Census Bureau Revises Census 2000 Accuracy Estimates; Los Angeles Sues For Data On Outdoor Enumeration Plus: Test Of Voluntary ACS Response Begins; Appropriations Update; New Congressional Committees Overseeing Census The Census Bureau has revised its estimates of undercounts and overcounts in Census 2000, based on further evaluation of the Accuracy and Coverage Evaluation (A.C.E.) survey conducted after the initial census enumeration. The latest findings are “dramatically different” from earlier calculations of census accuracy, Bureau officials said. They discussed the A.C.E. Revision II estimates – the second such revision to the original A.C.E. estimates of census accuracy – at a March 12th joint meeting of two National Academy of Sciences census panels. According to A.C.E. Revision II, the 2000 census overcounted the total household population by 0.5 percent, or 1.3 million people, instead of producing a net undercount of 1.18 percent (3 million people), which the Bureau had reported in October 2001. For race and ethnic groups, the new numbers showed a net overcount of 1.13 percent for non-Hispanic Whites and a net undercount of 1.84 percent for non-Hispanic Blacks. There was no statistically significant undercount or overcount for Hispanics, non-Hispanic Asians, Native Hawaiians and Other Pacific Islanders, and American Indians and Alaska Natives on and off reservations. However, the revised estimates also showed a net overcount of children aged 0 – 9. The additional review of the A.C.E. survey also revealed at least 5.8 million duplicates in the census enumeration, a much higher number than the Bureau found in its first analysis in 2001. Based on the additional review, Census Bureau Director C. Louis Kincannon reaffirmed the agency’s prior decision not to statistically adjust the census numbers used as a base for intercensal population estimates. While the new set of A.C.E. estimates “represent the most accurate assessment of Census 2000 coverage to date,” Kincannon said in a written statement, the Bureau “still [has] technical concerns about the limitations of the methodology and the data produced by A.C.E. Revision II.” The director cited efforts to account for correlation bias, the effects of synthetic estimation on small geographic areas, and remaining inconsistencies between independent Demographic Analysis estimates and A.C.E. estimates of the coverage of children in the census, as primary reasons for the decision to stick with the raw census counts for all official purposes. The Census Bureau tried for the first time to correct for correlation bias, a statistical concept that refers to the likelihood that certain segments of the population (for example, African American males, aged 18 – 34) will be missed both in the census enumeration and in the coverage measurement survey. The adjustment in the A.C.E. Revision II estimates was made based on comparisons with national Demographic Analysis estimates, which themselves are produced only for two race categories – Black and non-Black – by age and sex. Without the adjustment for correlation bias, the Bureau reported, the revised accuracy estimates would have shown a net national overcount of 1.12 percent, instead of 0.49 percent, and a net overcount for non-Hispanic Blacks of 0.53 percent, instead of a net undercount of 1.84 percent. Census Bureau staff said they were uncertain about the choice of scientific models for allocating assumptions about correlation bias among the various groupings (called “post strata) that are used to construct adjusted population numbers. They also expressed concern about their inability to determine whether correlation bias exists for children, adult women, and Hispanics. Research has led Bureau experts to assume that correlation bias is highest for adult Black men. Annual or biennial population figures are used to allocate funds for a range of federal programs, including Medicaid, Community Development Block Grants, Social Services Block Grants, and the Training Assistance-Dislocated Worker Program. Intercensal estimates are produced for states, counties, incorporated places, and minor civil divisions; many of these geographic areas have populations below 10,000. Census evaluators said the Bureau “cannot be confident of improvements in accuracy [of A.C.E. Revision II adjusted numbers] at the levels of geography for which estimates are produced.” The Bureau will post place-level “A.C.E. Revision II” estimates on its web site in the coming weeks. The data will reflect “collection places” instead of “tabulation places,” a distinction that should not affect comparisons with the official, unadjusted place-level data in most areas. Late last fall, under court order in a Freedom of Information Act case, the Census Bureau publicly released block-level A.C.E.