census 2000 Liaison,
said the state "support[s] the full development and rigorous evaluation
of the American Community Survey," but that it is "premature" to endorse
its use over the census long form. Ms. Gage noted that the survey would
not be fully implemented until 2007, when five years of accumulated data
would be available to produce the first estimates for areas and groups
of 20,000 population and below in 2008, a category that includes 92
percent of the nation's cities. She said it is important for the ACS
sample to increase as the number of housing units grows throughout the
decade. Ms. Gage also urged improvements in the Census Bureau's annual
(or 'intercensal') population estimates, which will be used as controls
for the ACS sample. The Bureau's national annual estimate for April 1,
2000 was nearly seven million below the census count tied to that date.
Other issues the Bureau should address, Ms. Gage suggested, include the
effect of voluntary versus mandatory response and assistance for
non-English speakers who receive the ACS questionnaire. "We recommend
that 2010 census planning include a contingency for a long-form
questionnaire until a positive decision to use the ACS can be made," Ms.
Gage told subcommittee members.
Don Hernandez, representing the Population Association of America and
the Association of Population Centers, called the ACS an "innovative
effort...to dramatically improve the timeliness of data" collected
previously on the census long form and said the two organizations
"strongly endorse the on-going development and evaluation of the ACS."
Historically, Mr. Hernandez said, the long form has been "the most
accurate, and often the only, source of data" for public and private
sector decision-making, particularly at the local level. He echoed the
concern that the ACS must receive full funding each year to sustain an
adequate sample size that reflected growth in population and housing
units. Mr. Hernandez suggested that the ACS content should be flexible
to collect data on "pressing, emerging state-level public needs for
local data." The Bureau must conduct broader and systematic evaluations
of the ACS data and operations through release of data for the smallest
areas in 2008, Mr. Hernandez said, before data users can assess fully
the program's viability.
The subcommittee also heard testimony from Marilyn McMillen, Chief
Statistician at the National Center for Education Statistics (NCES),
U.S. Department of Education. NCES "collects and disseminates data on
all aspects of American Education from pre-school through adult
education," using sample surveys, performance and skill assessments,
longitudinal studies, and universal surveys. NCES uses data from the
census long form to compile information at the state, local, and school
district levels, Ms. McMillen said. The ACS would allow the agency to
produce more timely data, as well as to identify households with "rare
populations," such as recent immigrants, in order to field more targeted
surveys to assess the educational characteristics of these populations.
Ms. McMillen noted that the Education Department distributes $12 billion
annually to state and local governments based in whole or in part on
data collected only on the census long form, but that lawmakers have
required the use of updated counts of school-age children in poverty for
the largest grant program. The ACS, she said, would help improve the
accuracy of those estimates for counties and school districts.
Subcommittee members expressed a range of views about the proposed ACS
design and content. Chairman Dan Miller (R-FL) expressed concern about
the survey's potential cost, noting a Bureau estimate of $131 million in
fiscal year 2003 (FY03), the first year of nationwide implementation.
He acknowledged the benefits of the new program, such as providing more
timely information than the once-a-decade long form, but said cost and
privacy concerns must be considered "in order to determine whether the
ACS is the best means by which to collect" the data policymakers need.
Rep. William L Clay (D-MO), the panel's new Ranking Minority Member,
said Congress must consider whether the ACS is an effective investment
in light of requests to improve statistical programs in several
agencies. Rep. Bob Barr (R-GA) said there were "very serious privacy
concerns" associated with the questions proposed for the ACS, which
currently includes 69 questions, compared to 53 asked on the Census 2000
long form. "There is no way that this is the bare minimum of
information that the government needs," Rep. Barr told Mr. Barron.
Mr. Barron acknowledged that the first year cost estimate of $131
million for the ACS was lower than projected costs for subsequent years,
because there would be only nine months of field follow-up required in
FY03. (Note: The federal fiscal year begins October 1, while the ACS
will run on a calendar year starting in January.) But the "life cycle
costs" for the 2010 census, Mr. Barron said, "will demonstrate cost
neutrality" between repeating the Census 2000 design in 2010 and
conducting a short-form only census in 2010, with the ACS replacing the
long form. In response to concerns about the intrusiveness of the ACS
questions, the director noted that all of the topics have a federal law
mandate or are required to administer federal programs or laws.