-adjusted numbers that reflected the October 2001 revision to the initial set of undercount and overcount estimates. Those data were posted on the Internet by several universities. In response to questions at the National Academy of Sciences meeting, senior Census Bureau officials said the tight schedule for issuing block level counts to the States for redistricting did not leave enough time to adjust those data based on a post-census survey in the 2010 census. The Bureau “will measure coverage aggressively” and use Demographic Analysis as an independent benchmark in 2010, J. Preston Waite, Associate Director for Decennial Census, told panel members. Using the results of a post enumeration survey to statistically adjust the base for intercensal population estimates was still “an open question,” Mr. Waite said. The Bureau must first “fix” problems related to duplication, correlation bias, and errors in synthetic estimates. If those and other concerns are resolved, then the Census Bureau is “very likely” to adjust intercensal population estimates, Mr. Waite predicted. Mr. Kincannon’s full statement on the revised accuracy estimates can be found on the Census Bureau’s web site at <http://www.census.gov/Press-Release/www/2003/cb03cs02.html>. The technical assessment document can be found at <http://www.census.gov/Press-Release/www/2003/ExecSumm.pdf>. Los Angeles Seeks Detailed Census Data on “Homeless” Population: The City of Los Angeles has sued the U.S. Department of Commerce (the Census Bureau’s parent agency) to gain public release of Census 2000 counts from “targeted non-shelter outdoor locations.” Last May, the Census Bureau denied the City’s Freedom of Information Act (FOIA) request for separate tabulations of people counted at selected outdoor sites during the service-based enumeration (SBE) March 27 – 29, 2000. The City filed its lawsuit on November 27, 2002, in federal district court; the Commerce Department filed its response last February. (City of Los Angeles v. U.S. Department of Commerce, U.S. District Court for the Central District of California, Western Division, Case No. 02-9122 WMB) In its lawsuit, Los Angeles noted its “extensive efforts to ensure an accurate and reliable count of the homeless population in the City,” including identifying more than 7,600 outdoor locations where homeless individuals could be found. It is seeking disaggregated numbers for people counted at outdoor sites, the City said, to help it “provide appropriate outreach and other types of assistance to [the City’s] homeless population.” In its response to the lawsuit, the Commerce Department countered that the Census Bureau did not produce separate tabulations, below the county level, of people counted at various locations during the three-night SBE. Initial Census Bureau plans called for the separate release of data on people living in emergency and transitional shelters. However, the agency changed course in January 2001 based on concerns about the quality of the data. The population counted in the SBE, including people at pre-identified outdoor locations, instead was combined with “Other non-institutional group quarters” in standard Census 2000 data products. The Bureau later issued a special report on the emergency and transitional shelter enumeration. The October 2001 report, which included shelter counts for states, metropolitan areas, and large places, emphasized that the numbers “should not be construed as a count of people without conventional housing” and “do not provide a count of the population experiencing homelessness.” Some national advocacy groups, including the National Coalition for the Homeless and National Alliance to End Homelessness, supported the Bureau’s decision not to release separate data on the homeless, saying efforts to define and count this population would be “inaccurate and misleading.” Other local service agencies said the data would help them address the needs of people in homeless situations more effectively. In a recent report to Congress, the U.S. General Accounting Office (GAO) reviewed the Census Bureau’s decision not to publish separate data from components of the Service-Based Enumeration. Its findings and recommendations are set out in a report entitled, “Decennial Census: Methods for Collecting Data on the Homeless and Others Without Conventional Housing Need Refinement,” available