Bureau set to release national, state ACS test data: This month, the
Census Bureau will start releasing data collected in the Census 2000
Supplementary Survey (C2SS), which surveyed 58,000 households a month
using the questionnaire and methods developed for the American Community
Survey. The
C2SS, administered in 1,203 counties during the decennial census year,
was designed "to demonstrate the operational feasibility of collecting
[census] long form information at the same time, but separately from the
decennial census," according to the Bureau.
The July release will include demographic, economic, and housing
information for the nation, the 50 states, and the District of
Columbia. The tables will be available through the Bureau's web-based
American FactFinder at <http://factfinder.census.gov>. By the end of
2001, similar data will be available for most counties and
municipalities with 250,000 population or greater. More information on
the C2SS can be found at www.census.gov/c2ss/www
<http://www.census.gov/c2ss/www>, beginning July 25.
Utah pursues legal, legislative remedies for loss of House district: The
State of Utah formally appealed a lower federal court ruling that the
Census Bureau was not required to count all American citizens living
abroad in Census 2000, a policy the State claims cost it a fourth
congressional seat in the post-census apportionment process. The appeal
goes directly to the U.S. Supreme Court.
Utah, its state legislative leaders, and its congressional delegation
filed a lawsuit in January, contending that the Census Bureau's failure
to count more than 10,000 Mormon missionaries serving in foreign
countries, while including members of the armed forces and federal
government employees stationed abroad, had shifted the last seat in the
U.S. House of Representatives from Utah to North Carolina. In April, a
three-judge panel unanimously rejected the State's argument that the
Census Bureau's policy was unconstitutional.
Meanwhile, Rep. Christopher Cannon (R-UT), a member of the Subcommittee
on the Census, introduced legislation that would require the Census
Bureau to submit a detailed plan for counting American citizens living
abroad in future censuses. H.R. 2171 (introduced on June 14) would
prohibit the Bureau from spending funds to implement the American
Community Survey until the Secretary of Commerce approves a written plan
for counting overseas Americans and puts regulations in place to carry
out the plan. In a written statement, Rep. Cannon said his bill "will
address the Bureau's perverse set
of priorities." The congressman told The (Salt Lake City) Deseret News
that, "The information [collected in the ACS] is nice. But the core
mission of the census according to the Constitution is for
apportionment." H.R. 2171 was referred to the Committee on Government
Reform, which has not yet scheduled hearings on the measure. The Census
Bureau is preparing a report to Congress, due September 30, on the
issues involved in counting overseas Americans. Acting Director Barron
told the Decennial Census Advisory Committee at a June 21 meeting that
the report would raise significant concerns about the feasibility of
such a count.
Utah is challenging its failure to gain a fourth congressional district
on other grounds, as well. It filed a second lawsuit in April, arguing
that the use of a statistical method called "imputation" violates the
Census Act and the Constitution. Imputation involves the use of
statistical models to assign occupants (or vacancy status) to housing
units for which census takers cannot gather any information, based on
data collected from nearby households. A hearing in the second case has
been set for August; the State filed a motion for summary judgment in
June, asking a new three-judge panel to declare the use of imputation
illegal without a prolonged trial. The Census Bureau has said
imputation added about 1.2 million people, or 0.2 percent of the
population, to the state population totals used for apportionment in
Census 2000.
Census Monitoring Board activities: The Congressional Members of the
U.S. Census Monitoring Board held a press briefing in Washington, D.C.
on June 7 to discuss their semi-annual report to Congress. A. Mark
Neuman, co-chair for the Board's three Republican members (there is one
Republican vacancy), called Census 2000 the "most successful" in
history, citing a 60 percent reduction over 1990 in the undercount gap
between African Americans and non-Hispanic Whites and a "dramatic
reduction" in the undercount of Hispanics. Mr. Neuman, who succeeded
Kenneth Blackwell as co-chair for the Congressional appointees, said the
Census 2000 results do not mean "we can't do better in the future," but
that analysis of the count "tells us how to do better." He credited
local partnerships and the use of enumerators indigenous to the
neighborhoods they counted for helping to increase participation among
traditionally hard-to-enumerate populations in 2000.
Republican Board Member David Murray called the possibility of using the
Accuracy and Coverage Evaluation (A.C.E.) survey results to
statistically correct under- and overcounts in the census a case of
"hope over experience." He criticized the adjustment methodology, which
is based on the A.C.E. survey results, for relying on "weak" assumptions
about the similarity of different demographic subgroups (i.e. racial
groups, renter/owner, age cohorts, male/female) and for having too small
a sample to measure the accuracy of the raw census data for numerically
small population groups. The A.C.E. program included a quality-check
survey of 314,000 households, designed to measure the numbers of people
missed and counted twice in the census.
The Congressional Members' report, entitled "A Guide to Statistical
Adjustment: How It Really Works," is available on their web site at
www.cmbc.gov <http://www.cmbc.gov>. The Presidential Members issued
their own report in April, which is available on their web site at
www.cmbp.gov <http://www.cmbp.gov>. Congress established the
eight-member Board in late 1997 to monitor Census 2000 activities.
Congressional Republican leaders and President Clinton each appointed
four members to the panel.
Earlier this week, the Board's four Presidential Members released
estimates of the Census 2000 undercount for counties with populations of
500,000 or greater. Dr. Eugene Ericksen, a professor of statistics at
Temple University and co-chair of the Special Advisory Panel on
Adjustment during the 1990 census, prepared the estimates based on
information from the A.C.E. survey. The co-chair for the Presidential
Members, Gilbert Cassellas, said the analysis was necessary in light of
the Census Bureau's decision not to release the A.C.E.-adjusted numbers,
which he said "could be of significant use to state and local
governments for planning decisions and funds distribution."
Dr. Ericksen described his county undercount estimates as "reasonable
approximations" of the Census Bureau's calculations and noted that they
do not include the 'group quarters' population. He previously estimated
the undercount rates for states, which he said, "equal the [Census
Bureau's] estimates." According to Dr. Rexene's analysis, Los Angeles
County had the highest number of people missed (176,000), while Bronx
County (NY) had the highest undercount rate (2.7 percent), of the
nation's largest counties.
The Presidential Members also criticized the Commerce Department and
Census Bureau for proposing new rules governing the release of
information to the Monitoring Board and other oversight bodies, which
they believe "plainly violates our enabling legislation" and "severely
impedes our statutory reporting requirements." In a letter to Commerce
Secretary Donald Evens, Mr. Cassellas said the proposed Memorandum of
Understanding was "designed to repress the flow of information" on the
census. In a subsequent letter to the chairmen of the Senate committees
that oversee and fund the Census Bureau, Mr. Cassellas said the Bureau
would not comply with requests for data from the Monitoring Board unless
its members agreed not to reveal the information publicly until 2002.
The Board's authority will expire on September 30, 2001. The final
report from the Presidential Members "will not be complete, having been
denied the data necessary for a thorough and complete analysis of Census
2000," Mr. Cassellas told the senators.
Census Bureau FY03 Funding Bill Starts Moving Through House: The House
of
Representatives began the process of funding Census Bureau activities in
Fiscal Year 2002 (FY03), as the Appropriations Subcommittee on Commerce,
Justice, and State, The Judiciary and Related Agencies approved a draft
spending bill by voice vote on June 27. The subcommittee "mark," as it
is called, includes $519.8 million for the Census Bureau in the fiscal
year beginning October 1, 2001, about $23.6 million less than the Bush
Administration's request of $543.4 million. The $38.5 billion spending
bill, one of 13 regular annual appropriations measures that fund all
federal government activities, allocates $4.76 billion for the Commerce
Department, which houses the Census Bureau.