through GAO’s web site at http://www.gao.gov/new.items/d03227.pdf. While acknowledging the “problematic” nature of counting people without conventional housing, congressional auditors found that the Bureau’s failure to “clearly and consistently communicate its plans” regarding SBE data products led to “expectation gaps” between the agency and local organizations and governments which helped the Bureau prepare for and implement the census at shelters, soup kitchens, and outdoor locations. “A key cause of the Bureau’s shifting position on reporting the Service-Based Enumeration data appears to be its lack of clear, documented, and consistently applied guidelines governing the release of data from the 2000 census,” the GAO concluded. The absence of such guidelines, the GAO wrote, “could undermine public confidence in the accuracy and credibility of Bureau data.” For the 2010 census, the GAO recommended thorough testing and evaluation of procedures, under census-like conditions, to enumerate people without conventional housing. It also urged the Bureau to develop “clearly documented, transparent, and consistently applied agency-wide guidelines for releasing all census data to the public,” which could help the agency “be more accountable and consistent in its dealings with data users and stakeholders.” American Community Survey update: The Census Bureau is evaluating the effects of voluntary response to the ACS in 31 test sites around the country. All households in the test site samples are receiving questionnaires that indicate response is voluntary, instead of mandatory under the Census Act (title 13, United States Code). (Under preliminary plans to assess voluntary response, respondents in only half of the test sites would have been told their participation was voluntary.) According to the Census Bureau, the ACS voluntary response experiment, which began in March, will continue for at least three months, and possibly longer, depending on funding. Based on lingering concerns over the length and content of the traditional census long form, Congress pressed the Census Bureau to determine how voluntary participation would affect costs and data quality in the ACS. Congress allocated $1 million in the current fiscal year to carry out the evaluation. Chip Walker, a key majority aide on census issues for the House Subcommittee on Technology, Information Policy, Intergovernmental Relations, and the Census, said at a recent association meeting that initial results from the voluntary response test are expected this month. The evaluation should be completed by August, Mr. Walker said, giving Congress time to consider the full results before finalizing the Census Bureau’s spending bill for fiscal year 2004. Mr. Walker also told the Council of Professional Associations on Federal Statistics quarterly meeting that the census oversight panel tentatively plans to hold a hearing on the American Community Survey in May. Appropriations update: The process of setting a federal budget and funding levels for federal agencies continues in Congress, with the House and Senate adopting budget resolutions for fiscal year 2004, which begins on October 1. The annual budget resolution is Congress’ blueprint for federal spending and does not become a law or require the President’s signature. The House-passed budget (House Concurrent Resolution 95) provides $775 billion for discretionary programs, $9 billion above current levels but below the President’s budget submission. The Senate version (Senate Concurrent Resolution 23) provides $791 billion in discretionary spending, slightly above the President’s recommendation. The two chambers are meeting to reconcile differences between their budget plans, which also include differing proposals for tax cuts that were part of the Administration’s economic stimulus package. Once the budget plan is finalized, the House and Senate Appropriations Committees will divvy up discretionary funds among the thirteen regular appropriations accounts. The Census Bureau is funded under the Commerce, Justice, State, and Judiciary account and will compete for money with those federal departments and other Commerce Department agencies. Commerce Secretary Donald Evans testified last month before the House and Senate appropriations subcommittees responsible for that account. He told lawmakers that Census Bureau and Bureau of Economic Analysis (BEA) programs help foster economic growth, one of the department’s five strategic goals. The Secretary noted that the proposed Census Bureau budget included increased funds “to fill gaps” in the