The Commerce-Justice-State panel approved $350.4 million for Periodic
Censuses and Programs, about $24.5 million below the President's request
of $374.8 million. The "Periodic" account, one of two main Census
Bureau funding categories, covers the decennial census and proposed
American Community Survey. The panel approved $169.4 million for the
Salaries and Expenses account (which funds ongoing surveys), an increase
of $863,000 over the President's request of $168.6 million. A more
detailed spending bill will emerge in the coming days, as the full
Appropriations panel prepares to consider it when Congress returns from
its July 4th recess next week. (For details of President Bush's FY03
budget request for the Census Bureau, please see our April 16, 2001 News
Alert.)
Stockholders continue calls for release of adjusted data: The U.S.
Conference of Mayors (USCG) adopted a resolution at its annual meeting
in Detroit, calling upon the Census Bureau to release the statistically
adjusted census numbers down to the block level. The resolution,
offered by Stanford (CT) Mayor Daniel Malay and approved on June 25,
also urged the U.S. Senate to hold hearings to address the mayors'
"grave concerns" about the implications of the Census 2000 undercount
for cities and to review the Census Bureau's plans to complete the
analysis of the Accuracy and Coverage Evaluation (A.C.E.) program. The
USCG "applauded" the Bureau for "an admirable job of conducting the 2000
Census under difficult conditions," but said the Administration has not
adequately explained how it will account for the 6.4 million people the
census missed.
Two state legislators from Oregon are seeking access to the adjusted
Census 2000 data through the Freedom of Information Act (FOE). Sen.
Margaret Carter (Portland) and Sen. Susan Castile (Deign) filed a
lawsuit in federal district court on June 11 after the U.S. Commerce
Department (the defendant in the lawsuit) declined to release the
statistically adjusted numbers in response to the senators' FOE
request. The department cited a FOE exemption for information that is
"pre decisional and deliberative."
Sen. Castile said the unadjusted numbers, which the Census Bureau sent
to the states in March for use in the redistricting process, "continue
historical patterns of undercounting," but that "we won't really know
how badly Oregonians are being shortchanged" unless the Bureau releases
the entire set of adjusted data. Sen. Carter said the adjusted census
numbers "will ensure that Oregon and other states get what they deserve
when it comes to the allocation of federal funds. ...We need to right a
wrong, not perpetuate one." Thomas Balmier, an attorney representing
the senators, noted that the 9th United States Circuit Court of Appeals,
which covers Oregon, granted California's FOE request for the state's
adjusted though unofficial data following the 1990 census.
In Washington, DC, 51 Democratic Representatives called upon the Senate
to "investigate the Commerce Department's refusal to release the
corrected data." In a letter spearheaded by Rep. Clay and Rep. Carolina
Mallory (DNA), the legislators asked Sen. Earnest Holdings (D.Sc.), who
chairs the Commerce appropriations subcommittee, and Sen. Joseph
Liberian (CDT), chairman of the Governmental Affairs Committee, to hold
hearings on the department's decision to withhold the adjusted numbers,
which they said could affect the allocation of $185 billion annually in
federal program funds.
Acting Bureau Director Barron told members of the agency's advisory
committees at their meetings in late June that the Executive Steering
Committee for A.C.E. Policy (ESCAPE) would recommend whether to use
statistically adjusted data to allocate federal program funds and for
other non redistricting purposes by October 15. Newly confirmed
Commerce Under Secretary Kathleen Cooper, who also spoke to advisory
committee members, said she was not sure who would make the final
decision about using the adjusted numbers.
Questions about the information contained in this News Alert may be
directed to Terry Ann Lowenthal at 202/484-2270 or, by e-mail at
mailto:terriann2k@aol.com. For copies of previous News Alerts and other
information, use our web site www.census2000.org
<http://www.census2000.org>. Please direct all requests to receive News
Alerts, and all changes in address/phone/fax/e-mail, to the Census 2000
Initiative at mailto:Census2000@ccmc.org or 202/326-8700. Please feel
free to circulate this information to colleagues and other interested
individuals.
--
Ed Christopher
Bureau of Transportation Statistics
U.S. Department of Transportation
400 Seventh Street SW
Washington DC 20590
202-366-0412