collection of demographic and economic statistics, to process data from the 2002 Economic Census, and to continue planning for the 2010 census. Additional funds requested for BEA, he said, would help speed up release of key economic statistics and improve the quality and timeliness of the data. Congressional Committees in the 108th Congress: The House and Senate have reorganized their committees and subcommittees for the 108th Congress, with several new faces emerging on panels that fund and oversee the census. In the House, Chairman Frank Wolf (R-VA) and Rep. Jose Serrano (D-NY) continue to hold the top slots on the appropriations Subcommittee on Commerce, Justice, State, and Judiciary, which funds census activities. The Committee on Government Reform, the Census Bureau’s authorizing panel, completely reorganized its subcommittees under a new chairman, Rep. Thomas M. Davis, III (R-VA); Rep. Henry Waxman (D-CA) continues as the committee’s senior Democrat. The newly formed Subcommittee on Technology, Information Policy, Intergovernmental Relations, and the Census is headed by a sophomore legislator, Rep. Adam Putnam (R-FL); Rep. William “Lacy” Clay, Jr. (D-MO), who served on a previous census oversight panel, returns as the new subcommittee’s ranking minority member. With Republicans resuming majority status in the Senate, Sen. Judd Gregg (R-NH) is again chairman of the appropriations Commerce subcommittee, while Ernest “Fritz” Hollings (D-SC) holds the ranking minority position. The Committee on Governmental Affairs has assigned responsibility for census oversight to a newly formed Subcommittee on Financial Management, the Budget, and International Security. Sen. Peter Fitzgerald R-IL), a new member of the full committee, chairs the panel; Sen. Daniel Akaka (D-HI) is the senior Democrat. The chairmen and ranking minority members of full committees usually serve as ex officio members of all of that panel’s subcommittees of which they are not regular members. House Committee on Government Reform Subcommittee on Technology, Information Policy, Intergovernmental Relations, and the Census Republicans Democrats B349A Rayburn H.O.B. B350A Rayburn H.O.B. Washington, DC 20515 Washington, DC 20515 Adam Putnam (FL) (Chairman) Wm. Lacy Clay (MO) (Ranking Minority Member) Candice Miller (MI) Diane Watson (CA) Doug Ose (CA) Stephen Lynch (MA) Tim Murphy (PA) Henry Waxman (CA) (ex officio) Michael Turner (OH) Tom Davis (VA) (ex officio) Senate Committee on Governmental Affairs Subcommittee on Financial Management, the Budget, and International Security Republicans Democrats SH-446 Hart S.O.B. SH-439 Hart S.O.B. Washington, DC 20510 Washington, DC 20510 Peter Fitzgerald (IL) (Chairman) Daniel Akaka (HI) (Ranking Minority Member) Ted Stevens (AK) Carl Levin (MI) George Voinovich (OH) Thomas Carper (DE) Arlen Specter (PA) Mark Dayton (MN) Robert Bennett (UT) Frank Lautenberg (NJ) John Sununu (NH) Mark Pryor (AR) Richard Shelby (AL) Joseph Lieberman (CT) (ex officio) Susan Collins (ME) (ex officio) House Committee on Appropriations Subcommittee on Commerce, Justice, State, and the Judiciary Republicans Democrats H-309 The Capitol 1016 Longworth H.O.B. Washington, DC 20515 Washington, DC 20515 Frank Wolf (VA) (Chairman) Jose Serrano (NY) (Ranking Minority Member) Harold Rogers (KY) Alan Mollohan (WV) Jim Kolbe (AZ) Robert Cramer, Jr. (AL) Charles Taylor (NC) Patrick Kennedy (RI) Ralph Regula (OH) Martin Sabo (MN) David Vitter (LA) David Obey (WI) (ex officio) John Sweeney (NY) Mark Kirk (IL) C.W. Bill Young (FL) (ex officio) Senate Committee on Appropriations Subcommittee on Commerce, Justice, State, and the Judiciary Republicans Democrats S-146A The Capitol SH-123 Hart S.O.B. Washington, DC 20510 Washington, DC 20510 Judd Gregg (NH) (Chairman) Ernest Hollings (SC) (Ranking Minority Member) Ted Stevens (AK)* Daniel Inouye (HI) Pete Domenici (NM) Barbara Mikulski (MD) Mitch McConnell (KY) Patrick Leahy (VT) Kay Bailey Hutchison (TX) Herbert Kohl (WI) Ben Nighthorse Campbell (CO) Patty Murray (WA) Sam Brownback (KS) Robert C. Byrd (WV) (ex officio) *Full committee chairman. Census News Briefs are prepared by Terri Ann Lowenthal, an independent consultant in Washington, DC. Please direct questions about the information in this News Brief to Ms. Lowenthal at 202/484-2270 or by e-mail at terriann2k(a)aol.com. Thank you to the Communications Consortium Media Center for posting the News Briefs on the Census 2000 Initiative web site, at www.census2000.org. Please feel free to circulate this information to colleagues and other interested individuals.